Research ArticleNovel Resource Allocation Algorithms for the Social Internet ofThings Based Fog Computing Paradigm

Sungwook Kim

Department of Computer Science Sogang University 35 Baekbeom-ro (Sinsu-dong) Mapo-gu Seoul 04107 Republic of Korea

Correspondence should be addressed to Sungwook Kim swkim01sogangackr

Received 19 December 2018 Revised 29 January 2019 Accepted 6 February 2019 Published 19 February 2019

Guest Editor Miriam Capretz

Copyright copy 2019 Sungwook Kim This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Social Internet ofThings (SIoT) is a control paradigm by the integration of social networking concepts into the Internet of Thingsand Fog Computing (FC) is an emerging technology that is aimed at moving the cloud computing facilities to the access networkRecently the SIoT and FC models are combined by using complementary features and a new Social Fog IoT (SFIoT) paradigmhas been developed In this paper we design novel resource allocation algorithms for the SFIoT system Considering the socialrelationship and each preference mobile devices in the SFIoT effectively share the limited computation and communicationresources of FC operators To formulate the interaction among mobile devices and FC operator we adopt the basic concept of twogame models voting and bargaining games Bicooperative voting approach can make control decisions for the resource allocationmethod andNash bargaining solution is used to effectively distribute the computation resource to different application tasks Basedon the two-phase game model the proposed scheme takes various benefits in a fair-efficient wayThrough the extensive simulationexperiments we can validate the superiority of our proposed approach by the fact that it produces a mutually acceptable agreementamong game players and significantly outperforms the existing protocols Last we point out the further challenges and futureresearch issues about the SFIoT paradigm opportunities

1 Introduction

Since its birth in the early 1960s the Internet has madeenormous progress in research development and innovationin the data communications Recently the Internet hasstarted to connect every thing in the physical world Thispervasive paradigm known as Internet ofThings (IoT) mightincrease the value of information exchanged by the number ofinterconnected things and can usher in a wide range of smartservices and applications for the benefit of mankind andsociety Accordingly the IoT technologies make it possible toprovide new services to end-users while establishing socialrelationships in an autonomous way Similar to what happenswith social networks among humans smart objects in the IoTconnected each other asymmetrically with their preferencesimilarity and interests [1ndash3]

The integration of social networking concepts into theIoT has led to a burgeoning topic of research the so-calledSocial IoT (SIoT) paradigm In the SIoT smart objectscan find the desired services through its social networking

in a decentralized manner The term lsquosocial networkingrsquorevolves around not only the online social networking butalso the natural social relations in the local area Thereforethe inter-thing social network can be exploited based onsocial device-to-device short-range wireless connections Asa consequence we have witnessed the rise of tremendousbusiness and social opportunities in the SIoT paradigm TheSIoT paradigm effectively addresses the scalability issue ofthe native IoT paradigm by enforcing the convergence ofsocial and communication networks However the socialconnectivity-based IoT services may cause the cascade effectTherefore for effective and efficient SIoToperations themainobjective pursued by the SIoT paradigm is to maximize thesocial welfare based on the social relations [2 4]

In the era of big date based computation-intensiveapplications data generated from SIoT devices are generallyprocessed in a cloud infrastructure However it may beinefficient to send the large data of IoT devices to the remotecloud system especially for time-sensitive applications Toaddress this issue Fog Computing (FC)method has exhibited

HindawiWireless Communications and Mobile ComputingVolume 2019 Article ID 3065438 11 pageshttpsdoiorg10115520193065438

2 Wireless Communications and Mobile Computing

the right features FC provides low-latency computing cloudservices at the edge of the networkwhere data is generated Toimplement the FC architecture edge agents are evolved to theedge clouds called cloudlets by being equippedwith a certaincomputation power capability Therefore it is an enabler forthe emerging SIoT systems while handling the rapid devel-opment of computation-intensive applications Comparedto the traditional cloud computing the FC method withSIoT system can greatly improve the efficiency by providingthe powerful computing resource through one-hop wirelessconnections [5 6]

Usually SIoT and FC are two stand-alone technologicalparadigmsunder the realmof the future generation networksTo guarantee the scalability of large IoT SIoT relies on theself-establishment and self-management of inter-thing socialrelationships To support the computation-intensive appli-cations in IoT devices FC extends cloud capabilities to theedges of access networks Motivated by these complementaryfeatures of the SIoT and FC methods a new paradigm calledSocial Fog IoT (SFIoT) has been introduced by integratingthe SIoT and FC technologies The main idea behind theSFIoT paradigm is to embrace the design principles relatedto both SIoT and FC methods [4]

In the SFIoT system each cloudlet is a small-size vir-tualized inter-connected resource-equipped data center Atthe edge of access networks the cloudlet hosts the thing-to-FC offloading tasks for SFIoT devices By using the intra-FC resource pooling technique each cloudlet partitionsthe available resource for each application type [4] In theSFIoT system there are two types of resources computationand communication However whereas in reality the com-putation and communication resources of each individualcloudlet are limited and raced When a lot of computation-intensive applications are offloaded the resources of cloudletwill become exhausted rapidly and the quality of experience(QoE) will be seriously degraded [7ndash9] To allocate limitedcomputation and communication resources to SFIoT ser-vices efficient resource allocation is a critical issue in orderto improve the total system performance while ensuring theQoE provisioning In this study we are solving the cloudletrsquosresource allocation problem for SIoT devices which areconnected according to interdevice social ties [10]

The past decade has witnessed a huge explosion ofinterest in game theory As a branch of Applied Mathematicsgame theory is concerned with decision-making in strategicsettings and it has been successfully applied to tackle manydistributed selfish optimization problems Game theory canbe divided into noncooperative game and cooperative gameIf game players can reach a binding agreement then the gamebecomes a cooperative game On the contrary this agreementcannot be reached in the noncooperative game In particularcooperative game theory can be used to analyze manydistributed selfish optimization problems in communicationnetworks [11]

In the SFIoT vision individual devices interact accordingto the peer-to-peer communication model by autonomouslybuilding up inter-thing social relationships with respect totheir preferences [4] To improve the total SFIoT systemperformance each individual device needs to cooperate with

each other To get the mutual advantages for themselves therelationship of interrelated SFIoT devices can be modeledas cooperative games Inspired by the basic concept ofcooperative games the major question to be answered ishow to reach a consensus for the effective SFIoT resourceallocation This issue is our main concern in this study

In 1950s John Nash proposed a simple cooperative gamemodel which predicted an outcome of bargaining based onlyon information about each playerrsquos preferences To allocateresources fairly and optimally Nash Bargaining Solution(NBS) is the unique bargaining solution that satisfies sixaxioms individual rationality feasibility symmetry paretooptimality independence of irrelevant alternatives and invari-ance with respect to utility transformations Therefore it canachieve a mutually desirable solution with a good balancebetween efficiency and fairness In addition the NBS doesnot require global objective functions unlike conventionaloptimization methods such as Lagrangian or dynamic pro-gramming [11]

As a kind of cooperative game simple voting game is veryuseful in modeling a decision-making process However itallows each voter only two choices to support or oppose adecisionThis restriction ignores that voters often can abstainfrom voting which is effectively different from the othertwo options [12 13] By considering abstainers Bicooper-ative Voting Game (BVG) can formulate a more realisticdecision-making process to adaptively control the SFIoTsystem

Based on these appealing properties we adopt the basicconcepts of NBS and BVG to solve the resource allocationproblem in the SFIoT system For the computation resourceallocation we adopt the NBS It is formulated by an expectedutility function over the set of feasible agreements basedonly on information about each playerrsquos preferences For thecommunication resource allocation we use the fundamentalidea of voting game However classical voting games cannotbe effectively applied to design our SFIoT resource allocationalgorithm As a special class of voting games the BVGcan formulate a more realistic decision-making process toadaptively allocate the communication resource of the SFIoTsystemWith desired properties of NBS and BVG we attemptto reach an outcome that meets our design goals while takingreciprocal advantages in a technologically more suitableway

11 Related Work Due to the driving force for the improve-ment of SFIoT system there have been considerableresearches about the implementation of social relation basedFC control algorithms and some SFIoT control schemeshave been published L Liu et al propose the socially awaredynamic fog computing (SDFC) scheme for the computationoffloading mechanism [14] This scheme advocates a gametheoretic model and proposes a new dynamic computationoffloading method to minimize the social group executioncost In particular a computation offloading problem in a FCsystem has been investigated and this problem is formulatedas a generalized Nash equilibrium problem To derive thedelay performance during the offloading process different

Wireless Communications and Mobile Computing 3

queuemodels are also applied Finally it is addressed by usingthe semi-smooth Newton method with Armijo line searchThe SDFC scheme can reduce the accumulated error andadaptively improves the calculation accuracy during iterationprocess The simulation results demonstrate the effectivenessof the SDFC scheme [14]

The social system through edge computing (SSEC) schemeis proposed as a new collaborative horizontal offloadingmethod for the distributed data processing mechanism [15]To discover and select devices which are able to performan offloaded task according to specific requirements thisscheme implements the clustered edge computing paradigmCurrently personal mobile devices are seen not only aspassive data generators and IoT service consumers butrather as active participants and contributors Based on theprinciples of volunteer computing and SIoT the SSEC schemepresents a novel approach to address network latency issuesat the very edge of an IoT network topology while utilizingidle resources of mobile devices Finally the simulationresults show that the SSEC scheme has a potential power tooutperform cloud-centric setups by keeping the computationlocally and by involving volunteering mobile devices inclustered computation at the edge [15]

Paper [16] presents the cloud-fog information-centric com-puting (CFIC) schemeThis scheme supports IoT applicationsin various network domains with the IoT middleware andthe cloud-fog computingmechanism By using two functionsie job-classification function and resource scheduling func-tion QoE in IoT applications can be ensured Specifically thejob-classification function classifies different IoT applicationsby authority data type data update rate and priority Theresource scheduling function generates a resource allocationplan based on the resource limitation Simulation resultsreveal that the CFIC scheme can reduce the cloud computingtime while guaranteeing the execution of real-time IoTapplications [16]

The earlier studies [14ndash16] have attracted considerableattentions while introducing unique challenges in handlingthe SFIoT resource allocation problem Therefore our pro-posed scheme may look similar to the existing schemesWhile these schemes have some similarities there are severalkey differences First our proposed scheme is designedas a two-phase game model based on the two differentcooperative game approaches Second in our game modelthe main concepts of NBS and BVG are adopted to addressthe SFIoT resource allocation problem Third to reduce thecomputation complexity we hierarchically implement theBVG and the NBS with cascade interactions Most of allthe main difference between the existing schemes in [14ndash16] and our proposed scheme is a control paradigm Theprinciple novelty of our protocol is a judicious mixture of twogame solutions and its feasible self-adaptability in the real-world SFIoT systemoperations In this paper we compare ourproposed scheme with the existing the SDFC [14] SSEC [15]and CFIC [16] schemes and demonstrate that our two-phaseinteractive game approach can significantly outperform theseexisting schemes

The main contributions of this study are (i) ability tomaintain the SFIoT resource efficiency as high as possible

(ii) ability to respond to current SFIoT situations basedon the interactive bargaining and voting process (iii) QoEprovisioning for different kinds of application tasks (iv)synergy effect based on the reciprocal combination of SIoTand FC and (v) ensuring a mutually desirable solution whileimproving the overall SFIoT system performance The keyadvantage of the proposed scheme is its adaptability feasi-bility and effectiveness for realistic SFIoT system operationsunder widely different and diversified application task loadsituations

The rest of this paper is organized as follows Section 2presents the SFIoT system infrastructure and the backgroundknowledge ofNBS andBVGconcepts And then the details ofour proposed scheme are explained based on the interactivetwo-phase game model In Section 3 we provide the numer-ical results from simulation experiments and demonstratethe effectiveness of the proposed scheme comparing withthe existing SDFC [14] SSEC [15] and CFIC [16] protocolsFinally conclusions and suggestions for future research aregiven in Section 4

2 The Proposed SFIoT ResourceAllocation Algorithms

In this section we describe a new two-phase game approachfor solving the resource allocation problem of SFIoT systemIn the first phase mobile devices take two votes to decidecommunication and computation resource allocation waysIn the second phase the computation capacity is distributedto each application task based on the NBS Finally weexplain in detail the proposed scheme in the nine-stepprocedures

21 Fog Controlled Social IoT System Infrastructure Tradi-tionally the SFIoT system infrastructure is consisting of mul-tiple cloudlets and a number of mobile devices As a small-scale cloud datacenter each cloudlet is located at the edge ofthe Internet The main purpose of the cloudlet is supportingresource-intensive and interactive mobile applications byproviding powerful computing resources with lower latencyIn the SFIoT system each individual mobile device is amember of both an underlying IoT and an overlay socialnetwork Usually the IoT accounts for more conventionalnetwork services and the social network reflects somewhatless conventional social ties induced by inter-thing socialrelationships which are formally described by a social graphG = (NE) N is the set of mobile devices and E = 119890119898119894119898119895 |0 le 119890119898119894119898119895 le 1 and (119898119894 119898119895) isin N times N is the set of the socialties 119890119898119894119898119895 is a real-valued nonnegative number to measurethe strength of the social tie between 119898119894 and 119898119895 ThereforeS119898119894 = 119898119895 | 119890119898119894119898119895 gt 0 and 119898119895 isin N formally defines the setof the social friends of119898119894 [4]

To calculate each userrsquos influential power we develop asocial power function F(119909) in line with some propertiesUnder diverse scenarios the function F(119909) should obeythe following three properties to effectively estimate eachindividual userrsquos social power [17]

4 Wireless Communications and Mobile Computing

(i) nonnegative property F(119909) ge 0(ii) nondecreasing property 119889F(119909)119889119909 ge 0(iii) saturation property 1198892F(119909)1198891199092 lt 0

The nonnegative property means that the outcome shouldbe larger than zero and if there is no social relationshipit is zero The nondecreasing property simply implies thatthe more users are socially connected the more outcomecan be obtained The saturation property suggests that whenthe social relationship is sufficient any further increase ofthe outcome remains marginal [17] In order to construct afunction while satisfying above three properties the socialpower function of119898119894 ieF119898119894(119909) is defined as follows

F119898119894 (119909) = Γ minus 1exp (120583 times 119909)

st 119909 = sum119898119895isinS119898119894

119890119898119894119898119895 (1)

where 120583 controls the speed of social power saturation and Γrepresents the maximum outcome of a user to normalize thesocial power [17]

In this study we assume that there are 119896 cloudlets and 119899mobile devices where 119896 ≪ 119899 In the SFIoT system infrastruc-ture cloudlets are evolved to the Fog-computing basedAccessPoint (F-AP) by being equipped with a certain computationpower and communication capability The design of F-APplatform has been introduced to deliver various devicesrequests which have different characteristics and resourcedemands that is CPU capacity for computation and wirelessbandwidth assignment for communication However dueto the limited CPU capacity and bandwidth scarcity in theF-AP it is impossible to guarantee all applicationsrsquo needs[18] Therefore an efficient resource management strategybecomes a key factor in enhancing the SFIoT performancewhile ensuring the required QoE

To tackle the SFIoT resource control problem we pro-pose a two-phase game based resource allocation schemeAt the first phase individual mobile devices participatein two BVGs one is to decide the allocation method ofcommunication resource and the other is to partition thecomputation resource according to the voting power of eachmobile device In each F-AP covering area the voting power iscalculated based on the pseudo-Banzhaf value At the secondphase the partitioned computation capacity is distributedfor same kinds of application tasks based on the NBS Topractically implement our resource allocation algorithmseach individual F-AP and its corresponding mobile devicesoperate their own two-phase game in an entirely distributedmanner For example the 119896th F-AP ie F119860119875119896 and the set ofcorresponding mobile devices in F119860119875119896 rsquos covering area ieNF119860119875119896 work together for their two-phase gameG119896 Therefore

multiple two-phase games G1le119894le119896 are operated in parallel Toformally define our proposed game model we characterizeG119896 as F119860119875119896 NF119860119875

119896 (S119898119894 F119898119894(119909) ΨQ

119898119894) | 119898119894 isin NF119860119875

119896 119909 =sum119898119895isinS119898119894 119890119898119894119898119895AF119860119875

119896 120572F119860119875119896 119880F119860119875

119896 Table 1 lists the notations

used in this paper

(i) F119860119875119896 NF119860119875119896 is the set of gameplayers for theG119896 game

where NF119860119875119896cap NF119860119875

119897= 0

(ii) (S119898119894 F119898119894(119909) ΨQ119898119894) is a three-pair for the 119898119894 isin NF119860119875

119896

where S119898119894 is the set of119898119894rsquos social friendsF119898119894(119909) andΨQ119898119894

represent the social power and the voting powerof119898119894 respectively

(iii) AF119860119875119896

is the set of generated application tasks whichare offloaded in theF119860119875119896 for FC services

(iv) 120572F119860119875119896

is the ratio of relative service weights givento different type applications in the F119860119875119896 where0 le 120572F119860119875

119896le 1 Based on the 120572F119860119875

119896value F119860119875119896 rsquos

computation power is partitioned

(v) 119880F119860119875119896

is the utility function of F119860119875119896 it is optimizedbased on the NBS

22 Bicooperative Game and Ternary Voting Model Usuallycooperative game theory studies situations where a group ofgame players are associated to obtain a profit as a result oftheir cooperation Thus a cooperative gamemodel is definedas a pair (119873V) where N is a finite set of players and V2119873 997888rarr R is a function verifying that V(0) = 0 For eachcoalition 119878 isin 2119873V(119878) can be interpreted as the coalition Srsquosmaximal gain that players in S can achieve themselves againstthe best offensive threat by the complementary coalition119873119878[12]

In 2008 Bilbao et al proposed a new idea called bicoop-erative game for themulticriteria decision-making process Inthe bicooperative game game players are allowed to chooseone among three alternatives To think the voting situationby analogy we consider ordered pairs (S T) with 119878 119879 sube119873 and 119878 cap 119879 = 0 Each pair (S T) yields a partition of119873 in three groups Players in S are in favor of the votingitem and players in T object to it The remaining players in119873 (119878 cup 119879) are nonvoters Formally we can define the set3119873 = (119878 119879) | 119878 119879 sube 119873 119878 cap 119879 = 0 and the functionQ 3119873 997888rarr R For each (119878 119879) isin 3119873 the outcome of Q(119878 119879)can be interpreted as the gain that 119878 can achieve when 119879 isthe opposite coalition and119873 (119878 cup 119879) is the neutral coalition[12 13]

The pair (0119873) represents the situation if all the playersobject to the voting item and (119873 0) represents the situationwhere all the players support the voting item More generallya relation of 3119873 is given by [12 13]

(119860 119861) ⊑ (119862119863) lArrrArr 119860 sube 119862 119861 supe 119863 (2)

The set (3119873 ⊑) is a finite distributive lattice and a partiallyordered set with the following properties [12 13]

(i) (0119873) is the first element (0119873) ⊑ (119860 119861) for all(119860 119861) isin 3119873(ii) (119873 0) is the last element (119860 119861) ⊑ (119873 0) for all(119860 119861) isin 3119873

Wireless Communications and Mobile Computing 5

Table 1 Symbols and parameters used in the proposed algorithm

Notations ExplanationN the set of mobile devicesE the set of the social ties119890119898119894 119898119895 a real-valued number to measure the strength of the social tie between119898119894 and119898119895S119898119894 the set of the social friendsof119898119894F119898119894 (119909) the social power function of 119898119894120583 a parameter to control the speed of social power saturationΓ the maximum outcome of a user to normalize the social powerF119860119875119896 the 119896th fog-computing based access pointNF119860119875119896

the set of corresponding mobile devices in theF119860119875119896 rsquos covering areaG119896 The two-phase game inF119860119875119896ΨQ119898119894

the social power and the voting power of 119898119894AF119860119875119896

the set of application tasks which are offloaded in theF119860119875119896 for FC services120572F119860119875119896

the ratio of relative service weights given to different type applications in theF119860119875119896119880F119860119875119896

the utility function ofF119860119875119896119873 a finite set of voting game playersV (119878) the coalition Srsquos maximal gainQ (119878 119879) the gain that 119878 can achieve when 119879 is the opposite coalitionBG119873 the set of all bicooperative games with players119873120601119894 (Q) the player 119894rsquos value in a game Q onBG119873

Q(119878119879) (119860 119861) the identity gameQ(119878119879) (119860 119861) the superior unanimity gameQ(119878119879) (119860 119861) the inferior unanimity gameQ119894 the payoff of voter 119894W (119878) the sum of voting weights of coalition 119878rsquos playersP the quota for the decision-taking120599 the quota for the decision-blockingX a weighted ternary voting gameB the total available bandwidth amountm119894 the pre-defined minimum bandwidth amount for the 119894119905ℎ service119877119894 the bandwidth request of mobile device119898119894119860 119894 the decided bandwidth amount for the119898119894119879119862L119862 the constraint thresholds which are chosen to satisfy each condition120572119868 the preference ratio of class I application servicesC each F-APrsquos total computation capacity120578119868120578119868119868 control parameters to adjust the payoff of119898CF119860119875119896

the total computation capacity ofF119860119875119896

(iii) Every fair of elements of 3119873 has a join(119860 119861)⋁ (119862119863) = (119860 cup 119862 119861 cap 119863)

and (119860 119861) and (119862119863) = (119860 cap 119862 119861 cup 119863) (3)

The set of all bicooperative games with players119873 is denotedbyBG119873 which is a real vector space [12 13]

BG119873 = Q | 3119873 997888rarr R and Q (0 0) = 0 (4)

A value on BG119873 is a mapping 120601 BG119873 997888rarr R119899 thatassociates with each game Q isin BG119873 In a vector (1206011(Q) 120601119899(Q)) isin R119899 1206011le119894le119899(Q) represents a real number whichis the player 119894rsquos value in a game Q on BG119873 Therefore themapping 120601119894isin119873(Q) BG119873 997888rarr R represents the player 119894rsquospayoff from playing bicooperative games [12 13] There arethree special collections of games in BG119873 taking values inminus1 0 1 the identity games the superior unanimity gamesand the inferior unanimity games which are defined for any(119878 119879) isin 3119873 [12 13]

The identity game Q(119878119879)(119860 119861) | 3119873 997888rarr R is defined by

Q(119878119879) (119860 119861)

6 Wireless Communications and Mobile Computing

= 1 if (119860 119861) = (119878 119879) and (119878 119879) = (0 0)0 otherwise

(5)

The superior unanimity game Q(119878119879)(119860 119861) | 3119873 997888rarr R isdefined by

Q(119878119879) (119860 119861)=

1 if (119878 119879) ⊑ (119860 119861) and (119860 119861) = (0 0)0 otherwise

(6)

The inferior unanimity game Q(119878119879)(119860 119861) | 3119873 997888rarr R isdefined by

Q(119878119879) (119860 119861)=

minus1 if (119860 119861) ⊑ (119878 119879) and (119860 119861) = (0 0)0 otherwise

(7)

Voting game is a cooperative game model that incorporateselements from social choice theory This game model can beused to analyze situations where voters would vote in favor ofor against a decision [11] Simple voting game with abstentioncan be formulated as a bicooperative game It has been studiedunder the name of ternary voting game [19 20]

Let N be a set of n voters that has to choose between apair of two alternatives 119878 and 119879 Every voter 119894 isin 119873 has apreference over the two alternatives 119878 and 119879 Thus the payoffof voter 119894 (Q119894) is defined by Q119894 3119873 997888rarr minus1 0 1 based onthe majority rule Formally a ternary voting game Q isin BG119873

satisfies the following conditions [19]

(I) For every bicoalition pair (119878 119879)isin 3119873 its worthQ(119878 119879) isin minus1 0 1

(II) If (119878 119879) (119878 ) isin 3119873 with (119878 119879) ⊑ (119878 ) thenQ(119878 119879) le Q(119878 )

(III)Q119894

=

1 if (119894 isin 119878 and |119878| gt |119879|) or (119894 isin 119879 and |119879| gt |119878|)minus1 if (119894 isin 119878 and |119878| lt |119879|) or (119894 isin 119879 and |119879| lt |119878|)0 if voter i abstains form voting or |119878| = |119879|

(8)

where |119878| is the cardinality of coalition 119878 For voting gamesthe analysis of voting power is the main issue In 1965Banzhaf value was introduced by John F Banzhaf III based onprobabilistic analysis of the individual voters in a block votingsystem It depends on the number of ways in which eachvoter can effect a swing in the game In 2010 J Bilbao et alintroduced a pseudo-Banzhaf value ie ΨQ

119894 for the ternaryvoting game Q it can be interpreted as a voterrsquos probabilisticpower index [19 20]

ΨQ119894 = 13119899minus1 times sum

(119878119879)isin3119873119894

(Q119894 (119878 cup 119894 119879) minus Q119894 (119878 119879 cup 119894)) (9)

The pseudo-Banzhaf value satisfies four axioms Null PlayerTotal Swings Transfer Property and Simple Additivity[19 20]

Axiom 1 (null player) If 119894 isin 119873 is a null player in Q isin BG119873then ΨQ

119894 = 0 Therefore if the player 119894 is the null player in Q

with any (119878 119879) isin 3119873119894 thenQ119894 (119894 0) = (Q119894 (119878 cup 119894 119879) minus Q119894 (119878 119879))Q119894 (0 119894) = (Q119894 (119878 119879) minus Q119894 (119878 119879 cup 119894))

120601119894 (Q) = (Q(119878119879) (119894 0) minus Q(119878119879) (0 119894))(10)

Axiom 2 (total swings) If Q isin BG119873 then

sum119894isin119873

ΨQ119894 = sum(119878119879)isin3119873119894

(Q119894 (119878 cup 119894 119879) minus Q119894 (119878 119879 cup 119894)) (11)

Axiom 3 (transfer property) For any QG isin BG119873 we havethat

(120601119894 (Q) + 120601119894 (G)) = (120601119894 (Q⋁G) + 120601119894 (Q andG)) (12)

Axiom 4 (simple additivity) For any Q isin BG119873 such thatQ = G +H we have that

120601119894 (Q) = (120601119894 (G) + 120601119894 (H)) (13)

As a special ternary voting gamemodel the weighted ternaryvoting game can formulate a decision-making process thatnot all voters have the same amount of influence overthe decision Therefore votes of different voters are givendifferent weight This type of voting game model is usedin shareholder meetings where votes are weighted by thenumber of shares that each shareholder owns [11] In thisstudy we develop a ternary voting game model based onthe weighted voting approach In our game model mobiledevices are treated differently according to their socialpowers (F119898(119909)) and give different amounts of influenceto different members Therefore each coalition of players119878 sube 119873 has the sum of voting weights of its compo-nents that is W(119878) = sum119898119894isin119878F119898119894(119909) If the value ofW(119878) exceeds a certain quota players in 119878 win the game[11 19]

For the weighted ternary voting game we need twoquotas one is to block a decision and the other is toapprove it Formally it is represented by the voting body[P 120599F1198981(119909) F119898119899(119909)] where P is the quota for thedecision-taking and 120599 is the quota for the decision-blockingit can be seen as a particular case of n-player games with

Wireless Communications and Mobile Computing 7

three choices This situation can be formulated by a weightedternary voting game (X) ieX(119878 119879) 3119873 997888rarr minus1 0 1 [12]

X (119878 119879) =

1 if W (119878) ge P and W (119879) lt 120599minus1 if W (119878) lt P and W (119879) ge 1205990 otherwise

(14)

st P = 120572 times ( 119899sum119894=1

F119898119894 (119909))

and 120599 = 120573 times ( 119899sum119894=1

F119898119894 (119909))(15)

23 Communication and Computation Resource AllocationMechanisms Wireless bandwidth is a valuable and scarceresource in the SFIoT system Therefore the limited wirelessbandwidth has to be shared fair-efficiently by mobile usersIf the allocation method is not considered explicitly at thedesign stage of bandwidth allocation algorithms differentbandwidth requests can result in very unfair bandwidthallocations it may lead to a system inefficiency Usuallyvarious bandwidth allocation methods can be categorizedinto three ways proportional allocation (PA) constrainedallocation (CA) and equal loss allocation (EA) ways

(I) Proportional allocation (PA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max(119877119894 times Bsum119899119894=1 119877119894) m119894 otherwise

(16)

(II) Constrained allocation (CA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max (min 119877119894 119879119862 m119894) otherwise

st119899sum119894=1

min 119877119894 119879119862 = B

(17)

(III) Equal loss allocation (EA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max (119877119894 minusL119862) m119894 otherwise

st119899sum119894=1

max (119877119894 minusL119862) 0 = B

(18)

where B is the total available bandwidth amount and m119894is the predefined minimum bandwidth amount for the 119894119905ℎ

service 1198771le119894le119873 is the bandwidth request of mobile device119898119894 and 1198601le119894le119873 represents the decided bandwidth amountfor 119898119894 119879119862 and L119862 are the constraint thresholds which arechosen to satisfy each condition If the available bandwidthis enough or the F-AP can secure the new request 119860 119894from 119898119894 through the allocated bandwidth readjustment thenew requested application is accepted Otherwise the newrequested application is rejected

In our bandwidth allocation algorithm each F-APdecides its resource allocation way based on the weightedternary voting game InF119860119875119896 consider theX gamewithNF119860119875

119896

Individual 119898119894 isin NF119860119875119896

participates in the voting procedurewith his voting weight Usually the PA way is the mostcommon method and it can be seen as implicit consentTherefore for each voter we can assume that there are threeoptions ie pro-CA pro-EA and abstention Accordingto (14) the coalition 119878 consists of pro-CA voters and thecoalition 119879 consists of pro-EA voters If the outcome ofX(119878 119879) is 1 or -1 F119860119875119896 simply adopts the CA way or the EAway respectively in the bandwidth allocation process If theoutcome ofX(119878 119879) is 0F119860119875119896 adopts the PAway Based on thedecisions of individual devices Each F-AP adaptively decidesits communication resource allocation way

Like as the communication resource computation capac-ity in each F-AP is also a naturally limited resourceThereforethe computation power of F-AP is adaptively assigned to eachrequested application task Usually different applications canbe categorized into two classes class I (real-time data applica-tions) and class II (non-real-time data applications) accord-ing to the required QoE They not only require differentcomputation capacity but also have different QoE In orderto design the computation resource allocation algorithmeach F-AP needs to partition its own computation powerto support different type applications In this paper we alsoadopt the weighted ternary voting game model However incontrast to the voting game in our communication allocationalgorithm the developed computation allocation algorithmhas focused on the basic concept of pseudo-Banzhaf values

Consider another X game inF119860119875119896 with NF119860119875119896 Individual119898119894 isin NF119860119875

119896participates in the second voting procedure At

this time there are three options for each voter ie proclass Iservice proclass II service and abstention According to (14)the coalition 119878 consists of pro-class I service voters and thecoalition 119879 consists of pro-class II service voters Based on theoutcome of X(119878 119879) the F-AP estimates the preference ratio(120572119868) of class I application services as follows

120572119868= sum119898119894isin119878 (ΨQ

119898119894timesF119898119894 (119909))

(sum119898119894isin119878 (ΨQ119898119894timesF119898119894 (119909)) + sum119898119895isin119879 (ΨQ

119898119895timesF119898119894 (119909)))

(19)

According to each userrsquos pseudo-Banzhaf value and socialpower the 120572119868 value is decided and each F-AP partitions itstotal computation capacity (C) The division ratio for class I(or class II) type application services is 120572119868 (or 1 minus 120572119868) Andthen the partitioned computation capacity is distributed forits corresponding type application services while maximizing

8 Wireless Communications and Mobile Computing

Table 2 Application service type and requirements and system parameters

Type Application Bandwidthrequirement

Minimumrequirement (m)

Computationrequirement Duration (Avesec)

Class I Application I 119860=512 Mbps 119860=256 Mbps 98 MHz 180 secApplication II 119860=640 Mbps 119860=320 Mbps 256MHz 120 sec

Class II

Application III 119860=768 Mbps 119860=384 Mbps 128 MHz 240 secApplication IV 119860=896 Mbps 119860=458 Mbps 256MHz 300 secApplication V 119860=128 Gbps 119860=640 Mbps 384 MHz 360 secApplication VI 119860=152 Gbps 119860=860 Mbps 512 MHz 480 sec

Parameter Value DescriptionΓ 5 the maximum outcome of a user to normalize the social power120583 09 a parameter to controls the speed of social power saturation120572 075 a factor to decide the quota for the decision-taking120573 025 a factor to decide the quota for the decision- blocking120578119868 09 a control parameter to adjust the payoff of class I service120578119868119868 07 a control parameter to adjust the payoff of class II service120576 10 a control parameter for the sharing activity between social friends

the systemperformance To implement this process we adoptthe fundamental notion of the Nash Bargaining Solution(NBS) This is captured by the following optimization prob-lem

max[119860119898119894 ]

prod119898119894isin119878

119880119868119898119894 (119860119898119894) + max[119860119898119895 ]

prod119898119895isin119879

119880119868119868119898119895 (119860119898119895) (20)

st 119880119868119898119894 (119860119898119894) = 120578119868 times log2 (1 + 119860119898119894119877119898119894 )

119880119868119868119898119895 (119860119898119895) = 120578119868119868 times (1 minus exp(minus119860119898119894119877119898119894 ))sum119898119894isin119878

119860119898119894 le (120572119868 timesCF119860119875119896)

and sum119898119895isin119879

119860119898119895 le ((1 minus 120572119868) timesCF119860119875119896)

(21)

where 120578119868 (or 120578119868119868) is a control parameter to adjust the payoffof 119898119894 (or 119898119895) and CF119860119875

119896is the total computation capacity of

F119860119875119896

24 Main Steps of Proposed SFIoT Resource AllocationScheme Under the vision of the future internet the SIoTparadigm aims to connect anything while relying on the self-establishment of inter-thing social relationships In additionthe emerging paradigm of FC exhibits the right features forcoping with the aforementioned technological 5G networkissues In order to address this challenge the SFIoT paradigmhas been gaining momentum To effectively operate theSFIoT system the interactive relationship between F-APs andmobile devices is an important research topic and should beconsidered to design the resource allocation algorithms Inthis study we provide the communication and computationresource allocation scheme which is formulated according

to the BVG and NBS Owing to our two-phase game modelwe can get the most fair-efficient system performance bycombining both game approaches The main steps of theproposed scheme are described as follows

Step 1 Application types requirements and system controlparameters are determined by the simulation scenario (seesimulation assumptions in Section 3 and Table 2)

Step 2 Each individual F-AP monitors its covering area andresponds to its corresponding mobile devices Mobile devicesgenerate their application service requests in a distributedfashion and ask FC services to their corresponding F-AP

Step 3 Each mobile device has its own social relationshipthrough a social network close friends can probabilisticallyshare the outcome of FC services

Step 4 To reduce computation complexity individual F-APsexecute separately their two-phase games in parallel

Step 5 During the first game process each mobile devicetakes two votes one is to decide the communication resourceallocation way and the other is to partition the F-APrsquoscomputation capacity

Step 6 According to (14) the F-AP decides its communi-cation allocation way Based on our weighted ternary BVGmodel one of the PA CA and EA ways is adaptivelyselected And then the F-AP estimates the division ratio ofcomputation capacity using (19)

Step 7 During the second game process each F-AP inde-pendently distributes the partitioned computation capacity tocorresponding type applications According to (20) differentapplications can retain their FC services while maximizingthe system performance

Wireless Communications and Mobile Computing 9

Step 8 Based on the two-phase game model F-APs andmobile devices are hierarchically interconnected and inter-acting with one another to operate the SFIoT system

Step 9 Constantly each F-AP is self-monitoring its coveringarea When the game players are changed due to the mobiledevice migration the proposed scheme proceeds to Step 5 forthe next two-phase game procedure

3 Simulation Results and Discussion

In this section we evaluate the effectiveness of our proposedprotocol and analyze the simulation result while compar-ing it with that of the SDFC [14] SSEC [15] and CFIC[16] schemes To ensure a fair comparison the experi-mental setup consists of the following specifications andscenario

(i) Simulated SFIoT system consists of 25 F-APs and 100mobile devices F-APs are laid out in regular patternandmobile is randomly located in the F-APsrsquo coveringareas

(ii) For the FC services total computation capacity ofeach F-AP is 30 GHz and total communicationresource of each F-AP is 50 Gbps

(iii) Each mobile device individually generates servicerequests Service type of applications is randomlydecided and service request rate is Poisson process(120588) The offered rate range is varied from 0 to 30

(iv) The social friend number (|S119898| gt 0) of each mobiledevice is a random variable which has the maximumprobability at 3 while following a normal distribution

(v) Each social tie (0 le 119890119898119898 le 1) between mobiledevices is also a random variable which has themaximum probability at 05 while following a normaldistribution

(vi) The probability (P119904) that social friends119898119894 and119898119895 canshare the service outcome is defined asP119904 = 119890119898119894119898119895120576where 120576 is a control parameter for the sharing activity

(vii) We assume that there are no physical obstacles in theexperiments and each mobile device can contact itscorresponding F-AP for FC services

(viii) For calculation simplicity the communicationresource is specified in terms of basic units (BUs)where one BU is the minimum amount (eg 128Mbps in our system) of resource management

(ix) Using the iterative water-filling method the value ofNBS in (20) is obtained

(x) Network performancemeasures obtained on the basisof 100 simulation runs are plotted as functions of theoffered service request rate (120588)

(xi) Six different applications are assumed based on thedifferent computation and communication require-ment they are generated with equal probability

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

05 1 15 2 25 30Offered Application Load (Task Request Rate)

0

05

1

15

Qua

lity

of E

xper

ienc

e (Q

oE)

Figure 1 SFIoT quality of experience (QoE)

(xii) Performance criteria obtained through simulation areQoE resource usability and task failure probabilitythese simulation metrics are evaluated mainly todemonstrate the validity of our proposed method

To facilitate the development and implementation of oursimulator Table 2 lists application service type requirementsand system control parameters

Figure 1 shows the QoE of each scheme in differentapplication load scenarios For this evaluationQoE is definedas the normalized resource loss ratio than the originallyrequested amount From the viewpoint of mobile users QoEis a key factor to evaluate their service satisfactions In theproposed scheme the system resource is differently allocatedbased on the voting approach According to the desirablefeatures of ternary voting approach the limited resourcecan be assigned while effectively reflecting the preferenceinclination of users Our two-phase game approach can wellorchestrate the SFIoT system infrastructure to maximize theQoE of mobile users The result of Figure 1 proves that ourtwo-step game method is a socially beneficial way to achievea better QoE than other existing schemes

The obtained results of task failure probabilities aredepicted in Figure 2 This performance criterion is stronglyrelated to the SFIoT system throughput which is estimatedbased on the ratio of successfully completed applications Asexpected the application load increases the available systemresource is exhausted to support new application servicesTherefore task failure probabilities increase proportionatelywith the application request rate The resulting curves inFigure 2 allow us to see that our proposed scheme has gaineda lower task failure probability than other existing schemes Itis therefore worth to say that under different application loadconditions our proposed scheme can perform excellentlyto maintain a superior system throughput based on the

10 Wireless Communications and Mobile Computing

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Task

Fai

lure

Pro

babi

lity

05 1 15 2 25 30Offered Application Load (Task Request Rate)

Figure 2 SFIoT task failure probability

0 05 1 15 2 25 3Offered Application Load (Task Request Rate)

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Syste

m R

esou

rce U

tiliz

atio

n

Figure 3 Resource utilization

self-controlled management policies It is a highly desirableproperty for the SFIoT system operations

Figure 3 illustrates the resource utilization of F-APs Wemeasure this performance metric to show and determinewhether our game-based approach can provide the systemefficiency As it was explained control decisions in theproposed scheme are made in an interactive two-phasegame model More specifically the proposed approach inthis paper can adapt the current SFIoT system conditionwhile dynamically adjusting the control decisions Whenthe application load is low (below 05) the performance ofeach scheme is not much difference As the application load

rate increases the resource utilization also increases All theschemes have similar trends however due to our approachrsquosflexibility adaptability and responsiveness to current SFIoTsystem conditions the F-AP resource in our proposed schemecan be used more efficiently under widely different anddiversified application load situations

Through simulation experiments the obtained resultsin Figures 1ndash3 clearly demonstrate that the suitability andfeasibility of our proposed scheme while revealing enhancedperspectives regarding QoE throughput and resource uti-lization in the SFIoT system In particular the BVG and NBSprocesses work together to reach an agreement that givesmutual advantages for mobile devices and F-APs two gamemodels are sophisticatedly combined and dependent on eachotherTherefore they act cooperatively to satisfy the differentperformance requirements

4 Summary and Conclusions

The past decade has witnessed an explosive growth of IoTwhere a variety of smart mobile devices offer a plethora ofdifferent applications Under the vision of future networksFC and social network principles are integrated into the IoTparadigm in order to enforce the distributed self-cooperationand self-management of mobile devices In this paper wepresent a novel resource allocation scheme for the SFIoTsystem According to the BVG and NBS approaches ourproposed scheme is designed as two phases (i) mobiledevices take votes to decide the resource allocation methodsbased on the weighted ternary BVG model and (ii) the par-titioned computation capacity is fair-efficiently distributed toeach application by using the NBS These two game modelscontinue iteratively in a step-by-step manner in accordancewith the current SFIoT system informationThe experimentalsimulation results prove that our proposed protocol canefficiently improve theQoE system throughput and resourceutilization compared with the existing SDFC SSEC andCFIC schemes As directions for future research we aim atinvestigating the privacy and energy issues for the SFIoTsystemWe especially plan to add a blockchain-based securitylayer to the system In addition we also plan to develop anoncooperative gamemodelwith theoretical analysis andwillfocus on the addition of more features to increase the SFIoTsystem reliability in an unconstrained environment Lastbut not least machine learning techniques can be appliedto complement our proposed scheme It will be a potentialdirection and another possible extension to this work

Data Availability

The data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

The author declares that there are no conflicts of interestregarding the publication of this paper

Wireless Communications and Mobile Computing 11

Authorsrsquo Contributions

The sole author contributes to all research work

Acknowledgments

This research was supported by the MSIT (Ministry ofScience and ICT) Korea under the ITRC (InformationTechnology Research Center) support program (IITP-2018-2018-0-01799) supervised by the IITP (Institute for Infor-mation amp communications Technology Promotion) and byInstitute for Information amp communications TechnologyPromotion (IITP) grant funded by the Korea government(MSIT) (No2017-0-00498 A Development of Deidentifica-tion Technique based on Differential privacy)

References

[1] A M Ortiz D Hussein S Park S N Han and N Crespi ldquoThecluster between internet of things and social networks reviewand research challengesrdquo IEEE Internet of ings Journal vol 1no 3 pp 206ndash215 2014

[2] Z Li R Chen L Liu andGMin ldquoDynamic resource discoverybased on preference andmovement pattern similarity for large-scale social internet of thingsrdquo IEEE Internet of ings Journalvol 3 no 4 pp 581ndash589 2016

[3] H Yang and L Zhang ldquoResponse time prediction of IoT servicebased on time similarityrdquo Journal of Computing Science andEngineering vol 11 no 3 pp 100ndash108 2017

[4] E Baccarelli M Scarpiniti P G Naranjo and L Vaca-Cardenas ldquoFog of social IoT when the fog becomes socialrdquoIEEE Network vol 32 no 4 pp 68ndash80 2018

[5] H Shah-Mansouri and V W S Wong ldquoHierarchical fog-cloudcomputing for iot systems a computation offloading gamerdquoIEEE Internet ofings Journal vol 5 no 4 pp 3246ndash3257 2018

[6] K Lee and I Shin ldquoUser mobility model based computationoffloading decision for mobile cloudrdquo Journal of ComputingScience and Engineering vol 9 no 3 pp 155ndash162 2015

[7] Y Liu M J Lee and Y Zheng ldquoAdaptive multi-resourceallocation for cloudlet-based mobile cloud computing systemrdquoIEEE Transactions on Mobile Computing vol 15 no 10 pp2398ndash2410 2016

[8] K Kim S Uno and M Kim ldquoAdaptive QoS mechanism forwireless mobile networkrdquo Journal of Computing Science andEngineering vol 4 no 2 pp 153ndash172 2010

[9] A Pande V Ramamurthi and P Mohapatra ldquoQuality-orientedvideo delivery over LTErdquo Journal of Computing Science andEngineering vol 7 no 3 pp 168ndash176 2013

[10] S Kim and P K Varshney ldquoAn integrated adaptive bandwidth-management framework for QoS-sensitive multimedia cellularnetworksrdquo IEEE Transactions on Vehicular Technology vol 53no 3 pp 835ndash846 2004

[11] K SungwookGameeory Applications in Network Design IGIGlobal Hershey PA USA 2014

[12] J M Bilbao J R Fernandez N Jimenez and J J LopezldquoBiprobabilistic values for bicooperative gamesrdquo DiscreteApplied Mathematics vol 156 pp 2698ndash2711 2008

[13] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheshapley value for bicooperative gamesrdquo Annals of OperationsResearch vol 158 pp 99ndash115 2008

[14] L Liu Z Chang and X Guo ldquoSocially-aware dynamic compu-tation offloading scheme for fog computing system with energyharvesting devicesrdquo IEEE Internet of ings Journal vol 5 no3 pp 1869ndash1879 2018

[15] R Dautov S Distefano D Bruneo F Longo G Merlino andA Puliafito ldquoData processing in cyber-physical-social systemsthrough edge computingrdquo IEEE Access vol 6 pp 29822ndash298352018

[16] Y-C Chen Y-C Chang C-H Chen Y-S Lin J-L Chen andY-Y Chang ldquoCloud-fog computing for information-centricinternet-of-things applicationsrdquo in Proceedings of the 2017 IEEEInternational Conference on Applied System Innovation ICASI2017 pp 637ndash640 May 2017

[17] C Jiang L Kuang Z Han Y Ren and L Hanzo ldquoInformationcredibility modeling in cooperative networks equilibrium andmechanism designrdquo IEEE Journal on Selected Areas in Commu-nications vol 35 no 2 pp 432ndash448 2017

[18] S Kim ldquoFog radio access network system control scheme basedon the embedded game modelrdquo EURASIP Journal on WirelessCommunications and Networking vol 2017 no 113 pp 1ndash142017

[19] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheBanzhaf power index for ternary bicooperative gamesrdquoDiscreteApplied Mathematics vol 158 pp 967ndash980 2010

[20] M TsurumiM Inuiguchi and A Nishimura ldquoPseudo-banzhafvalues in bicooperative gamesrdquo in Proceedings of the IEEEInternational Conference on Systems Man and Cybernetics pp1138ndash1143 2005

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Active and Passive Electronic Components

VLSI Design

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Shock and Vibration

Hindawiwwwhindawicom Volume 2018

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

SensorsJournal of

Hindawiwwwhindawicom Volume 2018

International Journal of

RotatingMachinery

Hindawiwwwhindawicom Volume 2018

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Navigation and Observation

International Journal of

Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom

Wireless Communications and Mobile Computing 3

queuemodels are also applied Finally it is addressed by usingthe semi-smooth Newton method with Armijo line searchThe SDFC scheme can reduce the accumulated error andadaptively improves the calculation accuracy during iterationprocess The simulation results demonstrate the effectivenessof the SDFC scheme [14]

The social system through edge computing (SSEC) schemeis proposed as a new collaborative horizontal offloadingmethod for the distributed data processing mechanism [15]To discover and select devices which are able to performan offloaded task according to specific requirements thisscheme implements the clustered edge computing paradigmCurrently personal mobile devices are seen not only aspassive data generators and IoT service consumers butrather as active participants and contributors Based on theprinciples of volunteer computing and SIoT the SSEC schemepresents a novel approach to address network latency issuesat the very edge of an IoT network topology while utilizingidle resources of mobile devices Finally the simulationresults show that the SSEC scheme has a potential power tooutperform cloud-centric setups by keeping the computationlocally and by involving volunteering mobile devices inclustered computation at the edge [15]

Paper [16] presents the cloud-fog information-centric com-puting (CFIC) schemeThis scheme supports IoT applicationsin various network domains with the IoT middleware andthe cloud-fog computingmechanism By using two functionsie job-classification function and resource scheduling func-tion QoE in IoT applications can be ensured Specifically thejob-classification function classifies different IoT applicationsby authority data type data update rate and priority Theresource scheduling function generates a resource allocationplan based on the resource limitation Simulation resultsreveal that the CFIC scheme can reduce the cloud computingtime while guaranteeing the execution of real-time IoTapplications [16]

The earlier studies [14ndash16] have attracted considerableattentions while introducing unique challenges in handlingthe SFIoT resource allocation problem Therefore our pro-posed scheme may look similar to the existing schemesWhile these schemes have some similarities there are severalkey differences First our proposed scheme is designedas a two-phase game model based on the two differentcooperative game approaches Second in our game modelthe main concepts of NBS and BVG are adopted to addressthe SFIoT resource allocation problem Third to reduce thecomputation complexity we hierarchically implement theBVG and the NBS with cascade interactions Most of allthe main difference between the existing schemes in [14ndash16] and our proposed scheme is a control paradigm Theprinciple novelty of our protocol is a judicious mixture of twogame solutions and its feasible self-adaptability in the real-world SFIoT systemoperations In this paper we compare ourproposed scheme with the existing the SDFC [14] SSEC [15]and CFIC [16] schemes and demonstrate that our two-phaseinteractive game approach can significantly outperform theseexisting schemes

The main contributions of this study are (i) ability tomaintain the SFIoT resource efficiency as high as possible

(ii) ability to respond to current SFIoT situations basedon the interactive bargaining and voting process (iii) QoEprovisioning for different kinds of application tasks (iv)synergy effect based on the reciprocal combination of SIoTand FC and (v) ensuring a mutually desirable solution whileimproving the overall SFIoT system performance The keyadvantage of the proposed scheme is its adaptability feasi-bility and effectiveness for realistic SFIoT system operationsunder widely different and diversified application task loadsituations

The rest of this paper is organized as follows Section 2presents the SFIoT system infrastructure and the backgroundknowledge ofNBS andBVGconcepts And then the details ofour proposed scheme are explained based on the interactivetwo-phase game model In Section 3 we provide the numer-ical results from simulation experiments and demonstratethe effectiveness of the proposed scheme comparing withthe existing SDFC [14] SSEC [15] and CFIC [16] protocolsFinally conclusions and suggestions for future research aregiven in Section 4

2 The Proposed SFIoT ResourceAllocation Algorithms

In this section we describe a new two-phase game approachfor solving the resource allocation problem of SFIoT systemIn the first phase mobile devices take two votes to decidecommunication and computation resource allocation waysIn the second phase the computation capacity is distributedto each application task based on the NBS Finally weexplain in detail the proposed scheme in the nine-stepprocedures

21 Fog Controlled Social IoT System Infrastructure Tradi-tionally the SFIoT system infrastructure is consisting of mul-tiple cloudlets and a number of mobile devices As a small-scale cloud datacenter each cloudlet is located at the edge ofthe Internet The main purpose of the cloudlet is supportingresource-intensive and interactive mobile applications byproviding powerful computing resources with lower latencyIn the SFIoT system each individual mobile device is amember of both an underlying IoT and an overlay socialnetwork Usually the IoT accounts for more conventionalnetwork services and the social network reflects somewhatless conventional social ties induced by inter-thing socialrelationships which are formally described by a social graphG = (NE) N is the set of mobile devices and E = 119890119898119894119898119895 |0 le 119890119898119894119898119895 le 1 and (119898119894 119898119895) isin N times N is the set of the socialties 119890119898119894119898119895 is a real-valued nonnegative number to measurethe strength of the social tie between 119898119894 and 119898119895 ThereforeS119898119894 = 119898119895 | 119890119898119894119898119895 gt 0 and 119898119895 isin N formally defines the setof the social friends of119898119894 [4]

To calculate each userrsquos influential power we develop asocial power function F(119909) in line with some propertiesUnder diverse scenarios the function F(119909) should obeythe following three properties to effectively estimate eachindividual userrsquos social power [17]

4 Wireless Communications and Mobile Computing

(i) nonnegative property F(119909) ge 0(ii) nondecreasing property 119889F(119909)119889119909 ge 0(iii) saturation property 1198892F(119909)1198891199092 lt 0

The nonnegative property means that the outcome shouldbe larger than zero and if there is no social relationshipit is zero The nondecreasing property simply implies thatthe more users are socially connected the more outcomecan be obtained The saturation property suggests that whenthe social relationship is sufficient any further increase ofthe outcome remains marginal [17] In order to construct afunction while satisfying above three properties the socialpower function of119898119894 ieF119898119894(119909) is defined as follows

F119898119894 (119909) = Γ minus 1exp (120583 times 119909)

st 119909 = sum119898119895isinS119898119894

119890119898119894119898119895 (1)

where 120583 controls the speed of social power saturation and Γrepresents the maximum outcome of a user to normalize thesocial power [17]

In this study we assume that there are 119896 cloudlets and 119899mobile devices where 119896 ≪ 119899 In the SFIoT system infrastruc-ture cloudlets are evolved to the Fog-computing basedAccessPoint (F-AP) by being equipped with a certain computationpower and communication capability The design of F-APplatform has been introduced to deliver various devicesrequests which have different characteristics and resourcedemands that is CPU capacity for computation and wirelessbandwidth assignment for communication However dueto the limited CPU capacity and bandwidth scarcity in theF-AP it is impossible to guarantee all applicationsrsquo needs[18] Therefore an efficient resource management strategybecomes a key factor in enhancing the SFIoT performancewhile ensuring the required QoE

To tackle the SFIoT resource control problem we pro-pose a two-phase game based resource allocation schemeAt the first phase individual mobile devices participatein two BVGs one is to decide the allocation method ofcommunication resource and the other is to partition thecomputation resource according to the voting power of eachmobile device In each F-AP covering area the voting power iscalculated based on the pseudo-Banzhaf value At the secondphase the partitioned computation capacity is distributedfor same kinds of application tasks based on the NBS Topractically implement our resource allocation algorithmseach individual F-AP and its corresponding mobile devicesoperate their own two-phase game in an entirely distributedmanner For example the 119896th F-AP ie F119860119875119896 and the set ofcorresponding mobile devices in F119860119875119896 rsquos covering area ieNF119860119875119896 work together for their two-phase gameG119896 Therefore

multiple two-phase games G1le119894le119896 are operated in parallel Toformally define our proposed game model we characterizeG119896 as F119860119875119896 NF119860119875

119896 (S119898119894 F119898119894(119909) ΨQ

119898119894) | 119898119894 isin NF119860119875

119896 119909 =sum119898119895isinS119898119894 119890119898119894119898119895AF119860119875

119896 120572F119860119875119896 119880F119860119875

119896 Table 1 lists the notations

used in this paper

(i) F119860119875119896 NF119860119875119896 is the set of gameplayers for theG119896 game

where NF119860119875119896cap NF119860119875

119897= 0

(ii) (S119898119894 F119898119894(119909) ΨQ119898119894) is a three-pair for the 119898119894 isin NF119860119875

119896

where S119898119894 is the set of119898119894rsquos social friendsF119898119894(119909) andΨQ119898119894

represent the social power and the voting powerof119898119894 respectively

(iii) AF119860119875119896

is the set of generated application tasks whichare offloaded in theF119860119875119896 for FC services

(iv) 120572F119860119875119896

is the ratio of relative service weights givento different type applications in the F119860119875119896 where0 le 120572F119860119875

119896le 1 Based on the 120572F119860119875

119896value F119860119875119896 rsquos

computation power is partitioned

(v) 119880F119860119875119896

is the utility function of F119860119875119896 it is optimizedbased on the NBS

22 Bicooperative Game and Ternary Voting Model Usuallycooperative game theory studies situations where a group ofgame players are associated to obtain a profit as a result oftheir cooperation Thus a cooperative gamemodel is definedas a pair (119873V) where N is a finite set of players and V2119873 997888rarr R is a function verifying that V(0) = 0 For eachcoalition 119878 isin 2119873V(119878) can be interpreted as the coalition Srsquosmaximal gain that players in S can achieve themselves againstthe best offensive threat by the complementary coalition119873119878[12]

In 2008 Bilbao et al proposed a new idea called bicoop-erative game for themulticriteria decision-making process Inthe bicooperative game game players are allowed to chooseone among three alternatives To think the voting situationby analogy we consider ordered pairs (S T) with 119878 119879 sube119873 and 119878 cap 119879 = 0 Each pair (S T) yields a partition of119873 in three groups Players in S are in favor of the votingitem and players in T object to it The remaining players in119873 (119878 cup 119879) are nonvoters Formally we can define the set3119873 = (119878 119879) | 119878 119879 sube 119873 119878 cap 119879 = 0 and the functionQ 3119873 997888rarr R For each (119878 119879) isin 3119873 the outcome of Q(119878 119879)can be interpreted as the gain that 119878 can achieve when 119879 isthe opposite coalition and119873 (119878 cup 119879) is the neutral coalition[12 13]

The pair (0119873) represents the situation if all the playersobject to the voting item and (119873 0) represents the situationwhere all the players support the voting item More generallya relation of 3119873 is given by [12 13]

(119860 119861) ⊑ (119862119863) lArrrArr 119860 sube 119862 119861 supe 119863 (2)

The set (3119873 ⊑) is a finite distributive lattice and a partiallyordered set with the following properties [12 13]

(i) (0119873) is the first element (0119873) ⊑ (119860 119861) for all(119860 119861) isin 3119873(ii) (119873 0) is the last element (119860 119861) ⊑ (119873 0) for all(119860 119861) isin 3119873

Wireless Communications and Mobile Computing 5

Table 1 Symbols and parameters used in the proposed algorithm

Notations ExplanationN the set of mobile devicesE the set of the social ties119890119898119894 119898119895 a real-valued number to measure the strength of the social tie between119898119894 and119898119895S119898119894 the set of the social friendsof119898119894F119898119894 (119909) the social power function of 119898119894120583 a parameter to control the speed of social power saturationΓ the maximum outcome of a user to normalize the social powerF119860119875119896 the 119896th fog-computing based access pointNF119860119875119896

the set of corresponding mobile devices in theF119860119875119896 rsquos covering areaG119896 The two-phase game inF119860119875119896ΨQ119898119894

the social power and the voting power of 119898119894AF119860119875119896

the set of application tasks which are offloaded in theF119860119875119896 for FC services120572F119860119875119896

the ratio of relative service weights given to different type applications in theF119860119875119896119880F119860119875119896

the utility function ofF119860119875119896119873 a finite set of voting game playersV (119878) the coalition Srsquos maximal gainQ (119878 119879) the gain that 119878 can achieve when 119879 is the opposite coalitionBG119873 the set of all bicooperative games with players119873120601119894 (Q) the player 119894rsquos value in a game Q onBG119873

Q(119878119879) (119860 119861) the identity gameQ(119878119879) (119860 119861) the superior unanimity gameQ(119878119879) (119860 119861) the inferior unanimity gameQ119894 the payoff of voter 119894W (119878) the sum of voting weights of coalition 119878rsquos playersP the quota for the decision-taking120599 the quota for the decision-blockingX a weighted ternary voting gameB the total available bandwidth amountm119894 the pre-defined minimum bandwidth amount for the 119894119905ℎ service119877119894 the bandwidth request of mobile device119898119894119860 119894 the decided bandwidth amount for the119898119894119879119862L119862 the constraint thresholds which are chosen to satisfy each condition120572119868 the preference ratio of class I application servicesC each F-APrsquos total computation capacity120578119868120578119868119868 control parameters to adjust the payoff of119898CF119860119875119896

the total computation capacity ofF119860119875119896

(iii) Every fair of elements of 3119873 has a join(119860 119861)⋁ (119862119863) = (119860 cup 119862 119861 cap 119863)

and (119860 119861) and (119862119863) = (119860 cap 119862 119861 cup 119863) (3)

The set of all bicooperative games with players119873 is denotedbyBG119873 which is a real vector space [12 13]

BG119873 = Q | 3119873 997888rarr R and Q (0 0) = 0 (4)

A value on BG119873 is a mapping 120601 BG119873 997888rarr R119899 thatassociates with each game Q isin BG119873 In a vector (1206011(Q) 120601119899(Q)) isin R119899 1206011le119894le119899(Q) represents a real number whichis the player 119894rsquos value in a game Q on BG119873 Therefore themapping 120601119894isin119873(Q) BG119873 997888rarr R represents the player 119894rsquospayoff from playing bicooperative games [12 13] There arethree special collections of games in BG119873 taking values inminus1 0 1 the identity games the superior unanimity gamesand the inferior unanimity games which are defined for any(119878 119879) isin 3119873 [12 13]

The identity game Q(119878119879)(119860 119861) | 3119873 997888rarr R is defined by

Q(119878119879) (119860 119861)

6 Wireless Communications and Mobile Computing

= 1 if (119860 119861) = (119878 119879) and (119878 119879) = (0 0)0 otherwise

(5)

The superior unanimity game Q(119878119879)(119860 119861) | 3119873 997888rarr R isdefined by

Q(119878119879) (119860 119861)=

1 if (119878 119879) ⊑ (119860 119861) and (119860 119861) = (0 0)0 otherwise

(6)

The inferior unanimity game Q(119878119879)(119860 119861) | 3119873 997888rarr R isdefined by

Q(119878119879) (119860 119861)=

minus1 if (119860 119861) ⊑ (119878 119879) and (119860 119861) = (0 0)0 otherwise

(7)

Voting game is a cooperative game model that incorporateselements from social choice theory This game model can beused to analyze situations where voters would vote in favor ofor against a decision [11] Simple voting game with abstentioncan be formulated as a bicooperative game It has been studiedunder the name of ternary voting game [19 20]

Let N be a set of n voters that has to choose between apair of two alternatives 119878 and 119879 Every voter 119894 isin 119873 has apreference over the two alternatives 119878 and 119879 Thus the payoffof voter 119894 (Q119894) is defined by Q119894 3119873 997888rarr minus1 0 1 based onthe majority rule Formally a ternary voting game Q isin BG119873

satisfies the following conditions [19]

(I) For every bicoalition pair (119878 119879)isin 3119873 its worthQ(119878 119879) isin minus1 0 1

(II) If (119878 119879) (119878 ) isin 3119873 with (119878 119879) ⊑ (119878 ) thenQ(119878 119879) le Q(119878 )

(III)Q119894

=

1 if (119894 isin 119878 and |119878| gt |119879|) or (119894 isin 119879 and |119879| gt |119878|)minus1 if (119894 isin 119878 and |119878| lt |119879|) or (119894 isin 119879 and |119879| lt |119878|)0 if voter i abstains form voting or |119878| = |119879|

(8)

where |119878| is the cardinality of coalition 119878 For voting gamesthe analysis of voting power is the main issue In 1965Banzhaf value was introduced by John F Banzhaf III based onprobabilistic analysis of the individual voters in a block votingsystem It depends on the number of ways in which eachvoter can effect a swing in the game In 2010 J Bilbao et alintroduced a pseudo-Banzhaf value ie ΨQ

119894 for the ternaryvoting game Q it can be interpreted as a voterrsquos probabilisticpower index [19 20]

ΨQ119894 = 13119899minus1 times sum

(119878119879)isin3119873119894

(Q119894 (119878 cup 119894 119879) minus Q119894 (119878 119879 cup 119894)) (9)

The pseudo-Banzhaf value satisfies four axioms Null PlayerTotal Swings Transfer Property and Simple Additivity[19 20]

Axiom 1 (null player) If 119894 isin 119873 is a null player in Q isin BG119873then ΨQ

119894 = 0 Therefore if the player 119894 is the null player in Q

with any (119878 119879) isin 3119873119894 thenQ119894 (119894 0) = (Q119894 (119878 cup 119894 119879) minus Q119894 (119878 119879))Q119894 (0 119894) = (Q119894 (119878 119879) minus Q119894 (119878 119879 cup 119894))

120601119894 (Q) = (Q(119878119879) (119894 0) minus Q(119878119879) (0 119894))(10)

Axiom 2 (total swings) If Q isin BG119873 then

sum119894isin119873

ΨQ119894 = sum(119878119879)isin3119873119894

(Q119894 (119878 cup 119894 119879) minus Q119894 (119878 119879 cup 119894)) (11)

Axiom 3 (transfer property) For any QG isin BG119873 we havethat

(120601119894 (Q) + 120601119894 (G)) = (120601119894 (Q⋁G) + 120601119894 (Q andG)) (12)

Axiom 4 (simple additivity) For any Q isin BG119873 such thatQ = G +H we have that

120601119894 (Q) = (120601119894 (G) + 120601119894 (H)) (13)

As a special ternary voting gamemodel the weighted ternaryvoting game can formulate a decision-making process thatnot all voters have the same amount of influence overthe decision Therefore votes of different voters are givendifferent weight This type of voting game model is usedin shareholder meetings where votes are weighted by thenumber of shares that each shareholder owns [11] In thisstudy we develop a ternary voting game model based onthe weighted voting approach In our game model mobiledevices are treated differently according to their socialpowers (F119898(119909)) and give different amounts of influenceto different members Therefore each coalition of players119878 sube 119873 has the sum of voting weights of its compo-nents that is W(119878) = sum119898119894isin119878F119898119894(119909) If the value ofW(119878) exceeds a certain quota players in 119878 win the game[11 19]

For the weighted ternary voting game we need twoquotas one is to block a decision and the other is toapprove it Formally it is represented by the voting body[P 120599F1198981(119909) F119898119899(119909)] where P is the quota for thedecision-taking and 120599 is the quota for the decision-blockingit can be seen as a particular case of n-player games with

Wireless Communications and Mobile Computing 7

three choices This situation can be formulated by a weightedternary voting game (X) ieX(119878 119879) 3119873 997888rarr minus1 0 1 [12]

X (119878 119879) =

1 if W (119878) ge P and W (119879) lt 120599minus1 if W (119878) lt P and W (119879) ge 1205990 otherwise

(14)

st P = 120572 times ( 119899sum119894=1

F119898119894 (119909))

and 120599 = 120573 times ( 119899sum119894=1

F119898119894 (119909))(15)

23 Communication and Computation Resource AllocationMechanisms Wireless bandwidth is a valuable and scarceresource in the SFIoT system Therefore the limited wirelessbandwidth has to be shared fair-efficiently by mobile usersIf the allocation method is not considered explicitly at thedesign stage of bandwidth allocation algorithms differentbandwidth requests can result in very unfair bandwidthallocations it may lead to a system inefficiency Usuallyvarious bandwidth allocation methods can be categorizedinto three ways proportional allocation (PA) constrainedallocation (CA) and equal loss allocation (EA) ways

(I) Proportional allocation (PA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max(119877119894 times Bsum119899119894=1 119877119894) m119894 otherwise

(16)

(II) Constrained allocation (CA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max (min 119877119894 119879119862 m119894) otherwise

st119899sum119894=1

min 119877119894 119879119862 = B

(17)

(III) Equal loss allocation (EA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max (119877119894 minusL119862) m119894 otherwise

st119899sum119894=1

max (119877119894 minusL119862) 0 = B

(18)

where B is the total available bandwidth amount and m119894is the predefined minimum bandwidth amount for the 119894119905ℎ

service 1198771le119894le119873 is the bandwidth request of mobile device119898119894 and 1198601le119894le119873 represents the decided bandwidth amountfor 119898119894 119879119862 and L119862 are the constraint thresholds which arechosen to satisfy each condition If the available bandwidthis enough or the F-AP can secure the new request 119860 119894from 119898119894 through the allocated bandwidth readjustment thenew requested application is accepted Otherwise the newrequested application is rejected

In our bandwidth allocation algorithm each F-APdecides its resource allocation way based on the weightedternary voting game InF119860119875119896 consider theX gamewithNF119860119875

119896

Individual 119898119894 isin NF119860119875119896

participates in the voting procedurewith his voting weight Usually the PA way is the mostcommon method and it can be seen as implicit consentTherefore for each voter we can assume that there are threeoptions ie pro-CA pro-EA and abstention Accordingto (14) the coalition 119878 consists of pro-CA voters and thecoalition 119879 consists of pro-EA voters If the outcome ofX(119878 119879) is 1 or -1 F119860119875119896 simply adopts the CA way or the EAway respectively in the bandwidth allocation process If theoutcome ofX(119878 119879) is 0F119860119875119896 adopts the PAway Based on thedecisions of individual devices Each F-AP adaptively decidesits communication resource allocation way

Like as the communication resource computation capac-ity in each F-AP is also a naturally limited resourceThereforethe computation power of F-AP is adaptively assigned to eachrequested application task Usually different applications canbe categorized into two classes class I (real-time data applica-tions) and class II (non-real-time data applications) accord-ing to the required QoE They not only require differentcomputation capacity but also have different QoE In orderto design the computation resource allocation algorithmeach F-AP needs to partition its own computation powerto support different type applications In this paper we alsoadopt the weighted ternary voting game model However incontrast to the voting game in our communication allocationalgorithm the developed computation allocation algorithmhas focused on the basic concept of pseudo-Banzhaf values

Consider another X game inF119860119875119896 with NF119860119875119896 Individual119898119894 isin NF119860119875

119896participates in the second voting procedure At

this time there are three options for each voter ie proclass Iservice proclass II service and abstention According to (14)the coalition 119878 consists of pro-class I service voters and thecoalition 119879 consists of pro-class II service voters Based on theoutcome of X(119878 119879) the F-AP estimates the preference ratio(120572119868) of class I application services as follows

120572119868= sum119898119894isin119878 (ΨQ

119898119894timesF119898119894 (119909))

(sum119898119894isin119878 (ΨQ119898119894timesF119898119894 (119909)) + sum119898119895isin119879 (ΨQ

119898119895timesF119898119894 (119909)))

(19)

According to each userrsquos pseudo-Banzhaf value and socialpower the 120572119868 value is decided and each F-AP partitions itstotal computation capacity (C) The division ratio for class I(or class II) type application services is 120572119868 (or 1 minus 120572119868) Andthen the partitioned computation capacity is distributed forits corresponding type application services while maximizing

8 Wireless Communications and Mobile Computing

Table 2 Application service type and requirements and system parameters

Type Application Bandwidthrequirement

Minimumrequirement (m)

Computationrequirement Duration (Avesec)

Class I Application I 119860=512 Mbps 119860=256 Mbps 98 MHz 180 secApplication II 119860=640 Mbps 119860=320 Mbps 256MHz 120 sec

Class II

Application III 119860=768 Mbps 119860=384 Mbps 128 MHz 240 secApplication IV 119860=896 Mbps 119860=458 Mbps 256MHz 300 secApplication V 119860=128 Gbps 119860=640 Mbps 384 MHz 360 secApplication VI 119860=152 Gbps 119860=860 Mbps 512 MHz 480 sec

Parameter Value DescriptionΓ 5 the maximum outcome of a user to normalize the social power120583 09 a parameter to controls the speed of social power saturation120572 075 a factor to decide the quota for the decision-taking120573 025 a factor to decide the quota for the decision- blocking120578119868 09 a control parameter to adjust the payoff of class I service120578119868119868 07 a control parameter to adjust the payoff of class II service120576 10 a control parameter for the sharing activity between social friends

the systemperformance To implement this process we adoptthe fundamental notion of the Nash Bargaining Solution(NBS) This is captured by the following optimization prob-lem

max[119860119898119894 ]

prod119898119894isin119878

119880119868119898119894 (119860119898119894) + max[119860119898119895 ]

prod119898119895isin119879

119880119868119868119898119895 (119860119898119895) (20)

st 119880119868119898119894 (119860119898119894) = 120578119868 times log2 (1 + 119860119898119894119877119898119894 )

119880119868119868119898119895 (119860119898119895) = 120578119868119868 times (1 minus exp(minus119860119898119894119877119898119894 ))sum119898119894isin119878

119860119898119894 le (120572119868 timesCF119860119875119896)

and sum119898119895isin119879

119860119898119895 le ((1 minus 120572119868) timesCF119860119875119896)

(21)

where 120578119868 (or 120578119868119868) is a control parameter to adjust the payoffof 119898119894 (or 119898119895) and CF119860119875

119896is the total computation capacity of

F119860119875119896

24 Main Steps of Proposed SFIoT Resource AllocationScheme Under the vision of the future internet the SIoTparadigm aims to connect anything while relying on the self-establishment of inter-thing social relationships In additionthe emerging paradigm of FC exhibits the right features forcoping with the aforementioned technological 5G networkissues In order to address this challenge the SFIoT paradigmhas been gaining momentum To effectively operate theSFIoT system the interactive relationship between F-APs andmobile devices is an important research topic and should beconsidered to design the resource allocation algorithms Inthis study we provide the communication and computationresource allocation scheme which is formulated according

to the BVG and NBS Owing to our two-phase game modelwe can get the most fair-efficient system performance bycombining both game approaches The main steps of theproposed scheme are described as follows

Step 1 Application types requirements and system controlparameters are determined by the simulation scenario (seesimulation assumptions in Section 3 and Table 2)

Step 2 Each individual F-AP monitors its covering area andresponds to its corresponding mobile devices Mobile devicesgenerate their application service requests in a distributedfashion and ask FC services to their corresponding F-AP

Step 3 Each mobile device has its own social relationshipthrough a social network close friends can probabilisticallyshare the outcome of FC services

Step 4 To reduce computation complexity individual F-APsexecute separately their two-phase games in parallel

Step 5 During the first game process each mobile devicetakes two votes one is to decide the communication resourceallocation way and the other is to partition the F-APrsquoscomputation capacity

Step 6 According to (14) the F-AP decides its communi-cation allocation way Based on our weighted ternary BVGmodel one of the PA CA and EA ways is adaptivelyselected And then the F-AP estimates the division ratio ofcomputation capacity using (19)

Step 7 During the second game process each F-AP inde-pendently distributes the partitioned computation capacity tocorresponding type applications According to (20) differentapplications can retain their FC services while maximizingthe system performance

Wireless Communications and Mobile Computing 9

Step 8 Based on the two-phase game model F-APs andmobile devices are hierarchically interconnected and inter-acting with one another to operate the SFIoT system

Step 9 Constantly each F-AP is self-monitoring its coveringarea When the game players are changed due to the mobiledevice migration the proposed scheme proceeds to Step 5 forthe next two-phase game procedure

3 Simulation Results and Discussion

In this section we evaluate the effectiveness of our proposedprotocol and analyze the simulation result while compar-ing it with that of the SDFC [14] SSEC [15] and CFIC[16] schemes To ensure a fair comparison the experi-mental setup consists of the following specifications andscenario

(i) Simulated SFIoT system consists of 25 F-APs and 100mobile devices F-APs are laid out in regular patternandmobile is randomly located in the F-APsrsquo coveringareas

(ii) For the FC services total computation capacity ofeach F-AP is 30 GHz and total communicationresource of each F-AP is 50 Gbps

(iii) Each mobile device individually generates servicerequests Service type of applications is randomlydecided and service request rate is Poisson process(120588) The offered rate range is varied from 0 to 30

(iv) The social friend number (|S119898| gt 0) of each mobiledevice is a random variable which has the maximumprobability at 3 while following a normal distribution

(v) Each social tie (0 le 119890119898119898 le 1) between mobiledevices is also a random variable which has themaximum probability at 05 while following a normaldistribution

(vi) The probability (P119904) that social friends119898119894 and119898119895 canshare the service outcome is defined asP119904 = 119890119898119894119898119895120576where 120576 is a control parameter for the sharing activity

(vii) We assume that there are no physical obstacles in theexperiments and each mobile device can contact itscorresponding F-AP for FC services

(viii) For calculation simplicity the communicationresource is specified in terms of basic units (BUs)where one BU is the minimum amount (eg 128Mbps in our system) of resource management

(ix) Using the iterative water-filling method the value ofNBS in (20) is obtained

(x) Network performancemeasures obtained on the basisof 100 simulation runs are plotted as functions of theoffered service request rate (120588)

(xi) Six different applications are assumed based on thedifferent computation and communication require-ment they are generated with equal probability

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

05 1 15 2 25 30Offered Application Load (Task Request Rate)

0

05

1

15

Qua

lity

of E

xper

ienc

e (Q

oE)

Figure 1 SFIoT quality of experience (QoE)

(xii) Performance criteria obtained through simulation areQoE resource usability and task failure probabilitythese simulation metrics are evaluated mainly todemonstrate the validity of our proposed method

To facilitate the development and implementation of oursimulator Table 2 lists application service type requirementsand system control parameters

Figure 1 shows the QoE of each scheme in differentapplication load scenarios For this evaluationQoE is definedas the normalized resource loss ratio than the originallyrequested amount From the viewpoint of mobile users QoEis a key factor to evaluate their service satisfactions In theproposed scheme the system resource is differently allocatedbased on the voting approach According to the desirablefeatures of ternary voting approach the limited resourcecan be assigned while effectively reflecting the preferenceinclination of users Our two-phase game approach can wellorchestrate the SFIoT system infrastructure to maximize theQoE of mobile users The result of Figure 1 proves that ourtwo-step game method is a socially beneficial way to achievea better QoE than other existing schemes

The obtained results of task failure probabilities aredepicted in Figure 2 This performance criterion is stronglyrelated to the SFIoT system throughput which is estimatedbased on the ratio of successfully completed applications Asexpected the application load increases the available systemresource is exhausted to support new application servicesTherefore task failure probabilities increase proportionatelywith the application request rate The resulting curves inFigure 2 allow us to see that our proposed scheme has gaineda lower task failure probability than other existing schemes Itis therefore worth to say that under different application loadconditions our proposed scheme can perform excellentlyto maintain a superior system throughput based on the

10 Wireless Communications and Mobile Computing

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Task

Fai

lure

Pro

babi

lity

05 1 15 2 25 30Offered Application Load (Task Request Rate)

Figure 2 SFIoT task failure probability

0 05 1 15 2 25 3Offered Application Load (Task Request Rate)

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Syste

m R

esou

rce U

tiliz

atio

n

Figure 3 Resource utilization

self-controlled management policies It is a highly desirableproperty for the SFIoT system operations

Figure 3 illustrates the resource utilization of F-APs Wemeasure this performance metric to show and determinewhether our game-based approach can provide the systemefficiency As it was explained control decisions in theproposed scheme are made in an interactive two-phasegame model More specifically the proposed approach inthis paper can adapt the current SFIoT system conditionwhile dynamically adjusting the control decisions Whenthe application load is low (below 05) the performance ofeach scheme is not much difference As the application load

rate increases the resource utilization also increases All theschemes have similar trends however due to our approachrsquosflexibility adaptability and responsiveness to current SFIoTsystem conditions the F-AP resource in our proposed schemecan be used more efficiently under widely different anddiversified application load situations

Through simulation experiments the obtained resultsin Figures 1ndash3 clearly demonstrate that the suitability andfeasibility of our proposed scheme while revealing enhancedperspectives regarding QoE throughput and resource uti-lization in the SFIoT system In particular the BVG and NBSprocesses work together to reach an agreement that givesmutual advantages for mobile devices and F-APs two gamemodels are sophisticatedly combined and dependent on eachotherTherefore they act cooperatively to satisfy the differentperformance requirements

4 Summary and Conclusions

The past decade has witnessed an explosive growth of IoTwhere a variety of smart mobile devices offer a plethora ofdifferent applications Under the vision of future networksFC and social network principles are integrated into the IoTparadigm in order to enforce the distributed self-cooperationand self-management of mobile devices In this paper wepresent a novel resource allocation scheme for the SFIoTsystem According to the BVG and NBS approaches ourproposed scheme is designed as two phases (i) mobiledevices take votes to decide the resource allocation methodsbased on the weighted ternary BVG model and (ii) the par-titioned computation capacity is fair-efficiently distributed toeach application by using the NBS These two game modelscontinue iteratively in a step-by-step manner in accordancewith the current SFIoT system informationThe experimentalsimulation results prove that our proposed protocol canefficiently improve theQoE system throughput and resourceutilization compared with the existing SDFC SSEC andCFIC schemes As directions for future research we aim atinvestigating the privacy and energy issues for the SFIoTsystemWe especially plan to add a blockchain-based securitylayer to the system In addition we also plan to develop anoncooperative gamemodelwith theoretical analysis andwillfocus on the addition of more features to increase the SFIoTsystem reliability in an unconstrained environment Lastbut not least machine learning techniques can be appliedto complement our proposed scheme It will be a potentialdirection and another possible extension to this work

Data Availability

The data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

The author declares that there are no conflicts of interestregarding the publication of this paper

Wireless Communications and Mobile Computing 11

Authorsrsquo Contributions

The sole author contributes to all research work

Acknowledgments

This research was supported by the MSIT (Ministry ofScience and ICT) Korea under the ITRC (InformationTechnology Research Center) support program (IITP-2018-2018-0-01799) supervised by the IITP (Institute for Infor-mation amp communications Technology Promotion) and byInstitute for Information amp communications TechnologyPromotion (IITP) grant funded by the Korea government(MSIT) (No2017-0-00498 A Development of Deidentifica-tion Technique based on Differential privacy)

References

[1] A M Ortiz D Hussein S Park S N Han and N Crespi ldquoThecluster between internet of things and social networks reviewand research challengesrdquo IEEE Internet of ings Journal vol 1no 3 pp 206ndash215 2014

[2] Z Li R Chen L Liu andGMin ldquoDynamic resource discoverybased on preference andmovement pattern similarity for large-scale social internet of thingsrdquo IEEE Internet of ings Journalvol 3 no 4 pp 581ndash589 2016

[3] H Yang and L Zhang ldquoResponse time prediction of IoT servicebased on time similarityrdquo Journal of Computing Science andEngineering vol 11 no 3 pp 100ndash108 2017

[4] E Baccarelli M Scarpiniti P G Naranjo and L Vaca-Cardenas ldquoFog of social IoT when the fog becomes socialrdquoIEEE Network vol 32 no 4 pp 68ndash80 2018

[5] H Shah-Mansouri and V W S Wong ldquoHierarchical fog-cloudcomputing for iot systems a computation offloading gamerdquoIEEE Internet ofings Journal vol 5 no 4 pp 3246ndash3257 2018

[6] K Lee and I Shin ldquoUser mobility model based computationoffloading decision for mobile cloudrdquo Journal of ComputingScience and Engineering vol 9 no 3 pp 155ndash162 2015

[7] Y Liu M J Lee and Y Zheng ldquoAdaptive multi-resourceallocation for cloudlet-based mobile cloud computing systemrdquoIEEE Transactions on Mobile Computing vol 15 no 10 pp2398ndash2410 2016

[8] K Kim S Uno and M Kim ldquoAdaptive QoS mechanism forwireless mobile networkrdquo Journal of Computing Science andEngineering vol 4 no 2 pp 153ndash172 2010

[9] A Pande V Ramamurthi and P Mohapatra ldquoQuality-orientedvideo delivery over LTErdquo Journal of Computing Science andEngineering vol 7 no 3 pp 168ndash176 2013

[10] S Kim and P K Varshney ldquoAn integrated adaptive bandwidth-management framework for QoS-sensitive multimedia cellularnetworksrdquo IEEE Transactions on Vehicular Technology vol 53no 3 pp 835ndash846 2004

[11] K SungwookGameeory Applications in Network Design IGIGlobal Hershey PA USA 2014

[12] J M Bilbao J R Fernandez N Jimenez and J J LopezldquoBiprobabilistic values for bicooperative gamesrdquo DiscreteApplied Mathematics vol 156 pp 2698ndash2711 2008

[13] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheshapley value for bicooperative gamesrdquo Annals of OperationsResearch vol 158 pp 99ndash115 2008

[14] L Liu Z Chang and X Guo ldquoSocially-aware dynamic compu-tation offloading scheme for fog computing system with energyharvesting devicesrdquo IEEE Internet of ings Journal vol 5 no3 pp 1869ndash1879 2018

[15] R Dautov S Distefano D Bruneo F Longo G Merlino andA Puliafito ldquoData processing in cyber-physical-social systemsthrough edge computingrdquo IEEE Access vol 6 pp 29822ndash298352018

[16] Y-C Chen Y-C Chang C-H Chen Y-S Lin J-L Chen andY-Y Chang ldquoCloud-fog computing for information-centricinternet-of-things applicationsrdquo in Proceedings of the 2017 IEEEInternational Conference on Applied System Innovation ICASI2017 pp 637ndash640 May 2017

[17] C Jiang L Kuang Z Han Y Ren and L Hanzo ldquoInformationcredibility modeling in cooperative networks equilibrium andmechanism designrdquo IEEE Journal on Selected Areas in Commu-nications vol 35 no 2 pp 432ndash448 2017

[18] S Kim ldquoFog radio access network system control scheme basedon the embedded game modelrdquo EURASIP Journal on WirelessCommunications and Networking vol 2017 no 113 pp 1ndash142017

[19] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheBanzhaf power index for ternary bicooperative gamesrdquoDiscreteApplied Mathematics vol 158 pp 967ndash980 2010

[20] M TsurumiM Inuiguchi and A Nishimura ldquoPseudo-banzhafvalues in bicooperative gamesrdquo in Proceedings of the IEEEInternational Conference on Systems Man and Cybernetics pp1138ndash1143 2005

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Active and Passive Electronic Components

VLSI Design

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Shock and Vibration

Hindawiwwwhindawicom Volume 2018

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

SensorsJournal of

Hindawiwwwhindawicom Volume 2018

International Journal of

RotatingMachinery

Hindawiwwwhindawicom Volume 2018

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Navigation and Observation

International Journal of

Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom

4 Wireless Communications and Mobile Computing

(i) nonnegative property F(119909) ge 0(ii) nondecreasing property 119889F(119909)119889119909 ge 0(iii) saturation property 1198892F(119909)1198891199092 lt 0

The nonnegative property means that the outcome shouldbe larger than zero and if there is no social relationshipit is zero The nondecreasing property simply implies thatthe more users are socially connected the more outcomecan be obtained The saturation property suggests that whenthe social relationship is sufficient any further increase ofthe outcome remains marginal [17] In order to construct afunction while satisfying above three properties the socialpower function of119898119894 ieF119898119894(119909) is defined as follows

F119898119894 (119909) = Γ minus 1exp (120583 times 119909)

st 119909 = sum119898119895isinS119898119894

119890119898119894119898119895 (1)

where 120583 controls the speed of social power saturation and Γrepresents the maximum outcome of a user to normalize thesocial power [17]

In this study we assume that there are 119896 cloudlets and 119899mobile devices where 119896 ≪ 119899 In the SFIoT system infrastruc-ture cloudlets are evolved to the Fog-computing basedAccessPoint (F-AP) by being equipped with a certain computationpower and communication capability The design of F-APplatform has been introduced to deliver various devicesrequests which have different characteristics and resourcedemands that is CPU capacity for computation and wirelessbandwidth assignment for communication However dueto the limited CPU capacity and bandwidth scarcity in theF-AP it is impossible to guarantee all applicationsrsquo needs[18] Therefore an efficient resource management strategybecomes a key factor in enhancing the SFIoT performancewhile ensuring the required QoE

To tackle the SFIoT resource control problem we pro-pose a two-phase game based resource allocation schemeAt the first phase individual mobile devices participatein two BVGs one is to decide the allocation method ofcommunication resource and the other is to partition thecomputation resource according to the voting power of eachmobile device In each F-AP covering area the voting power iscalculated based on the pseudo-Banzhaf value At the secondphase the partitioned computation capacity is distributedfor same kinds of application tasks based on the NBS Topractically implement our resource allocation algorithmseach individual F-AP and its corresponding mobile devicesoperate their own two-phase game in an entirely distributedmanner For example the 119896th F-AP ie F119860119875119896 and the set ofcorresponding mobile devices in F119860119875119896 rsquos covering area ieNF119860119875119896 work together for their two-phase gameG119896 Therefore

multiple two-phase games G1le119894le119896 are operated in parallel Toformally define our proposed game model we characterizeG119896 as F119860119875119896 NF119860119875

119896 (S119898119894 F119898119894(119909) ΨQ

119898119894) | 119898119894 isin NF119860119875

119896 119909 =sum119898119895isinS119898119894 119890119898119894119898119895AF119860119875

119896 120572F119860119875119896 119880F119860119875

119896 Table 1 lists the notations

used in this paper

(i) F119860119875119896 NF119860119875119896 is the set of gameplayers for theG119896 game

where NF119860119875119896cap NF119860119875

119897= 0

(ii) (S119898119894 F119898119894(119909) ΨQ119898119894) is a three-pair for the 119898119894 isin NF119860119875

119896

where S119898119894 is the set of119898119894rsquos social friendsF119898119894(119909) andΨQ119898119894

represent the social power and the voting powerof119898119894 respectively

(iii) AF119860119875119896

is the set of generated application tasks whichare offloaded in theF119860119875119896 for FC services

(iv) 120572F119860119875119896

is the ratio of relative service weights givento different type applications in the F119860119875119896 where0 le 120572F119860119875

119896le 1 Based on the 120572F119860119875

119896value F119860119875119896 rsquos

computation power is partitioned

(v) 119880F119860119875119896

is the utility function of F119860119875119896 it is optimizedbased on the NBS

22 Bicooperative Game and Ternary Voting Model Usuallycooperative game theory studies situations where a group ofgame players are associated to obtain a profit as a result oftheir cooperation Thus a cooperative gamemodel is definedas a pair (119873V) where N is a finite set of players and V2119873 997888rarr R is a function verifying that V(0) = 0 For eachcoalition 119878 isin 2119873V(119878) can be interpreted as the coalition Srsquosmaximal gain that players in S can achieve themselves againstthe best offensive threat by the complementary coalition119873119878[12]

In 2008 Bilbao et al proposed a new idea called bicoop-erative game for themulticriteria decision-making process Inthe bicooperative game game players are allowed to chooseone among three alternatives To think the voting situationby analogy we consider ordered pairs (S T) with 119878 119879 sube119873 and 119878 cap 119879 = 0 Each pair (S T) yields a partition of119873 in three groups Players in S are in favor of the votingitem and players in T object to it The remaining players in119873 (119878 cup 119879) are nonvoters Formally we can define the set3119873 = (119878 119879) | 119878 119879 sube 119873 119878 cap 119879 = 0 and the functionQ 3119873 997888rarr R For each (119878 119879) isin 3119873 the outcome of Q(119878 119879)can be interpreted as the gain that 119878 can achieve when 119879 isthe opposite coalition and119873 (119878 cup 119879) is the neutral coalition[12 13]

The pair (0119873) represents the situation if all the playersobject to the voting item and (119873 0) represents the situationwhere all the players support the voting item More generallya relation of 3119873 is given by [12 13]

(119860 119861) ⊑ (119862119863) lArrrArr 119860 sube 119862 119861 supe 119863 (2)

The set (3119873 ⊑) is a finite distributive lattice and a partiallyordered set with the following properties [12 13]

(i) (0119873) is the first element (0119873) ⊑ (119860 119861) for all(119860 119861) isin 3119873(ii) (119873 0) is the last element (119860 119861) ⊑ (119873 0) for all(119860 119861) isin 3119873

Wireless Communications and Mobile Computing 5

Table 1 Symbols and parameters used in the proposed algorithm

Notations ExplanationN the set of mobile devicesE the set of the social ties119890119898119894 119898119895 a real-valued number to measure the strength of the social tie between119898119894 and119898119895S119898119894 the set of the social friendsof119898119894F119898119894 (119909) the social power function of 119898119894120583 a parameter to control the speed of social power saturationΓ the maximum outcome of a user to normalize the social powerF119860119875119896 the 119896th fog-computing based access pointNF119860119875119896

the set of corresponding mobile devices in theF119860119875119896 rsquos covering areaG119896 The two-phase game inF119860119875119896ΨQ119898119894

the social power and the voting power of 119898119894AF119860119875119896

the set of application tasks which are offloaded in theF119860119875119896 for FC services120572F119860119875119896

the ratio of relative service weights given to different type applications in theF119860119875119896119880F119860119875119896

the utility function ofF119860119875119896119873 a finite set of voting game playersV (119878) the coalition Srsquos maximal gainQ (119878 119879) the gain that 119878 can achieve when 119879 is the opposite coalitionBG119873 the set of all bicooperative games with players119873120601119894 (Q) the player 119894rsquos value in a game Q onBG119873

Q(119878119879) (119860 119861) the identity gameQ(119878119879) (119860 119861) the superior unanimity gameQ(119878119879) (119860 119861) the inferior unanimity gameQ119894 the payoff of voter 119894W (119878) the sum of voting weights of coalition 119878rsquos playersP the quota for the decision-taking120599 the quota for the decision-blockingX a weighted ternary voting gameB the total available bandwidth amountm119894 the pre-defined minimum bandwidth amount for the 119894119905ℎ service119877119894 the bandwidth request of mobile device119898119894119860 119894 the decided bandwidth amount for the119898119894119879119862L119862 the constraint thresholds which are chosen to satisfy each condition120572119868 the preference ratio of class I application servicesC each F-APrsquos total computation capacity120578119868120578119868119868 control parameters to adjust the payoff of119898CF119860119875119896

the total computation capacity ofF119860119875119896

(iii) Every fair of elements of 3119873 has a join(119860 119861)⋁ (119862119863) = (119860 cup 119862 119861 cap 119863)

and (119860 119861) and (119862119863) = (119860 cap 119862 119861 cup 119863) (3)

The set of all bicooperative games with players119873 is denotedbyBG119873 which is a real vector space [12 13]

BG119873 = Q | 3119873 997888rarr R and Q (0 0) = 0 (4)

A value on BG119873 is a mapping 120601 BG119873 997888rarr R119899 thatassociates with each game Q isin BG119873 In a vector (1206011(Q) 120601119899(Q)) isin R119899 1206011le119894le119899(Q) represents a real number whichis the player 119894rsquos value in a game Q on BG119873 Therefore themapping 120601119894isin119873(Q) BG119873 997888rarr R represents the player 119894rsquospayoff from playing bicooperative games [12 13] There arethree special collections of games in BG119873 taking values inminus1 0 1 the identity games the superior unanimity gamesand the inferior unanimity games which are defined for any(119878 119879) isin 3119873 [12 13]

The identity game Q(119878119879)(119860 119861) | 3119873 997888rarr R is defined by

Q(119878119879) (119860 119861)

6 Wireless Communications and Mobile Computing

= 1 if (119860 119861) = (119878 119879) and (119878 119879) = (0 0)0 otherwise

(5)

The superior unanimity game Q(119878119879)(119860 119861) | 3119873 997888rarr R isdefined by

Q(119878119879) (119860 119861)=

1 if (119878 119879) ⊑ (119860 119861) and (119860 119861) = (0 0)0 otherwise

(6)

The inferior unanimity game Q(119878119879)(119860 119861) | 3119873 997888rarr R isdefined by

Q(119878119879) (119860 119861)=

minus1 if (119860 119861) ⊑ (119878 119879) and (119860 119861) = (0 0)0 otherwise

(7)

Voting game is a cooperative game model that incorporateselements from social choice theory This game model can beused to analyze situations where voters would vote in favor ofor against a decision [11] Simple voting game with abstentioncan be formulated as a bicooperative game It has been studiedunder the name of ternary voting game [19 20]

Let N be a set of n voters that has to choose between apair of two alternatives 119878 and 119879 Every voter 119894 isin 119873 has apreference over the two alternatives 119878 and 119879 Thus the payoffof voter 119894 (Q119894) is defined by Q119894 3119873 997888rarr minus1 0 1 based onthe majority rule Formally a ternary voting game Q isin BG119873

satisfies the following conditions [19]

(I) For every bicoalition pair (119878 119879)isin 3119873 its worthQ(119878 119879) isin minus1 0 1

(II) If (119878 119879) (119878 ) isin 3119873 with (119878 119879) ⊑ (119878 ) thenQ(119878 119879) le Q(119878 )

(III)Q119894

=

1 if (119894 isin 119878 and |119878| gt |119879|) or (119894 isin 119879 and |119879| gt |119878|)minus1 if (119894 isin 119878 and |119878| lt |119879|) or (119894 isin 119879 and |119879| lt |119878|)0 if voter i abstains form voting or |119878| = |119879|

(8)

where |119878| is the cardinality of coalition 119878 For voting gamesthe analysis of voting power is the main issue In 1965Banzhaf value was introduced by John F Banzhaf III based onprobabilistic analysis of the individual voters in a block votingsystem It depends on the number of ways in which eachvoter can effect a swing in the game In 2010 J Bilbao et alintroduced a pseudo-Banzhaf value ie ΨQ

119894 for the ternaryvoting game Q it can be interpreted as a voterrsquos probabilisticpower index [19 20]

ΨQ119894 = 13119899minus1 times sum

(119878119879)isin3119873119894

(Q119894 (119878 cup 119894 119879) minus Q119894 (119878 119879 cup 119894)) (9)

The pseudo-Banzhaf value satisfies four axioms Null PlayerTotal Swings Transfer Property and Simple Additivity[19 20]

Axiom 1 (null player) If 119894 isin 119873 is a null player in Q isin BG119873then ΨQ

119894 = 0 Therefore if the player 119894 is the null player in Q

with any (119878 119879) isin 3119873119894 thenQ119894 (119894 0) = (Q119894 (119878 cup 119894 119879) minus Q119894 (119878 119879))Q119894 (0 119894) = (Q119894 (119878 119879) minus Q119894 (119878 119879 cup 119894))

120601119894 (Q) = (Q(119878119879) (119894 0) minus Q(119878119879) (0 119894))(10)

Axiom 2 (total swings) If Q isin BG119873 then

sum119894isin119873

ΨQ119894 = sum(119878119879)isin3119873119894

(Q119894 (119878 cup 119894 119879) minus Q119894 (119878 119879 cup 119894)) (11)

Axiom 3 (transfer property) For any QG isin BG119873 we havethat

(120601119894 (Q) + 120601119894 (G)) = (120601119894 (Q⋁G) + 120601119894 (Q andG)) (12)

Axiom 4 (simple additivity) For any Q isin BG119873 such thatQ = G +H we have that

120601119894 (Q) = (120601119894 (G) + 120601119894 (H)) (13)

As a special ternary voting gamemodel the weighted ternaryvoting game can formulate a decision-making process thatnot all voters have the same amount of influence overthe decision Therefore votes of different voters are givendifferent weight This type of voting game model is usedin shareholder meetings where votes are weighted by thenumber of shares that each shareholder owns [11] In thisstudy we develop a ternary voting game model based onthe weighted voting approach In our game model mobiledevices are treated differently according to their socialpowers (F119898(119909)) and give different amounts of influenceto different members Therefore each coalition of players119878 sube 119873 has the sum of voting weights of its compo-nents that is W(119878) = sum119898119894isin119878F119898119894(119909) If the value ofW(119878) exceeds a certain quota players in 119878 win the game[11 19]

For the weighted ternary voting game we need twoquotas one is to block a decision and the other is toapprove it Formally it is represented by the voting body[P 120599F1198981(119909) F119898119899(119909)] where P is the quota for thedecision-taking and 120599 is the quota for the decision-blockingit can be seen as a particular case of n-player games with

Wireless Communications and Mobile Computing 7

three choices This situation can be formulated by a weightedternary voting game (X) ieX(119878 119879) 3119873 997888rarr minus1 0 1 [12]

X (119878 119879) =

1 if W (119878) ge P and W (119879) lt 120599minus1 if W (119878) lt P and W (119879) ge 1205990 otherwise

(14)

st P = 120572 times ( 119899sum119894=1

F119898119894 (119909))

and 120599 = 120573 times ( 119899sum119894=1

F119898119894 (119909))(15)

23 Communication and Computation Resource AllocationMechanisms Wireless bandwidth is a valuable and scarceresource in the SFIoT system Therefore the limited wirelessbandwidth has to be shared fair-efficiently by mobile usersIf the allocation method is not considered explicitly at thedesign stage of bandwidth allocation algorithms differentbandwidth requests can result in very unfair bandwidthallocations it may lead to a system inefficiency Usuallyvarious bandwidth allocation methods can be categorizedinto three ways proportional allocation (PA) constrainedallocation (CA) and equal loss allocation (EA) ways

(I) Proportional allocation (PA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max(119877119894 times Bsum119899119894=1 119877119894) m119894 otherwise

(16)

(II) Constrained allocation (CA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max (min 119877119894 119879119862 m119894) otherwise

st119899sum119894=1

min 119877119894 119879119862 = B

(17)

(III) Equal loss allocation (EA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max (119877119894 minusL119862) m119894 otherwise

st119899sum119894=1

max (119877119894 minusL119862) 0 = B

(18)

where B is the total available bandwidth amount and m119894is the predefined minimum bandwidth amount for the 119894119905ℎ

service 1198771le119894le119873 is the bandwidth request of mobile device119898119894 and 1198601le119894le119873 represents the decided bandwidth amountfor 119898119894 119879119862 and L119862 are the constraint thresholds which arechosen to satisfy each condition If the available bandwidthis enough or the F-AP can secure the new request 119860 119894from 119898119894 through the allocated bandwidth readjustment thenew requested application is accepted Otherwise the newrequested application is rejected

In our bandwidth allocation algorithm each F-APdecides its resource allocation way based on the weightedternary voting game InF119860119875119896 consider theX gamewithNF119860119875

119896

Individual 119898119894 isin NF119860119875119896

participates in the voting procedurewith his voting weight Usually the PA way is the mostcommon method and it can be seen as implicit consentTherefore for each voter we can assume that there are threeoptions ie pro-CA pro-EA and abstention Accordingto (14) the coalition 119878 consists of pro-CA voters and thecoalition 119879 consists of pro-EA voters If the outcome ofX(119878 119879) is 1 or -1 F119860119875119896 simply adopts the CA way or the EAway respectively in the bandwidth allocation process If theoutcome ofX(119878 119879) is 0F119860119875119896 adopts the PAway Based on thedecisions of individual devices Each F-AP adaptively decidesits communication resource allocation way

Like as the communication resource computation capac-ity in each F-AP is also a naturally limited resourceThereforethe computation power of F-AP is adaptively assigned to eachrequested application task Usually different applications canbe categorized into two classes class I (real-time data applica-tions) and class II (non-real-time data applications) accord-ing to the required QoE They not only require differentcomputation capacity but also have different QoE In orderto design the computation resource allocation algorithmeach F-AP needs to partition its own computation powerto support different type applications In this paper we alsoadopt the weighted ternary voting game model However incontrast to the voting game in our communication allocationalgorithm the developed computation allocation algorithmhas focused on the basic concept of pseudo-Banzhaf values

Consider another X game inF119860119875119896 with NF119860119875119896 Individual119898119894 isin NF119860119875

119896participates in the second voting procedure At

this time there are three options for each voter ie proclass Iservice proclass II service and abstention According to (14)the coalition 119878 consists of pro-class I service voters and thecoalition 119879 consists of pro-class II service voters Based on theoutcome of X(119878 119879) the F-AP estimates the preference ratio(120572119868) of class I application services as follows

120572119868= sum119898119894isin119878 (ΨQ

119898119894timesF119898119894 (119909))

(sum119898119894isin119878 (ΨQ119898119894timesF119898119894 (119909)) + sum119898119895isin119879 (ΨQ

119898119895timesF119898119894 (119909)))

(19)

According to each userrsquos pseudo-Banzhaf value and socialpower the 120572119868 value is decided and each F-AP partitions itstotal computation capacity (C) The division ratio for class I(or class II) type application services is 120572119868 (or 1 minus 120572119868) Andthen the partitioned computation capacity is distributed forits corresponding type application services while maximizing

8 Wireless Communications and Mobile Computing

Table 2 Application service type and requirements and system parameters

Type Application Bandwidthrequirement

Minimumrequirement (m)

Computationrequirement Duration (Avesec)

Class I Application I 119860=512 Mbps 119860=256 Mbps 98 MHz 180 secApplication II 119860=640 Mbps 119860=320 Mbps 256MHz 120 sec

Class II

Application III 119860=768 Mbps 119860=384 Mbps 128 MHz 240 secApplication IV 119860=896 Mbps 119860=458 Mbps 256MHz 300 secApplication V 119860=128 Gbps 119860=640 Mbps 384 MHz 360 secApplication VI 119860=152 Gbps 119860=860 Mbps 512 MHz 480 sec

Parameter Value DescriptionΓ 5 the maximum outcome of a user to normalize the social power120583 09 a parameter to controls the speed of social power saturation120572 075 a factor to decide the quota for the decision-taking120573 025 a factor to decide the quota for the decision- blocking120578119868 09 a control parameter to adjust the payoff of class I service120578119868119868 07 a control parameter to adjust the payoff of class II service120576 10 a control parameter for the sharing activity between social friends

the systemperformance To implement this process we adoptthe fundamental notion of the Nash Bargaining Solution(NBS) This is captured by the following optimization prob-lem

max[119860119898119894 ]

prod119898119894isin119878

119880119868119898119894 (119860119898119894) + max[119860119898119895 ]

prod119898119895isin119879

119880119868119868119898119895 (119860119898119895) (20)

st 119880119868119898119894 (119860119898119894) = 120578119868 times log2 (1 + 119860119898119894119877119898119894 )

119880119868119868119898119895 (119860119898119895) = 120578119868119868 times (1 minus exp(minus119860119898119894119877119898119894 ))sum119898119894isin119878

119860119898119894 le (120572119868 timesCF119860119875119896)

and sum119898119895isin119879

119860119898119895 le ((1 minus 120572119868) timesCF119860119875119896)

(21)

where 120578119868 (or 120578119868119868) is a control parameter to adjust the payoffof 119898119894 (or 119898119895) and CF119860119875

119896is the total computation capacity of

F119860119875119896

24 Main Steps of Proposed SFIoT Resource AllocationScheme Under the vision of the future internet the SIoTparadigm aims to connect anything while relying on the self-establishment of inter-thing social relationships In additionthe emerging paradigm of FC exhibits the right features forcoping with the aforementioned technological 5G networkissues In order to address this challenge the SFIoT paradigmhas been gaining momentum To effectively operate theSFIoT system the interactive relationship between F-APs andmobile devices is an important research topic and should beconsidered to design the resource allocation algorithms Inthis study we provide the communication and computationresource allocation scheme which is formulated according

to the BVG and NBS Owing to our two-phase game modelwe can get the most fair-efficient system performance bycombining both game approaches The main steps of theproposed scheme are described as follows

Step 1 Application types requirements and system controlparameters are determined by the simulation scenario (seesimulation assumptions in Section 3 and Table 2)

Step 2 Each individual F-AP monitors its covering area andresponds to its corresponding mobile devices Mobile devicesgenerate their application service requests in a distributedfashion and ask FC services to their corresponding F-AP

Step 3 Each mobile device has its own social relationshipthrough a social network close friends can probabilisticallyshare the outcome of FC services

Step 4 To reduce computation complexity individual F-APsexecute separately their two-phase games in parallel

Step 5 During the first game process each mobile devicetakes two votes one is to decide the communication resourceallocation way and the other is to partition the F-APrsquoscomputation capacity

Step 6 According to (14) the F-AP decides its communi-cation allocation way Based on our weighted ternary BVGmodel one of the PA CA and EA ways is adaptivelyselected And then the F-AP estimates the division ratio ofcomputation capacity using (19)

Step 7 During the second game process each F-AP inde-pendently distributes the partitioned computation capacity tocorresponding type applications According to (20) differentapplications can retain their FC services while maximizingthe system performance

Wireless Communications and Mobile Computing 9

Step 8 Based on the two-phase game model F-APs andmobile devices are hierarchically interconnected and inter-acting with one another to operate the SFIoT system

Step 9 Constantly each F-AP is self-monitoring its coveringarea When the game players are changed due to the mobiledevice migration the proposed scheme proceeds to Step 5 forthe next two-phase game procedure

3 Simulation Results and Discussion

In this section we evaluate the effectiveness of our proposedprotocol and analyze the simulation result while compar-ing it with that of the SDFC [14] SSEC [15] and CFIC[16] schemes To ensure a fair comparison the experi-mental setup consists of the following specifications andscenario

(i) Simulated SFIoT system consists of 25 F-APs and 100mobile devices F-APs are laid out in regular patternandmobile is randomly located in the F-APsrsquo coveringareas

(ii) For the FC services total computation capacity ofeach F-AP is 30 GHz and total communicationresource of each F-AP is 50 Gbps

(iii) Each mobile device individually generates servicerequests Service type of applications is randomlydecided and service request rate is Poisson process(120588) The offered rate range is varied from 0 to 30

(iv) The social friend number (|S119898| gt 0) of each mobiledevice is a random variable which has the maximumprobability at 3 while following a normal distribution

(v) Each social tie (0 le 119890119898119898 le 1) between mobiledevices is also a random variable which has themaximum probability at 05 while following a normaldistribution

(vi) The probability (P119904) that social friends119898119894 and119898119895 canshare the service outcome is defined asP119904 = 119890119898119894119898119895120576where 120576 is a control parameter for the sharing activity

(vii) We assume that there are no physical obstacles in theexperiments and each mobile device can contact itscorresponding F-AP for FC services

(viii) For calculation simplicity the communicationresource is specified in terms of basic units (BUs)where one BU is the minimum amount (eg 128Mbps in our system) of resource management

(ix) Using the iterative water-filling method the value ofNBS in (20) is obtained

(x) Network performancemeasures obtained on the basisof 100 simulation runs are plotted as functions of theoffered service request rate (120588)

(xi) Six different applications are assumed based on thedifferent computation and communication require-ment they are generated with equal probability

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

05 1 15 2 25 30Offered Application Load (Task Request Rate)

0

05

1

15

Qua

lity

of E

xper

ienc

e (Q

oE)

Figure 1 SFIoT quality of experience (QoE)

(xii) Performance criteria obtained through simulation areQoE resource usability and task failure probabilitythese simulation metrics are evaluated mainly todemonstrate the validity of our proposed method

To facilitate the development and implementation of oursimulator Table 2 lists application service type requirementsand system control parameters

Figure 1 shows the QoE of each scheme in differentapplication load scenarios For this evaluationQoE is definedas the normalized resource loss ratio than the originallyrequested amount From the viewpoint of mobile users QoEis a key factor to evaluate their service satisfactions In theproposed scheme the system resource is differently allocatedbased on the voting approach According to the desirablefeatures of ternary voting approach the limited resourcecan be assigned while effectively reflecting the preferenceinclination of users Our two-phase game approach can wellorchestrate the SFIoT system infrastructure to maximize theQoE of mobile users The result of Figure 1 proves that ourtwo-step game method is a socially beneficial way to achievea better QoE than other existing schemes

The obtained results of task failure probabilities aredepicted in Figure 2 This performance criterion is stronglyrelated to the SFIoT system throughput which is estimatedbased on the ratio of successfully completed applications Asexpected the application load increases the available systemresource is exhausted to support new application servicesTherefore task failure probabilities increase proportionatelywith the application request rate The resulting curves inFigure 2 allow us to see that our proposed scheme has gaineda lower task failure probability than other existing schemes Itis therefore worth to say that under different application loadconditions our proposed scheme can perform excellentlyto maintain a superior system throughput based on the

10 Wireless Communications and Mobile Computing

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Task

Fai

lure

Pro

babi

lity

05 1 15 2 25 30Offered Application Load (Task Request Rate)

Figure 2 SFIoT task failure probability

0 05 1 15 2 25 3Offered Application Load (Task Request Rate)

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Syste

m R

esou

rce U

tiliz

atio

n

Figure 3 Resource utilization

self-controlled management policies It is a highly desirableproperty for the SFIoT system operations

Figure 3 illustrates the resource utilization of F-APs Wemeasure this performance metric to show and determinewhether our game-based approach can provide the systemefficiency As it was explained control decisions in theproposed scheme are made in an interactive two-phasegame model More specifically the proposed approach inthis paper can adapt the current SFIoT system conditionwhile dynamically adjusting the control decisions Whenthe application load is low (below 05) the performance ofeach scheme is not much difference As the application load

rate increases the resource utilization also increases All theschemes have similar trends however due to our approachrsquosflexibility adaptability and responsiveness to current SFIoTsystem conditions the F-AP resource in our proposed schemecan be used more efficiently under widely different anddiversified application load situations

Through simulation experiments the obtained resultsin Figures 1ndash3 clearly demonstrate that the suitability andfeasibility of our proposed scheme while revealing enhancedperspectives regarding QoE throughput and resource uti-lization in the SFIoT system In particular the BVG and NBSprocesses work together to reach an agreement that givesmutual advantages for mobile devices and F-APs two gamemodels are sophisticatedly combined and dependent on eachotherTherefore they act cooperatively to satisfy the differentperformance requirements

4 Summary and Conclusions

The past decade has witnessed an explosive growth of IoTwhere a variety of smart mobile devices offer a plethora ofdifferent applications Under the vision of future networksFC and social network principles are integrated into the IoTparadigm in order to enforce the distributed self-cooperationand self-management of mobile devices In this paper wepresent a novel resource allocation scheme for the SFIoTsystem According to the BVG and NBS approaches ourproposed scheme is designed as two phases (i) mobiledevices take votes to decide the resource allocation methodsbased on the weighted ternary BVG model and (ii) the par-titioned computation capacity is fair-efficiently distributed toeach application by using the NBS These two game modelscontinue iteratively in a step-by-step manner in accordancewith the current SFIoT system informationThe experimentalsimulation results prove that our proposed protocol canefficiently improve theQoE system throughput and resourceutilization compared with the existing SDFC SSEC andCFIC schemes As directions for future research we aim atinvestigating the privacy and energy issues for the SFIoTsystemWe especially plan to add a blockchain-based securitylayer to the system In addition we also plan to develop anoncooperative gamemodelwith theoretical analysis andwillfocus on the addition of more features to increase the SFIoTsystem reliability in an unconstrained environment Lastbut not least machine learning techniques can be appliedto complement our proposed scheme It will be a potentialdirection and another possible extension to this work

Data Availability

The data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

The author declares that there are no conflicts of interestregarding the publication of this paper

Wireless Communications and Mobile Computing 11

Authorsrsquo Contributions

The sole author contributes to all research work

Acknowledgments

This research was supported by the MSIT (Ministry ofScience and ICT) Korea under the ITRC (InformationTechnology Research Center) support program (IITP-2018-2018-0-01799) supervised by the IITP (Institute for Infor-mation amp communications Technology Promotion) and byInstitute for Information amp communications TechnologyPromotion (IITP) grant funded by the Korea government(MSIT) (No2017-0-00498 A Development of Deidentifica-tion Technique based on Differential privacy)

References

[1] A M Ortiz D Hussein S Park S N Han and N Crespi ldquoThecluster between internet of things and social networks reviewand research challengesrdquo IEEE Internet of ings Journal vol 1no 3 pp 206ndash215 2014

[2] Z Li R Chen L Liu andGMin ldquoDynamic resource discoverybased on preference andmovement pattern similarity for large-scale social internet of thingsrdquo IEEE Internet of ings Journalvol 3 no 4 pp 581ndash589 2016

[3] H Yang and L Zhang ldquoResponse time prediction of IoT servicebased on time similarityrdquo Journal of Computing Science andEngineering vol 11 no 3 pp 100ndash108 2017

[4] E Baccarelli M Scarpiniti P G Naranjo and L Vaca-Cardenas ldquoFog of social IoT when the fog becomes socialrdquoIEEE Network vol 32 no 4 pp 68ndash80 2018

[5] H Shah-Mansouri and V W S Wong ldquoHierarchical fog-cloudcomputing for iot systems a computation offloading gamerdquoIEEE Internet ofings Journal vol 5 no 4 pp 3246ndash3257 2018

[6] K Lee and I Shin ldquoUser mobility model based computationoffloading decision for mobile cloudrdquo Journal of ComputingScience and Engineering vol 9 no 3 pp 155ndash162 2015

[7] Y Liu M J Lee and Y Zheng ldquoAdaptive multi-resourceallocation for cloudlet-based mobile cloud computing systemrdquoIEEE Transactions on Mobile Computing vol 15 no 10 pp2398ndash2410 2016

[8] K Kim S Uno and M Kim ldquoAdaptive QoS mechanism forwireless mobile networkrdquo Journal of Computing Science andEngineering vol 4 no 2 pp 153ndash172 2010

[9] A Pande V Ramamurthi and P Mohapatra ldquoQuality-orientedvideo delivery over LTErdquo Journal of Computing Science andEngineering vol 7 no 3 pp 168ndash176 2013

[10] S Kim and P K Varshney ldquoAn integrated adaptive bandwidth-management framework for QoS-sensitive multimedia cellularnetworksrdquo IEEE Transactions on Vehicular Technology vol 53no 3 pp 835ndash846 2004

[11] K SungwookGameeory Applications in Network Design IGIGlobal Hershey PA USA 2014

[12] J M Bilbao J R Fernandez N Jimenez and J J LopezldquoBiprobabilistic values for bicooperative gamesrdquo DiscreteApplied Mathematics vol 156 pp 2698ndash2711 2008

[13] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheshapley value for bicooperative gamesrdquo Annals of OperationsResearch vol 158 pp 99ndash115 2008

[14] L Liu Z Chang and X Guo ldquoSocially-aware dynamic compu-tation offloading scheme for fog computing system with energyharvesting devicesrdquo IEEE Internet of ings Journal vol 5 no3 pp 1869ndash1879 2018

[15] R Dautov S Distefano D Bruneo F Longo G Merlino andA Puliafito ldquoData processing in cyber-physical-social systemsthrough edge computingrdquo IEEE Access vol 6 pp 29822ndash298352018

[16] Y-C Chen Y-C Chang C-H Chen Y-S Lin J-L Chen andY-Y Chang ldquoCloud-fog computing for information-centricinternet-of-things applicationsrdquo in Proceedings of the 2017 IEEEInternational Conference on Applied System Innovation ICASI2017 pp 637ndash640 May 2017

[17] C Jiang L Kuang Z Han Y Ren and L Hanzo ldquoInformationcredibility modeling in cooperative networks equilibrium andmechanism designrdquo IEEE Journal on Selected Areas in Commu-nications vol 35 no 2 pp 432ndash448 2017

[18] S Kim ldquoFog radio access network system control scheme basedon the embedded game modelrdquo EURASIP Journal on WirelessCommunications and Networking vol 2017 no 113 pp 1ndash142017

[19] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheBanzhaf power index for ternary bicooperative gamesrdquoDiscreteApplied Mathematics vol 158 pp 967ndash980 2010

[20] M TsurumiM Inuiguchi and A Nishimura ldquoPseudo-banzhafvalues in bicooperative gamesrdquo in Proceedings of the IEEEInternational Conference on Systems Man and Cybernetics pp1138ndash1143 2005

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Active and Passive Electronic Components

VLSI Design

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Shock and Vibration

Hindawiwwwhindawicom Volume 2018

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

SensorsJournal of

Hindawiwwwhindawicom Volume 2018

International Journal of

RotatingMachinery

Hindawiwwwhindawicom Volume 2018

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Navigation and Observation

International Journal of

Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom

Wireless Communications and Mobile Computing 5

Table 1 Symbols and parameters used in the proposed algorithm

Notations ExplanationN the set of mobile devicesE the set of the social ties119890119898119894 119898119895 a real-valued number to measure the strength of the social tie between119898119894 and119898119895S119898119894 the set of the social friendsof119898119894F119898119894 (119909) the social power function of 119898119894120583 a parameter to control the speed of social power saturationΓ the maximum outcome of a user to normalize the social powerF119860119875119896 the 119896th fog-computing based access pointNF119860119875119896

the set of corresponding mobile devices in theF119860119875119896 rsquos covering areaG119896 The two-phase game inF119860119875119896ΨQ119898119894

the social power and the voting power of 119898119894AF119860119875119896

the set of application tasks which are offloaded in theF119860119875119896 for FC services120572F119860119875119896

the ratio of relative service weights given to different type applications in theF119860119875119896119880F119860119875119896

the utility function ofF119860119875119896119873 a finite set of voting game playersV (119878) the coalition Srsquos maximal gainQ (119878 119879) the gain that 119878 can achieve when 119879 is the opposite coalitionBG119873 the set of all bicooperative games with players119873120601119894 (Q) the player 119894rsquos value in a game Q onBG119873

Q(119878119879) (119860 119861) the identity gameQ(119878119879) (119860 119861) the superior unanimity gameQ(119878119879) (119860 119861) the inferior unanimity gameQ119894 the payoff of voter 119894W (119878) the sum of voting weights of coalition 119878rsquos playersP the quota for the decision-taking120599 the quota for the decision-blockingX a weighted ternary voting gameB the total available bandwidth amountm119894 the pre-defined minimum bandwidth amount for the 119894119905ℎ service119877119894 the bandwidth request of mobile device119898119894119860 119894 the decided bandwidth amount for the119898119894119879119862L119862 the constraint thresholds which are chosen to satisfy each condition120572119868 the preference ratio of class I application servicesC each F-APrsquos total computation capacity120578119868120578119868119868 control parameters to adjust the payoff of119898CF119860119875119896

the total computation capacity ofF119860119875119896

(iii) Every fair of elements of 3119873 has a join(119860 119861)⋁ (119862119863) = (119860 cup 119862 119861 cap 119863)

and (119860 119861) and (119862119863) = (119860 cap 119862 119861 cup 119863) (3)

The set of all bicooperative games with players119873 is denotedbyBG119873 which is a real vector space [12 13]

BG119873 = Q | 3119873 997888rarr R and Q (0 0) = 0 (4)

A value on BG119873 is a mapping 120601 BG119873 997888rarr R119899 thatassociates with each game Q isin BG119873 In a vector (1206011(Q) 120601119899(Q)) isin R119899 1206011le119894le119899(Q) represents a real number whichis the player 119894rsquos value in a game Q on BG119873 Therefore themapping 120601119894isin119873(Q) BG119873 997888rarr R represents the player 119894rsquospayoff from playing bicooperative games [12 13] There arethree special collections of games in BG119873 taking values inminus1 0 1 the identity games the superior unanimity gamesand the inferior unanimity games which are defined for any(119878 119879) isin 3119873 [12 13]

The identity game Q(119878119879)(119860 119861) | 3119873 997888rarr R is defined by

Q(119878119879) (119860 119861)

6 Wireless Communications and Mobile Computing

= 1 if (119860 119861) = (119878 119879) and (119878 119879) = (0 0)0 otherwise

(5)

The superior unanimity game Q(119878119879)(119860 119861) | 3119873 997888rarr R isdefined by

Q(119878119879) (119860 119861)=

1 if (119878 119879) ⊑ (119860 119861) and (119860 119861) = (0 0)0 otherwise

(6)

The inferior unanimity game Q(119878119879)(119860 119861) | 3119873 997888rarr R isdefined by

Q(119878119879) (119860 119861)=

minus1 if (119860 119861) ⊑ (119878 119879) and (119860 119861) = (0 0)0 otherwise

(7)

Voting game is a cooperative game model that incorporateselements from social choice theory This game model can beused to analyze situations where voters would vote in favor ofor against a decision [11] Simple voting game with abstentioncan be formulated as a bicooperative game It has been studiedunder the name of ternary voting game [19 20]

Let N be a set of n voters that has to choose between apair of two alternatives 119878 and 119879 Every voter 119894 isin 119873 has apreference over the two alternatives 119878 and 119879 Thus the payoffof voter 119894 (Q119894) is defined by Q119894 3119873 997888rarr minus1 0 1 based onthe majority rule Formally a ternary voting game Q isin BG119873

satisfies the following conditions [19]

(I) For every bicoalition pair (119878 119879)isin 3119873 its worthQ(119878 119879) isin minus1 0 1

(II) If (119878 119879) (119878 ) isin 3119873 with (119878 119879) ⊑ (119878 ) thenQ(119878 119879) le Q(119878 )

(III)Q119894

=

1 if (119894 isin 119878 and |119878| gt |119879|) or (119894 isin 119879 and |119879| gt |119878|)minus1 if (119894 isin 119878 and |119878| lt |119879|) or (119894 isin 119879 and |119879| lt |119878|)0 if voter i abstains form voting or |119878| = |119879|

(8)

where |119878| is the cardinality of coalition 119878 For voting gamesthe analysis of voting power is the main issue In 1965Banzhaf value was introduced by John F Banzhaf III based onprobabilistic analysis of the individual voters in a block votingsystem It depends on the number of ways in which eachvoter can effect a swing in the game In 2010 J Bilbao et alintroduced a pseudo-Banzhaf value ie ΨQ

119894 for the ternaryvoting game Q it can be interpreted as a voterrsquos probabilisticpower index [19 20]

ΨQ119894 = 13119899minus1 times sum

(119878119879)isin3119873119894

(Q119894 (119878 cup 119894 119879) minus Q119894 (119878 119879 cup 119894)) (9)

The pseudo-Banzhaf value satisfies four axioms Null PlayerTotal Swings Transfer Property and Simple Additivity[19 20]

Axiom 1 (null player) If 119894 isin 119873 is a null player in Q isin BG119873then ΨQ

119894 = 0 Therefore if the player 119894 is the null player in Q

with any (119878 119879) isin 3119873119894 thenQ119894 (119894 0) = (Q119894 (119878 cup 119894 119879) minus Q119894 (119878 119879))Q119894 (0 119894) = (Q119894 (119878 119879) minus Q119894 (119878 119879 cup 119894))

120601119894 (Q) = (Q(119878119879) (119894 0) minus Q(119878119879) (0 119894))(10)

Axiom 2 (total swings) If Q isin BG119873 then

sum119894isin119873

ΨQ119894 = sum(119878119879)isin3119873119894

(Q119894 (119878 cup 119894 119879) minus Q119894 (119878 119879 cup 119894)) (11)

Axiom 3 (transfer property) For any QG isin BG119873 we havethat

(120601119894 (Q) + 120601119894 (G)) = (120601119894 (Q⋁G) + 120601119894 (Q andG)) (12)

Axiom 4 (simple additivity) For any Q isin BG119873 such thatQ = G +H we have that

120601119894 (Q) = (120601119894 (G) + 120601119894 (H)) (13)

As a special ternary voting gamemodel the weighted ternaryvoting game can formulate a decision-making process thatnot all voters have the same amount of influence overthe decision Therefore votes of different voters are givendifferent weight This type of voting game model is usedin shareholder meetings where votes are weighted by thenumber of shares that each shareholder owns [11] In thisstudy we develop a ternary voting game model based onthe weighted voting approach In our game model mobiledevices are treated differently according to their socialpowers (F119898(119909)) and give different amounts of influenceto different members Therefore each coalition of players119878 sube 119873 has the sum of voting weights of its compo-nents that is W(119878) = sum119898119894isin119878F119898119894(119909) If the value ofW(119878) exceeds a certain quota players in 119878 win the game[11 19]

For the weighted ternary voting game we need twoquotas one is to block a decision and the other is toapprove it Formally it is represented by the voting body[P 120599F1198981(119909) F119898119899(119909)] where P is the quota for thedecision-taking and 120599 is the quota for the decision-blockingit can be seen as a particular case of n-player games with

Wireless Communications and Mobile Computing 7

three choices This situation can be formulated by a weightedternary voting game (X) ieX(119878 119879) 3119873 997888rarr minus1 0 1 [12]

X (119878 119879) =

1 if W (119878) ge P and W (119879) lt 120599minus1 if W (119878) lt P and W (119879) ge 1205990 otherwise

(14)

st P = 120572 times ( 119899sum119894=1

F119898119894 (119909))

and 120599 = 120573 times ( 119899sum119894=1

F119898119894 (119909))(15)

23 Communication and Computation Resource AllocationMechanisms Wireless bandwidth is a valuable and scarceresource in the SFIoT system Therefore the limited wirelessbandwidth has to be shared fair-efficiently by mobile usersIf the allocation method is not considered explicitly at thedesign stage of bandwidth allocation algorithms differentbandwidth requests can result in very unfair bandwidthallocations it may lead to a system inefficiency Usuallyvarious bandwidth allocation methods can be categorizedinto three ways proportional allocation (PA) constrainedallocation (CA) and equal loss allocation (EA) ways

(I) Proportional allocation (PA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max(119877119894 times Bsum119899119894=1 119877119894) m119894 otherwise

(16)

(II) Constrained allocation (CA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max (min 119877119894 119879119862 m119894) otherwise

st119899sum119894=1

min 119877119894 119879119862 = B

(17)

(III) Equal loss allocation (EA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max (119877119894 minusL119862) m119894 otherwise

st119899sum119894=1

max (119877119894 minusL119862) 0 = B

(18)

where B is the total available bandwidth amount and m119894is the predefined minimum bandwidth amount for the 119894119905ℎ

service 1198771le119894le119873 is the bandwidth request of mobile device119898119894 and 1198601le119894le119873 represents the decided bandwidth amountfor 119898119894 119879119862 and L119862 are the constraint thresholds which arechosen to satisfy each condition If the available bandwidthis enough or the F-AP can secure the new request 119860 119894from 119898119894 through the allocated bandwidth readjustment thenew requested application is accepted Otherwise the newrequested application is rejected

In our bandwidth allocation algorithm each F-APdecides its resource allocation way based on the weightedternary voting game InF119860119875119896 consider theX gamewithNF119860119875

119896

Individual 119898119894 isin NF119860119875119896

participates in the voting procedurewith his voting weight Usually the PA way is the mostcommon method and it can be seen as implicit consentTherefore for each voter we can assume that there are threeoptions ie pro-CA pro-EA and abstention Accordingto (14) the coalition 119878 consists of pro-CA voters and thecoalition 119879 consists of pro-EA voters If the outcome ofX(119878 119879) is 1 or -1 F119860119875119896 simply adopts the CA way or the EAway respectively in the bandwidth allocation process If theoutcome ofX(119878 119879) is 0F119860119875119896 adopts the PAway Based on thedecisions of individual devices Each F-AP adaptively decidesits communication resource allocation way

Like as the communication resource computation capac-ity in each F-AP is also a naturally limited resourceThereforethe computation power of F-AP is adaptively assigned to eachrequested application task Usually different applications canbe categorized into two classes class I (real-time data applica-tions) and class II (non-real-time data applications) accord-ing to the required QoE They not only require differentcomputation capacity but also have different QoE In orderto design the computation resource allocation algorithmeach F-AP needs to partition its own computation powerto support different type applications In this paper we alsoadopt the weighted ternary voting game model However incontrast to the voting game in our communication allocationalgorithm the developed computation allocation algorithmhas focused on the basic concept of pseudo-Banzhaf values

Consider another X game inF119860119875119896 with NF119860119875119896 Individual119898119894 isin NF119860119875

119896participates in the second voting procedure At

this time there are three options for each voter ie proclass Iservice proclass II service and abstention According to (14)the coalition 119878 consists of pro-class I service voters and thecoalition 119879 consists of pro-class II service voters Based on theoutcome of X(119878 119879) the F-AP estimates the preference ratio(120572119868) of class I application services as follows

120572119868= sum119898119894isin119878 (ΨQ

119898119894timesF119898119894 (119909))

(sum119898119894isin119878 (ΨQ119898119894timesF119898119894 (119909)) + sum119898119895isin119879 (ΨQ

119898119895timesF119898119894 (119909)))

(19)

According to each userrsquos pseudo-Banzhaf value and socialpower the 120572119868 value is decided and each F-AP partitions itstotal computation capacity (C) The division ratio for class I(or class II) type application services is 120572119868 (or 1 minus 120572119868) Andthen the partitioned computation capacity is distributed forits corresponding type application services while maximizing

8 Wireless Communications and Mobile Computing

Table 2 Application service type and requirements and system parameters

Type Application Bandwidthrequirement

Minimumrequirement (m)

Computationrequirement Duration (Avesec)

Class I Application I 119860=512 Mbps 119860=256 Mbps 98 MHz 180 secApplication II 119860=640 Mbps 119860=320 Mbps 256MHz 120 sec

Class II

Application III 119860=768 Mbps 119860=384 Mbps 128 MHz 240 secApplication IV 119860=896 Mbps 119860=458 Mbps 256MHz 300 secApplication V 119860=128 Gbps 119860=640 Mbps 384 MHz 360 secApplication VI 119860=152 Gbps 119860=860 Mbps 512 MHz 480 sec

Parameter Value DescriptionΓ 5 the maximum outcome of a user to normalize the social power120583 09 a parameter to controls the speed of social power saturation120572 075 a factor to decide the quota for the decision-taking120573 025 a factor to decide the quota for the decision- blocking120578119868 09 a control parameter to adjust the payoff of class I service120578119868119868 07 a control parameter to adjust the payoff of class II service120576 10 a control parameter for the sharing activity between social friends

the systemperformance To implement this process we adoptthe fundamental notion of the Nash Bargaining Solution(NBS) This is captured by the following optimization prob-lem

max[119860119898119894 ]

prod119898119894isin119878

119880119868119898119894 (119860119898119894) + max[119860119898119895 ]

prod119898119895isin119879

119880119868119868119898119895 (119860119898119895) (20)

st 119880119868119898119894 (119860119898119894) = 120578119868 times log2 (1 + 119860119898119894119877119898119894 )

119880119868119868119898119895 (119860119898119895) = 120578119868119868 times (1 minus exp(minus119860119898119894119877119898119894 ))sum119898119894isin119878

119860119898119894 le (120572119868 timesCF119860119875119896)

and sum119898119895isin119879

119860119898119895 le ((1 minus 120572119868) timesCF119860119875119896)

(21)

where 120578119868 (or 120578119868119868) is a control parameter to adjust the payoffof 119898119894 (or 119898119895) and CF119860119875

119896is the total computation capacity of

F119860119875119896

24 Main Steps of Proposed SFIoT Resource AllocationScheme Under the vision of the future internet the SIoTparadigm aims to connect anything while relying on the self-establishment of inter-thing social relationships In additionthe emerging paradigm of FC exhibits the right features forcoping with the aforementioned technological 5G networkissues In order to address this challenge the SFIoT paradigmhas been gaining momentum To effectively operate theSFIoT system the interactive relationship between F-APs andmobile devices is an important research topic and should beconsidered to design the resource allocation algorithms Inthis study we provide the communication and computationresource allocation scheme which is formulated according

to the BVG and NBS Owing to our two-phase game modelwe can get the most fair-efficient system performance bycombining both game approaches The main steps of theproposed scheme are described as follows

Step 1 Application types requirements and system controlparameters are determined by the simulation scenario (seesimulation assumptions in Section 3 and Table 2)

Step 2 Each individual F-AP monitors its covering area andresponds to its corresponding mobile devices Mobile devicesgenerate their application service requests in a distributedfashion and ask FC services to their corresponding F-AP

Step 3 Each mobile device has its own social relationshipthrough a social network close friends can probabilisticallyshare the outcome of FC services

Step 4 To reduce computation complexity individual F-APsexecute separately their two-phase games in parallel

Step 5 During the first game process each mobile devicetakes two votes one is to decide the communication resourceallocation way and the other is to partition the F-APrsquoscomputation capacity

Step 6 According to (14) the F-AP decides its communi-cation allocation way Based on our weighted ternary BVGmodel one of the PA CA and EA ways is adaptivelyselected And then the F-AP estimates the division ratio ofcomputation capacity using (19)

Step 7 During the second game process each F-AP inde-pendently distributes the partitioned computation capacity tocorresponding type applications According to (20) differentapplications can retain their FC services while maximizingthe system performance

Wireless Communications and Mobile Computing 9

Step 8 Based on the two-phase game model F-APs andmobile devices are hierarchically interconnected and inter-acting with one another to operate the SFIoT system

Step 9 Constantly each F-AP is self-monitoring its coveringarea When the game players are changed due to the mobiledevice migration the proposed scheme proceeds to Step 5 forthe next two-phase game procedure

3 Simulation Results and Discussion

In this section we evaluate the effectiveness of our proposedprotocol and analyze the simulation result while compar-ing it with that of the SDFC [14] SSEC [15] and CFIC[16] schemes To ensure a fair comparison the experi-mental setup consists of the following specifications andscenario

(i) Simulated SFIoT system consists of 25 F-APs and 100mobile devices F-APs are laid out in regular patternandmobile is randomly located in the F-APsrsquo coveringareas

(ii) For the FC services total computation capacity ofeach F-AP is 30 GHz and total communicationresource of each F-AP is 50 Gbps

(iii) Each mobile device individually generates servicerequests Service type of applications is randomlydecided and service request rate is Poisson process(120588) The offered rate range is varied from 0 to 30

(iv) The social friend number (|S119898| gt 0) of each mobiledevice is a random variable which has the maximumprobability at 3 while following a normal distribution

(v) Each social tie (0 le 119890119898119898 le 1) between mobiledevices is also a random variable which has themaximum probability at 05 while following a normaldistribution

(vi) The probability (P119904) that social friends119898119894 and119898119895 canshare the service outcome is defined asP119904 = 119890119898119894119898119895120576where 120576 is a control parameter for the sharing activity

(vii) We assume that there are no physical obstacles in theexperiments and each mobile device can contact itscorresponding F-AP for FC services

(viii) For calculation simplicity the communicationresource is specified in terms of basic units (BUs)where one BU is the minimum amount (eg 128Mbps in our system) of resource management

(ix) Using the iterative water-filling method the value ofNBS in (20) is obtained

(x) Network performancemeasures obtained on the basisof 100 simulation runs are plotted as functions of theoffered service request rate (120588)

(xi) Six different applications are assumed based on thedifferent computation and communication require-ment they are generated with equal probability

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

05 1 15 2 25 30Offered Application Load (Task Request Rate)

0

05

1

15

Qua

lity

of E

xper

ienc

e (Q

oE)

Figure 1 SFIoT quality of experience (QoE)

(xii) Performance criteria obtained through simulation areQoE resource usability and task failure probabilitythese simulation metrics are evaluated mainly todemonstrate the validity of our proposed method

To facilitate the development and implementation of oursimulator Table 2 lists application service type requirementsand system control parameters

Figure 1 shows the QoE of each scheme in differentapplication load scenarios For this evaluationQoE is definedas the normalized resource loss ratio than the originallyrequested amount From the viewpoint of mobile users QoEis a key factor to evaluate their service satisfactions In theproposed scheme the system resource is differently allocatedbased on the voting approach According to the desirablefeatures of ternary voting approach the limited resourcecan be assigned while effectively reflecting the preferenceinclination of users Our two-phase game approach can wellorchestrate the SFIoT system infrastructure to maximize theQoE of mobile users The result of Figure 1 proves that ourtwo-step game method is a socially beneficial way to achievea better QoE than other existing schemes

The obtained results of task failure probabilities aredepicted in Figure 2 This performance criterion is stronglyrelated to the SFIoT system throughput which is estimatedbased on the ratio of successfully completed applications Asexpected the application load increases the available systemresource is exhausted to support new application servicesTherefore task failure probabilities increase proportionatelywith the application request rate The resulting curves inFigure 2 allow us to see that our proposed scheme has gaineda lower task failure probability than other existing schemes Itis therefore worth to say that under different application loadconditions our proposed scheme can perform excellentlyto maintain a superior system throughput based on the

10 Wireless Communications and Mobile Computing

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Task

Fai

lure

Pro

babi

lity

05 1 15 2 25 30Offered Application Load (Task Request Rate)

Figure 2 SFIoT task failure probability

0 05 1 15 2 25 3Offered Application Load (Task Request Rate)

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Syste

m R

esou

rce U

tiliz

atio

n

Figure 3 Resource utilization

self-controlled management policies It is a highly desirableproperty for the SFIoT system operations

Figure 3 illustrates the resource utilization of F-APs Wemeasure this performance metric to show and determinewhether our game-based approach can provide the systemefficiency As it was explained control decisions in theproposed scheme are made in an interactive two-phasegame model More specifically the proposed approach inthis paper can adapt the current SFIoT system conditionwhile dynamically adjusting the control decisions Whenthe application load is low (below 05) the performance ofeach scheme is not much difference As the application load

rate increases the resource utilization also increases All theschemes have similar trends however due to our approachrsquosflexibility adaptability and responsiveness to current SFIoTsystem conditions the F-AP resource in our proposed schemecan be used more efficiently under widely different anddiversified application load situations

Through simulation experiments the obtained resultsin Figures 1ndash3 clearly demonstrate that the suitability andfeasibility of our proposed scheme while revealing enhancedperspectives regarding QoE throughput and resource uti-lization in the SFIoT system In particular the BVG and NBSprocesses work together to reach an agreement that givesmutual advantages for mobile devices and F-APs two gamemodels are sophisticatedly combined and dependent on eachotherTherefore they act cooperatively to satisfy the differentperformance requirements

4 Summary and Conclusions

The past decade has witnessed an explosive growth of IoTwhere a variety of smart mobile devices offer a plethora ofdifferent applications Under the vision of future networksFC and social network principles are integrated into the IoTparadigm in order to enforce the distributed self-cooperationand self-management of mobile devices In this paper wepresent a novel resource allocation scheme for the SFIoTsystem According to the BVG and NBS approaches ourproposed scheme is designed as two phases (i) mobiledevices take votes to decide the resource allocation methodsbased on the weighted ternary BVG model and (ii) the par-titioned computation capacity is fair-efficiently distributed toeach application by using the NBS These two game modelscontinue iteratively in a step-by-step manner in accordancewith the current SFIoT system informationThe experimentalsimulation results prove that our proposed protocol canefficiently improve theQoE system throughput and resourceutilization compared with the existing SDFC SSEC andCFIC schemes As directions for future research we aim atinvestigating the privacy and energy issues for the SFIoTsystemWe especially plan to add a blockchain-based securitylayer to the system In addition we also plan to develop anoncooperative gamemodelwith theoretical analysis andwillfocus on the addition of more features to increase the SFIoTsystem reliability in an unconstrained environment Lastbut not least machine learning techniques can be appliedto complement our proposed scheme It will be a potentialdirection and another possible extension to this work

Data Availability

The data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

The author declares that there are no conflicts of interestregarding the publication of this paper

Wireless Communications and Mobile Computing 11

Authorsrsquo Contributions

The sole author contributes to all research work

Acknowledgments

This research was supported by the MSIT (Ministry ofScience and ICT) Korea under the ITRC (InformationTechnology Research Center) support program (IITP-2018-2018-0-01799) supervised by the IITP (Institute for Infor-mation amp communications Technology Promotion) and byInstitute for Information amp communications TechnologyPromotion (IITP) grant funded by the Korea government(MSIT) (No2017-0-00498 A Development of Deidentifica-tion Technique based on Differential privacy)

References

[1] A M Ortiz D Hussein S Park S N Han and N Crespi ldquoThecluster between internet of things and social networks reviewand research challengesrdquo IEEE Internet of ings Journal vol 1no 3 pp 206ndash215 2014

[2] Z Li R Chen L Liu andGMin ldquoDynamic resource discoverybased on preference andmovement pattern similarity for large-scale social internet of thingsrdquo IEEE Internet of ings Journalvol 3 no 4 pp 581ndash589 2016

[3] H Yang and L Zhang ldquoResponse time prediction of IoT servicebased on time similarityrdquo Journal of Computing Science andEngineering vol 11 no 3 pp 100ndash108 2017

[4] E Baccarelli M Scarpiniti P G Naranjo and L Vaca-Cardenas ldquoFog of social IoT when the fog becomes socialrdquoIEEE Network vol 32 no 4 pp 68ndash80 2018

[5] H Shah-Mansouri and V W S Wong ldquoHierarchical fog-cloudcomputing for iot systems a computation offloading gamerdquoIEEE Internet ofings Journal vol 5 no 4 pp 3246ndash3257 2018

[6] K Lee and I Shin ldquoUser mobility model based computationoffloading decision for mobile cloudrdquo Journal of ComputingScience and Engineering vol 9 no 3 pp 155ndash162 2015

[7] Y Liu M J Lee and Y Zheng ldquoAdaptive multi-resourceallocation for cloudlet-based mobile cloud computing systemrdquoIEEE Transactions on Mobile Computing vol 15 no 10 pp2398ndash2410 2016

[8] K Kim S Uno and M Kim ldquoAdaptive QoS mechanism forwireless mobile networkrdquo Journal of Computing Science andEngineering vol 4 no 2 pp 153ndash172 2010

[9] A Pande V Ramamurthi and P Mohapatra ldquoQuality-orientedvideo delivery over LTErdquo Journal of Computing Science andEngineering vol 7 no 3 pp 168ndash176 2013

[10] S Kim and P K Varshney ldquoAn integrated adaptive bandwidth-management framework for QoS-sensitive multimedia cellularnetworksrdquo IEEE Transactions on Vehicular Technology vol 53no 3 pp 835ndash846 2004

[11] K SungwookGameeory Applications in Network Design IGIGlobal Hershey PA USA 2014

[12] J M Bilbao J R Fernandez N Jimenez and J J LopezldquoBiprobabilistic values for bicooperative gamesrdquo DiscreteApplied Mathematics vol 156 pp 2698ndash2711 2008

[13] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheshapley value for bicooperative gamesrdquo Annals of OperationsResearch vol 158 pp 99ndash115 2008

[14] L Liu Z Chang and X Guo ldquoSocially-aware dynamic compu-tation offloading scheme for fog computing system with energyharvesting devicesrdquo IEEE Internet of ings Journal vol 5 no3 pp 1869ndash1879 2018

[15] R Dautov S Distefano D Bruneo F Longo G Merlino andA Puliafito ldquoData processing in cyber-physical-social systemsthrough edge computingrdquo IEEE Access vol 6 pp 29822ndash298352018

[16] Y-C Chen Y-C Chang C-H Chen Y-S Lin J-L Chen andY-Y Chang ldquoCloud-fog computing for information-centricinternet-of-things applicationsrdquo in Proceedings of the 2017 IEEEInternational Conference on Applied System Innovation ICASI2017 pp 637ndash640 May 2017

[17] C Jiang L Kuang Z Han Y Ren and L Hanzo ldquoInformationcredibility modeling in cooperative networks equilibrium andmechanism designrdquo IEEE Journal on Selected Areas in Commu-nications vol 35 no 2 pp 432ndash448 2017

[18] S Kim ldquoFog radio access network system control scheme basedon the embedded game modelrdquo EURASIP Journal on WirelessCommunications and Networking vol 2017 no 113 pp 1ndash142017

[19] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheBanzhaf power index for ternary bicooperative gamesrdquoDiscreteApplied Mathematics vol 158 pp 967ndash980 2010

[20] M TsurumiM Inuiguchi and A Nishimura ldquoPseudo-banzhafvalues in bicooperative gamesrdquo in Proceedings of the IEEEInternational Conference on Systems Man and Cybernetics pp1138ndash1143 2005

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Active and Passive Electronic Components

VLSI Design

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Shock and Vibration

Hindawiwwwhindawicom Volume 2018

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

SensorsJournal of

Hindawiwwwhindawicom Volume 2018

International Journal of

RotatingMachinery

Hindawiwwwhindawicom Volume 2018

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Navigation and Observation

International Journal of

Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom

6 Wireless Communications and Mobile Computing

= 1 if (119860 119861) = (119878 119879) and (119878 119879) = (0 0)0 otherwise

(5)

The superior unanimity game Q(119878119879)(119860 119861) | 3119873 997888rarr R isdefined by

Q(119878119879) (119860 119861)=

1 if (119878 119879) ⊑ (119860 119861) and (119860 119861) = (0 0)0 otherwise

(6)

The inferior unanimity game Q(119878119879)(119860 119861) | 3119873 997888rarr R isdefined by

Q(119878119879) (119860 119861)=

minus1 if (119860 119861) ⊑ (119878 119879) and (119860 119861) = (0 0)0 otherwise

(7)

Voting game is a cooperative game model that incorporateselements from social choice theory This game model can beused to analyze situations where voters would vote in favor ofor against a decision [11] Simple voting game with abstentioncan be formulated as a bicooperative game It has been studiedunder the name of ternary voting game [19 20]

Let N be a set of n voters that has to choose between apair of two alternatives 119878 and 119879 Every voter 119894 isin 119873 has apreference over the two alternatives 119878 and 119879 Thus the payoffof voter 119894 (Q119894) is defined by Q119894 3119873 997888rarr minus1 0 1 based onthe majority rule Formally a ternary voting game Q isin BG119873

satisfies the following conditions [19]

(I) For every bicoalition pair (119878 119879)isin 3119873 its worthQ(119878 119879) isin minus1 0 1

(II) If (119878 119879) (119878 ) isin 3119873 with (119878 119879) ⊑ (119878 ) thenQ(119878 119879) le Q(119878 )

(III)Q119894

=

1 if (119894 isin 119878 and |119878| gt |119879|) or (119894 isin 119879 and |119879| gt |119878|)minus1 if (119894 isin 119878 and |119878| lt |119879|) or (119894 isin 119879 and |119879| lt |119878|)0 if voter i abstains form voting or |119878| = |119879|

(8)

where |119878| is the cardinality of coalition 119878 For voting gamesthe analysis of voting power is the main issue In 1965Banzhaf value was introduced by John F Banzhaf III based onprobabilistic analysis of the individual voters in a block votingsystem It depends on the number of ways in which eachvoter can effect a swing in the game In 2010 J Bilbao et alintroduced a pseudo-Banzhaf value ie ΨQ

119894 for the ternaryvoting game Q it can be interpreted as a voterrsquos probabilisticpower index [19 20]

ΨQ119894 = 13119899minus1 times sum

(119878119879)isin3119873119894

(Q119894 (119878 cup 119894 119879) minus Q119894 (119878 119879 cup 119894)) (9)

The pseudo-Banzhaf value satisfies four axioms Null PlayerTotal Swings Transfer Property and Simple Additivity[19 20]

Axiom 1 (null player) If 119894 isin 119873 is a null player in Q isin BG119873then ΨQ

119894 = 0 Therefore if the player 119894 is the null player in Q

with any (119878 119879) isin 3119873119894 thenQ119894 (119894 0) = (Q119894 (119878 cup 119894 119879) minus Q119894 (119878 119879))Q119894 (0 119894) = (Q119894 (119878 119879) minus Q119894 (119878 119879 cup 119894))

120601119894 (Q) = (Q(119878119879) (119894 0) minus Q(119878119879) (0 119894))(10)

Axiom 2 (total swings) If Q isin BG119873 then

sum119894isin119873

ΨQ119894 = sum(119878119879)isin3119873119894

(Q119894 (119878 cup 119894 119879) minus Q119894 (119878 119879 cup 119894)) (11)

Axiom 3 (transfer property) For any QG isin BG119873 we havethat

(120601119894 (Q) + 120601119894 (G)) = (120601119894 (Q⋁G) + 120601119894 (Q andG)) (12)

Axiom 4 (simple additivity) For any Q isin BG119873 such thatQ = G +H we have that

120601119894 (Q) = (120601119894 (G) + 120601119894 (H)) (13)

As a special ternary voting gamemodel the weighted ternaryvoting game can formulate a decision-making process thatnot all voters have the same amount of influence overthe decision Therefore votes of different voters are givendifferent weight This type of voting game model is usedin shareholder meetings where votes are weighted by thenumber of shares that each shareholder owns [11] In thisstudy we develop a ternary voting game model based onthe weighted voting approach In our game model mobiledevices are treated differently according to their socialpowers (F119898(119909)) and give different amounts of influenceto different members Therefore each coalition of players119878 sube 119873 has the sum of voting weights of its compo-nents that is W(119878) = sum119898119894isin119878F119898119894(119909) If the value ofW(119878) exceeds a certain quota players in 119878 win the game[11 19]

For the weighted ternary voting game we need twoquotas one is to block a decision and the other is toapprove it Formally it is represented by the voting body[P 120599F1198981(119909) F119898119899(119909)] where P is the quota for thedecision-taking and 120599 is the quota for the decision-blockingit can be seen as a particular case of n-player games with

Wireless Communications and Mobile Computing 7

three choices This situation can be formulated by a weightedternary voting game (X) ieX(119878 119879) 3119873 997888rarr minus1 0 1 [12]

X (119878 119879) =

1 if W (119878) ge P and W (119879) lt 120599minus1 if W (119878) lt P and W (119879) ge 1205990 otherwise

(14)

st P = 120572 times ( 119899sum119894=1

F119898119894 (119909))

and 120599 = 120573 times ( 119899sum119894=1

F119898119894 (119909))(15)

23 Communication and Computation Resource AllocationMechanisms Wireless bandwidth is a valuable and scarceresource in the SFIoT system Therefore the limited wirelessbandwidth has to be shared fair-efficiently by mobile usersIf the allocation method is not considered explicitly at thedesign stage of bandwidth allocation algorithms differentbandwidth requests can result in very unfair bandwidthallocations it may lead to a system inefficiency Usuallyvarious bandwidth allocation methods can be categorizedinto three ways proportional allocation (PA) constrainedallocation (CA) and equal loss allocation (EA) ways

(I) Proportional allocation (PA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max(119877119894 times Bsum119899119894=1 119877119894) m119894 otherwise

(16)

(II) Constrained allocation (CA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max (min 119877119894 119879119862 m119894) otherwise

st119899sum119894=1

min 119877119894 119879119862 = B

(17)

(III) Equal loss allocation (EA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max (119877119894 minusL119862) m119894 otherwise

st119899sum119894=1

max (119877119894 minusL119862) 0 = B

(18)

where B is the total available bandwidth amount and m119894is the predefined minimum bandwidth amount for the 119894119905ℎ

service 1198771le119894le119873 is the bandwidth request of mobile device119898119894 and 1198601le119894le119873 represents the decided bandwidth amountfor 119898119894 119879119862 and L119862 are the constraint thresholds which arechosen to satisfy each condition If the available bandwidthis enough or the F-AP can secure the new request 119860 119894from 119898119894 through the allocated bandwidth readjustment thenew requested application is accepted Otherwise the newrequested application is rejected

In our bandwidth allocation algorithm each F-APdecides its resource allocation way based on the weightedternary voting game InF119860119875119896 consider theX gamewithNF119860119875

119896

Individual 119898119894 isin NF119860119875119896

participates in the voting procedurewith his voting weight Usually the PA way is the mostcommon method and it can be seen as implicit consentTherefore for each voter we can assume that there are threeoptions ie pro-CA pro-EA and abstention Accordingto (14) the coalition 119878 consists of pro-CA voters and thecoalition 119879 consists of pro-EA voters If the outcome ofX(119878 119879) is 1 or -1 F119860119875119896 simply adopts the CA way or the EAway respectively in the bandwidth allocation process If theoutcome ofX(119878 119879) is 0F119860119875119896 adopts the PAway Based on thedecisions of individual devices Each F-AP adaptively decidesits communication resource allocation way

Like as the communication resource computation capac-ity in each F-AP is also a naturally limited resourceThereforethe computation power of F-AP is adaptively assigned to eachrequested application task Usually different applications canbe categorized into two classes class I (real-time data applica-tions) and class II (non-real-time data applications) accord-ing to the required QoE They not only require differentcomputation capacity but also have different QoE In orderto design the computation resource allocation algorithmeach F-AP needs to partition its own computation powerto support different type applications In this paper we alsoadopt the weighted ternary voting game model However incontrast to the voting game in our communication allocationalgorithm the developed computation allocation algorithmhas focused on the basic concept of pseudo-Banzhaf values

Consider another X game inF119860119875119896 with NF119860119875119896 Individual119898119894 isin NF119860119875

119896participates in the second voting procedure At

this time there are three options for each voter ie proclass Iservice proclass II service and abstention According to (14)the coalition 119878 consists of pro-class I service voters and thecoalition 119879 consists of pro-class II service voters Based on theoutcome of X(119878 119879) the F-AP estimates the preference ratio(120572119868) of class I application services as follows

120572119868= sum119898119894isin119878 (ΨQ

119898119894timesF119898119894 (119909))

(sum119898119894isin119878 (ΨQ119898119894timesF119898119894 (119909)) + sum119898119895isin119879 (ΨQ

119898119895timesF119898119894 (119909)))

(19)

According to each userrsquos pseudo-Banzhaf value and socialpower the 120572119868 value is decided and each F-AP partitions itstotal computation capacity (C) The division ratio for class I(or class II) type application services is 120572119868 (or 1 minus 120572119868) Andthen the partitioned computation capacity is distributed forits corresponding type application services while maximizing

8 Wireless Communications and Mobile Computing

Table 2 Application service type and requirements and system parameters

Type Application Bandwidthrequirement

Minimumrequirement (m)

Computationrequirement Duration (Avesec)

Class I Application I 119860=512 Mbps 119860=256 Mbps 98 MHz 180 secApplication II 119860=640 Mbps 119860=320 Mbps 256MHz 120 sec

Class II

Application III 119860=768 Mbps 119860=384 Mbps 128 MHz 240 secApplication IV 119860=896 Mbps 119860=458 Mbps 256MHz 300 secApplication V 119860=128 Gbps 119860=640 Mbps 384 MHz 360 secApplication VI 119860=152 Gbps 119860=860 Mbps 512 MHz 480 sec

Parameter Value DescriptionΓ 5 the maximum outcome of a user to normalize the social power120583 09 a parameter to controls the speed of social power saturation120572 075 a factor to decide the quota for the decision-taking120573 025 a factor to decide the quota for the decision- blocking120578119868 09 a control parameter to adjust the payoff of class I service120578119868119868 07 a control parameter to adjust the payoff of class II service120576 10 a control parameter for the sharing activity between social friends

the systemperformance To implement this process we adoptthe fundamental notion of the Nash Bargaining Solution(NBS) This is captured by the following optimization prob-lem

max[119860119898119894 ]

prod119898119894isin119878

119880119868119898119894 (119860119898119894) + max[119860119898119895 ]

prod119898119895isin119879

119880119868119868119898119895 (119860119898119895) (20)

st 119880119868119898119894 (119860119898119894) = 120578119868 times log2 (1 + 119860119898119894119877119898119894 )

119880119868119868119898119895 (119860119898119895) = 120578119868119868 times (1 minus exp(minus119860119898119894119877119898119894 ))sum119898119894isin119878

119860119898119894 le (120572119868 timesCF119860119875119896)

and sum119898119895isin119879

119860119898119895 le ((1 minus 120572119868) timesCF119860119875119896)

(21)

where 120578119868 (or 120578119868119868) is a control parameter to adjust the payoffof 119898119894 (or 119898119895) and CF119860119875

119896is the total computation capacity of

F119860119875119896

24 Main Steps of Proposed SFIoT Resource AllocationScheme Under the vision of the future internet the SIoTparadigm aims to connect anything while relying on the self-establishment of inter-thing social relationships In additionthe emerging paradigm of FC exhibits the right features forcoping with the aforementioned technological 5G networkissues In order to address this challenge the SFIoT paradigmhas been gaining momentum To effectively operate theSFIoT system the interactive relationship between F-APs andmobile devices is an important research topic and should beconsidered to design the resource allocation algorithms Inthis study we provide the communication and computationresource allocation scheme which is formulated according

to the BVG and NBS Owing to our two-phase game modelwe can get the most fair-efficient system performance bycombining both game approaches The main steps of theproposed scheme are described as follows

Step 1 Application types requirements and system controlparameters are determined by the simulation scenario (seesimulation assumptions in Section 3 and Table 2)

Step 2 Each individual F-AP monitors its covering area andresponds to its corresponding mobile devices Mobile devicesgenerate their application service requests in a distributedfashion and ask FC services to their corresponding F-AP

Step 3 Each mobile device has its own social relationshipthrough a social network close friends can probabilisticallyshare the outcome of FC services

Step 4 To reduce computation complexity individual F-APsexecute separately their two-phase games in parallel

Step 5 During the first game process each mobile devicetakes two votes one is to decide the communication resourceallocation way and the other is to partition the F-APrsquoscomputation capacity

Step 6 According to (14) the F-AP decides its communi-cation allocation way Based on our weighted ternary BVGmodel one of the PA CA and EA ways is adaptivelyselected And then the F-AP estimates the division ratio ofcomputation capacity using (19)

Step 7 During the second game process each F-AP inde-pendently distributes the partitioned computation capacity tocorresponding type applications According to (20) differentapplications can retain their FC services while maximizingthe system performance

Wireless Communications and Mobile Computing 9

Step 8 Based on the two-phase game model F-APs andmobile devices are hierarchically interconnected and inter-acting with one another to operate the SFIoT system

Step 9 Constantly each F-AP is self-monitoring its coveringarea When the game players are changed due to the mobiledevice migration the proposed scheme proceeds to Step 5 forthe next two-phase game procedure

3 Simulation Results and Discussion

In this section we evaluate the effectiveness of our proposedprotocol and analyze the simulation result while compar-ing it with that of the SDFC [14] SSEC [15] and CFIC[16] schemes To ensure a fair comparison the experi-mental setup consists of the following specifications andscenario

(i) Simulated SFIoT system consists of 25 F-APs and 100mobile devices F-APs are laid out in regular patternandmobile is randomly located in the F-APsrsquo coveringareas

(ii) For the FC services total computation capacity ofeach F-AP is 30 GHz and total communicationresource of each F-AP is 50 Gbps

(iii) Each mobile device individually generates servicerequests Service type of applications is randomlydecided and service request rate is Poisson process(120588) The offered rate range is varied from 0 to 30

(iv) The social friend number (|S119898| gt 0) of each mobiledevice is a random variable which has the maximumprobability at 3 while following a normal distribution

(v) Each social tie (0 le 119890119898119898 le 1) between mobiledevices is also a random variable which has themaximum probability at 05 while following a normaldistribution

(vi) The probability (P119904) that social friends119898119894 and119898119895 canshare the service outcome is defined asP119904 = 119890119898119894119898119895120576where 120576 is a control parameter for the sharing activity

(vii) We assume that there are no physical obstacles in theexperiments and each mobile device can contact itscorresponding F-AP for FC services

(viii) For calculation simplicity the communicationresource is specified in terms of basic units (BUs)where one BU is the minimum amount (eg 128Mbps in our system) of resource management

(ix) Using the iterative water-filling method the value ofNBS in (20) is obtained

(x) Network performancemeasures obtained on the basisof 100 simulation runs are plotted as functions of theoffered service request rate (120588)

(xi) Six different applications are assumed based on thedifferent computation and communication require-ment they are generated with equal probability

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

05 1 15 2 25 30Offered Application Load (Task Request Rate)

0

05

1

15

Qua

lity

of E

xper

ienc

e (Q

oE)

Figure 1 SFIoT quality of experience (QoE)

(xii) Performance criteria obtained through simulation areQoE resource usability and task failure probabilitythese simulation metrics are evaluated mainly todemonstrate the validity of our proposed method

To facilitate the development and implementation of oursimulator Table 2 lists application service type requirementsand system control parameters

Figure 1 shows the QoE of each scheme in differentapplication load scenarios For this evaluationQoE is definedas the normalized resource loss ratio than the originallyrequested amount From the viewpoint of mobile users QoEis a key factor to evaluate their service satisfactions In theproposed scheme the system resource is differently allocatedbased on the voting approach According to the desirablefeatures of ternary voting approach the limited resourcecan be assigned while effectively reflecting the preferenceinclination of users Our two-phase game approach can wellorchestrate the SFIoT system infrastructure to maximize theQoE of mobile users The result of Figure 1 proves that ourtwo-step game method is a socially beneficial way to achievea better QoE than other existing schemes

The obtained results of task failure probabilities aredepicted in Figure 2 This performance criterion is stronglyrelated to the SFIoT system throughput which is estimatedbased on the ratio of successfully completed applications Asexpected the application load increases the available systemresource is exhausted to support new application servicesTherefore task failure probabilities increase proportionatelywith the application request rate The resulting curves inFigure 2 allow us to see that our proposed scheme has gaineda lower task failure probability than other existing schemes Itis therefore worth to say that under different application loadconditions our proposed scheme can perform excellentlyto maintain a superior system throughput based on the

10 Wireless Communications and Mobile Computing

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Task

Fai

lure

Pro

babi

lity

05 1 15 2 25 30Offered Application Load (Task Request Rate)

Figure 2 SFIoT task failure probability

0 05 1 15 2 25 3Offered Application Load (Task Request Rate)

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Syste

m R

esou

rce U

tiliz

atio

n

Figure 3 Resource utilization

self-controlled management policies It is a highly desirableproperty for the SFIoT system operations

Figure 3 illustrates the resource utilization of F-APs Wemeasure this performance metric to show and determinewhether our game-based approach can provide the systemefficiency As it was explained control decisions in theproposed scheme are made in an interactive two-phasegame model More specifically the proposed approach inthis paper can adapt the current SFIoT system conditionwhile dynamically adjusting the control decisions Whenthe application load is low (below 05) the performance ofeach scheme is not much difference As the application load

rate increases the resource utilization also increases All theschemes have similar trends however due to our approachrsquosflexibility adaptability and responsiveness to current SFIoTsystem conditions the F-AP resource in our proposed schemecan be used more efficiently under widely different anddiversified application load situations

Through simulation experiments the obtained resultsin Figures 1ndash3 clearly demonstrate that the suitability andfeasibility of our proposed scheme while revealing enhancedperspectives regarding QoE throughput and resource uti-lization in the SFIoT system In particular the BVG and NBSprocesses work together to reach an agreement that givesmutual advantages for mobile devices and F-APs two gamemodels are sophisticatedly combined and dependent on eachotherTherefore they act cooperatively to satisfy the differentperformance requirements

4 Summary and Conclusions

The past decade has witnessed an explosive growth of IoTwhere a variety of smart mobile devices offer a plethora ofdifferent applications Under the vision of future networksFC and social network principles are integrated into the IoTparadigm in order to enforce the distributed self-cooperationand self-management of mobile devices In this paper wepresent a novel resource allocation scheme for the SFIoTsystem According to the BVG and NBS approaches ourproposed scheme is designed as two phases (i) mobiledevices take votes to decide the resource allocation methodsbased on the weighted ternary BVG model and (ii) the par-titioned computation capacity is fair-efficiently distributed toeach application by using the NBS These two game modelscontinue iteratively in a step-by-step manner in accordancewith the current SFIoT system informationThe experimentalsimulation results prove that our proposed protocol canefficiently improve theQoE system throughput and resourceutilization compared with the existing SDFC SSEC andCFIC schemes As directions for future research we aim atinvestigating the privacy and energy issues for the SFIoTsystemWe especially plan to add a blockchain-based securitylayer to the system In addition we also plan to develop anoncooperative gamemodelwith theoretical analysis andwillfocus on the addition of more features to increase the SFIoTsystem reliability in an unconstrained environment Lastbut not least machine learning techniques can be appliedto complement our proposed scheme It will be a potentialdirection and another possible extension to this work

Data Availability

The data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

The author declares that there are no conflicts of interestregarding the publication of this paper

Wireless Communications and Mobile Computing 11

Authorsrsquo Contributions

The sole author contributes to all research work

Acknowledgments

This research was supported by the MSIT (Ministry ofScience and ICT) Korea under the ITRC (InformationTechnology Research Center) support program (IITP-2018-2018-0-01799) supervised by the IITP (Institute for Infor-mation amp communications Technology Promotion) and byInstitute for Information amp communications TechnologyPromotion (IITP) grant funded by the Korea government(MSIT) (No2017-0-00498 A Development of Deidentifica-tion Technique based on Differential privacy)

References

[1] A M Ortiz D Hussein S Park S N Han and N Crespi ldquoThecluster between internet of things and social networks reviewand research challengesrdquo IEEE Internet of ings Journal vol 1no 3 pp 206ndash215 2014

[2] Z Li R Chen L Liu andGMin ldquoDynamic resource discoverybased on preference andmovement pattern similarity for large-scale social internet of thingsrdquo IEEE Internet of ings Journalvol 3 no 4 pp 581ndash589 2016

[3] H Yang and L Zhang ldquoResponse time prediction of IoT servicebased on time similarityrdquo Journal of Computing Science andEngineering vol 11 no 3 pp 100ndash108 2017

[4] E Baccarelli M Scarpiniti P G Naranjo and L Vaca-Cardenas ldquoFog of social IoT when the fog becomes socialrdquoIEEE Network vol 32 no 4 pp 68ndash80 2018

[5] H Shah-Mansouri and V W S Wong ldquoHierarchical fog-cloudcomputing for iot systems a computation offloading gamerdquoIEEE Internet ofings Journal vol 5 no 4 pp 3246ndash3257 2018

[6] K Lee and I Shin ldquoUser mobility model based computationoffloading decision for mobile cloudrdquo Journal of ComputingScience and Engineering vol 9 no 3 pp 155ndash162 2015

[7] Y Liu M J Lee and Y Zheng ldquoAdaptive multi-resourceallocation for cloudlet-based mobile cloud computing systemrdquoIEEE Transactions on Mobile Computing vol 15 no 10 pp2398ndash2410 2016

[8] K Kim S Uno and M Kim ldquoAdaptive QoS mechanism forwireless mobile networkrdquo Journal of Computing Science andEngineering vol 4 no 2 pp 153ndash172 2010

[9] A Pande V Ramamurthi and P Mohapatra ldquoQuality-orientedvideo delivery over LTErdquo Journal of Computing Science andEngineering vol 7 no 3 pp 168ndash176 2013

[10] S Kim and P K Varshney ldquoAn integrated adaptive bandwidth-management framework for QoS-sensitive multimedia cellularnetworksrdquo IEEE Transactions on Vehicular Technology vol 53no 3 pp 835ndash846 2004

[11] K SungwookGameeory Applications in Network Design IGIGlobal Hershey PA USA 2014

[12] J M Bilbao J R Fernandez N Jimenez and J J LopezldquoBiprobabilistic values for bicooperative gamesrdquo DiscreteApplied Mathematics vol 156 pp 2698ndash2711 2008

[13] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheshapley value for bicooperative gamesrdquo Annals of OperationsResearch vol 158 pp 99ndash115 2008

[14] L Liu Z Chang and X Guo ldquoSocially-aware dynamic compu-tation offloading scheme for fog computing system with energyharvesting devicesrdquo IEEE Internet of ings Journal vol 5 no3 pp 1869ndash1879 2018

[15] R Dautov S Distefano D Bruneo F Longo G Merlino andA Puliafito ldquoData processing in cyber-physical-social systemsthrough edge computingrdquo IEEE Access vol 6 pp 29822ndash298352018

[16] Y-C Chen Y-C Chang C-H Chen Y-S Lin J-L Chen andY-Y Chang ldquoCloud-fog computing for information-centricinternet-of-things applicationsrdquo in Proceedings of the 2017 IEEEInternational Conference on Applied System Innovation ICASI2017 pp 637ndash640 May 2017

[17] C Jiang L Kuang Z Han Y Ren and L Hanzo ldquoInformationcredibility modeling in cooperative networks equilibrium andmechanism designrdquo IEEE Journal on Selected Areas in Commu-nications vol 35 no 2 pp 432ndash448 2017

[18] S Kim ldquoFog radio access network system control scheme basedon the embedded game modelrdquo EURASIP Journal on WirelessCommunications and Networking vol 2017 no 113 pp 1ndash142017

[19] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheBanzhaf power index for ternary bicooperative gamesrdquoDiscreteApplied Mathematics vol 158 pp 967ndash980 2010

[20] M TsurumiM Inuiguchi and A Nishimura ldquoPseudo-banzhafvalues in bicooperative gamesrdquo in Proceedings of the IEEEInternational Conference on Systems Man and Cybernetics pp1138ndash1143 2005

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Active and Passive Electronic Components

VLSI Design

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Shock and Vibration

Hindawiwwwhindawicom Volume 2018

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

SensorsJournal of

Hindawiwwwhindawicom Volume 2018

International Journal of

RotatingMachinery

Hindawiwwwhindawicom Volume 2018

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Navigation and Observation

International Journal of

Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom

Wireless Communications and Mobile Computing 7

three choices This situation can be formulated by a weightedternary voting game (X) ieX(119878 119879) 3119873 997888rarr minus1 0 1 [12]

X (119878 119879) =

1 if W (119878) ge P and W (119879) lt 120599minus1 if W (119878) lt P and W (119879) ge 1205990 otherwise

(14)

st P = 120572 times ( 119899sum119894=1

F119898119894 (119909))

and 120599 = 120573 times ( 119899sum119894=1

F119898119894 (119909))(15)

23 Communication and Computation Resource AllocationMechanisms Wireless bandwidth is a valuable and scarceresource in the SFIoT system Therefore the limited wirelessbandwidth has to be shared fair-efficiently by mobile usersIf the allocation method is not considered explicitly at thedesign stage of bandwidth allocation algorithms differentbandwidth requests can result in very unfair bandwidthallocations it may lead to a system inefficiency Usuallyvarious bandwidth allocation methods can be categorizedinto three ways proportional allocation (PA) constrainedallocation (CA) and equal loss allocation (EA) ways

(I) Proportional allocation (PA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max(119877119894 times Bsum119899119894=1 119877119894) m119894 otherwise

(16)

(II) Constrained allocation (CA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max (min 119877119894 119879119862 m119894) otherwise

st119899sum119894=1

min 119877119894 119879119862 = B

(17)

(III) Equal loss allocation (EA)

119860 119894 = 119877119894 if119899sum119894=1

119877119894 le B

119860 119894 = max (119877119894 minusL119862) m119894 otherwise

st119899sum119894=1

max (119877119894 minusL119862) 0 = B

(18)

where B is the total available bandwidth amount and m119894is the predefined minimum bandwidth amount for the 119894119905ℎ

service 1198771le119894le119873 is the bandwidth request of mobile device119898119894 and 1198601le119894le119873 represents the decided bandwidth amountfor 119898119894 119879119862 and L119862 are the constraint thresholds which arechosen to satisfy each condition If the available bandwidthis enough or the F-AP can secure the new request 119860 119894from 119898119894 through the allocated bandwidth readjustment thenew requested application is accepted Otherwise the newrequested application is rejected

In our bandwidth allocation algorithm each F-APdecides its resource allocation way based on the weightedternary voting game InF119860119875119896 consider theX gamewithNF119860119875

119896

Individual 119898119894 isin NF119860119875119896

participates in the voting procedurewith his voting weight Usually the PA way is the mostcommon method and it can be seen as implicit consentTherefore for each voter we can assume that there are threeoptions ie pro-CA pro-EA and abstention Accordingto (14) the coalition 119878 consists of pro-CA voters and thecoalition 119879 consists of pro-EA voters If the outcome ofX(119878 119879) is 1 or -1 F119860119875119896 simply adopts the CA way or the EAway respectively in the bandwidth allocation process If theoutcome ofX(119878 119879) is 0F119860119875119896 adopts the PAway Based on thedecisions of individual devices Each F-AP adaptively decidesits communication resource allocation way

Like as the communication resource computation capac-ity in each F-AP is also a naturally limited resourceThereforethe computation power of F-AP is adaptively assigned to eachrequested application task Usually different applications canbe categorized into two classes class I (real-time data applica-tions) and class II (non-real-time data applications) accord-ing to the required QoE They not only require differentcomputation capacity but also have different QoE In orderto design the computation resource allocation algorithmeach F-AP needs to partition its own computation powerto support different type applications In this paper we alsoadopt the weighted ternary voting game model However incontrast to the voting game in our communication allocationalgorithm the developed computation allocation algorithmhas focused on the basic concept of pseudo-Banzhaf values

Consider another X game inF119860119875119896 with NF119860119875119896 Individual119898119894 isin NF119860119875

119896participates in the second voting procedure At

this time there are three options for each voter ie proclass Iservice proclass II service and abstention According to (14)the coalition 119878 consists of pro-class I service voters and thecoalition 119879 consists of pro-class II service voters Based on theoutcome of X(119878 119879) the F-AP estimates the preference ratio(120572119868) of class I application services as follows

120572119868= sum119898119894isin119878 (ΨQ

119898119894timesF119898119894 (119909))

(sum119898119894isin119878 (ΨQ119898119894timesF119898119894 (119909)) + sum119898119895isin119879 (ΨQ

119898119895timesF119898119894 (119909)))

(19)

According to each userrsquos pseudo-Banzhaf value and socialpower the 120572119868 value is decided and each F-AP partitions itstotal computation capacity (C) The division ratio for class I(or class II) type application services is 120572119868 (or 1 minus 120572119868) Andthen the partitioned computation capacity is distributed forits corresponding type application services while maximizing

8 Wireless Communications and Mobile Computing

Table 2 Application service type and requirements and system parameters

Type Application Bandwidthrequirement

Minimumrequirement (m)

Computationrequirement Duration (Avesec)

Class I Application I 119860=512 Mbps 119860=256 Mbps 98 MHz 180 secApplication II 119860=640 Mbps 119860=320 Mbps 256MHz 120 sec

Class II

Application III 119860=768 Mbps 119860=384 Mbps 128 MHz 240 secApplication IV 119860=896 Mbps 119860=458 Mbps 256MHz 300 secApplication V 119860=128 Gbps 119860=640 Mbps 384 MHz 360 secApplication VI 119860=152 Gbps 119860=860 Mbps 512 MHz 480 sec

Parameter Value DescriptionΓ 5 the maximum outcome of a user to normalize the social power120583 09 a parameter to controls the speed of social power saturation120572 075 a factor to decide the quota for the decision-taking120573 025 a factor to decide the quota for the decision- blocking120578119868 09 a control parameter to adjust the payoff of class I service120578119868119868 07 a control parameter to adjust the payoff of class II service120576 10 a control parameter for the sharing activity between social friends

the systemperformance To implement this process we adoptthe fundamental notion of the Nash Bargaining Solution(NBS) This is captured by the following optimization prob-lem

max[119860119898119894 ]

prod119898119894isin119878

119880119868119898119894 (119860119898119894) + max[119860119898119895 ]

prod119898119895isin119879

119880119868119868119898119895 (119860119898119895) (20)

st 119880119868119898119894 (119860119898119894) = 120578119868 times log2 (1 + 119860119898119894119877119898119894 )

119880119868119868119898119895 (119860119898119895) = 120578119868119868 times (1 minus exp(minus119860119898119894119877119898119894 ))sum119898119894isin119878

119860119898119894 le (120572119868 timesCF119860119875119896)

and sum119898119895isin119879

119860119898119895 le ((1 minus 120572119868) timesCF119860119875119896)

(21)

where 120578119868 (or 120578119868119868) is a control parameter to adjust the payoffof 119898119894 (or 119898119895) and CF119860119875

119896is the total computation capacity of

F119860119875119896

24 Main Steps of Proposed SFIoT Resource AllocationScheme Under the vision of the future internet the SIoTparadigm aims to connect anything while relying on the self-establishment of inter-thing social relationships In additionthe emerging paradigm of FC exhibits the right features forcoping with the aforementioned technological 5G networkissues In order to address this challenge the SFIoT paradigmhas been gaining momentum To effectively operate theSFIoT system the interactive relationship between F-APs andmobile devices is an important research topic and should beconsidered to design the resource allocation algorithms Inthis study we provide the communication and computationresource allocation scheme which is formulated according

to the BVG and NBS Owing to our two-phase game modelwe can get the most fair-efficient system performance bycombining both game approaches The main steps of theproposed scheme are described as follows

Step 1 Application types requirements and system controlparameters are determined by the simulation scenario (seesimulation assumptions in Section 3 and Table 2)

Step 2 Each individual F-AP monitors its covering area andresponds to its corresponding mobile devices Mobile devicesgenerate their application service requests in a distributedfashion and ask FC services to their corresponding F-AP

Step 3 Each mobile device has its own social relationshipthrough a social network close friends can probabilisticallyshare the outcome of FC services

Step 4 To reduce computation complexity individual F-APsexecute separately their two-phase games in parallel

Step 5 During the first game process each mobile devicetakes two votes one is to decide the communication resourceallocation way and the other is to partition the F-APrsquoscomputation capacity

Step 6 According to (14) the F-AP decides its communi-cation allocation way Based on our weighted ternary BVGmodel one of the PA CA and EA ways is adaptivelyselected And then the F-AP estimates the division ratio ofcomputation capacity using (19)

Step 7 During the second game process each F-AP inde-pendently distributes the partitioned computation capacity tocorresponding type applications According to (20) differentapplications can retain their FC services while maximizingthe system performance

Wireless Communications and Mobile Computing 9

Step 8 Based on the two-phase game model F-APs andmobile devices are hierarchically interconnected and inter-acting with one another to operate the SFIoT system

Step 9 Constantly each F-AP is self-monitoring its coveringarea When the game players are changed due to the mobiledevice migration the proposed scheme proceeds to Step 5 forthe next two-phase game procedure

3 Simulation Results and Discussion

In this section we evaluate the effectiveness of our proposedprotocol and analyze the simulation result while compar-ing it with that of the SDFC [14] SSEC [15] and CFIC[16] schemes To ensure a fair comparison the experi-mental setup consists of the following specifications andscenario

(i) Simulated SFIoT system consists of 25 F-APs and 100mobile devices F-APs are laid out in regular patternandmobile is randomly located in the F-APsrsquo coveringareas

(ii) For the FC services total computation capacity ofeach F-AP is 30 GHz and total communicationresource of each F-AP is 50 Gbps

(iii) Each mobile device individually generates servicerequests Service type of applications is randomlydecided and service request rate is Poisson process(120588) The offered rate range is varied from 0 to 30

(iv) The social friend number (|S119898| gt 0) of each mobiledevice is a random variable which has the maximumprobability at 3 while following a normal distribution

(v) Each social tie (0 le 119890119898119898 le 1) between mobiledevices is also a random variable which has themaximum probability at 05 while following a normaldistribution

(vi) The probability (P119904) that social friends119898119894 and119898119895 canshare the service outcome is defined asP119904 = 119890119898119894119898119895120576where 120576 is a control parameter for the sharing activity

(vii) We assume that there are no physical obstacles in theexperiments and each mobile device can contact itscorresponding F-AP for FC services

(viii) For calculation simplicity the communicationresource is specified in terms of basic units (BUs)where one BU is the minimum amount (eg 128Mbps in our system) of resource management

(ix) Using the iterative water-filling method the value ofNBS in (20) is obtained

(x) Network performancemeasures obtained on the basisof 100 simulation runs are plotted as functions of theoffered service request rate (120588)

(xi) Six different applications are assumed based on thedifferent computation and communication require-ment they are generated with equal probability

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

05 1 15 2 25 30Offered Application Load (Task Request Rate)

0

05

1

15

Qua

lity

of E

xper

ienc

e (Q

oE)

Figure 1 SFIoT quality of experience (QoE)

(xii) Performance criteria obtained through simulation areQoE resource usability and task failure probabilitythese simulation metrics are evaluated mainly todemonstrate the validity of our proposed method

To facilitate the development and implementation of oursimulator Table 2 lists application service type requirementsand system control parameters

Figure 1 shows the QoE of each scheme in differentapplication load scenarios For this evaluationQoE is definedas the normalized resource loss ratio than the originallyrequested amount From the viewpoint of mobile users QoEis a key factor to evaluate their service satisfactions In theproposed scheme the system resource is differently allocatedbased on the voting approach According to the desirablefeatures of ternary voting approach the limited resourcecan be assigned while effectively reflecting the preferenceinclination of users Our two-phase game approach can wellorchestrate the SFIoT system infrastructure to maximize theQoE of mobile users The result of Figure 1 proves that ourtwo-step game method is a socially beneficial way to achievea better QoE than other existing schemes

The obtained results of task failure probabilities aredepicted in Figure 2 This performance criterion is stronglyrelated to the SFIoT system throughput which is estimatedbased on the ratio of successfully completed applications Asexpected the application load increases the available systemresource is exhausted to support new application servicesTherefore task failure probabilities increase proportionatelywith the application request rate The resulting curves inFigure 2 allow us to see that our proposed scheme has gaineda lower task failure probability than other existing schemes Itis therefore worth to say that under different application loadconditions our proposed scheme can perform excellentlyto maintain a superior system throughput based on the

10 Wireless Communications and Mobile Computing

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Task

Fai

lure

Pro

babi

lity

05 1 15 2 25 30Offered Application Load (Task Request Rate)

Figure 2 SFIoT task failure probability

0 05 1 15 2 25 3Offered Application Load (Task Request Rate)

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Syste

m R

esou

rce U

tiliz

atio

n

Figure 3 Resource utilization

self-controlled management policies It is a highly desirableproperty for the SFIoT system operations

Figure 3 illustrates the resource utilization of F-APs Wemeasure this performance metric to show and determinewhether our game-based approach can provide the systemefficiency As it was explained control decisions in theproposed scheme are made in an interactive two-phasegame model More specifically the proposed approach inthis paper can adapt the current SFIoT system conditionwhile dynamically adjusting the control decisions Whenthe application load is low (below 05) the performance ofeach scheme is not much difference As the application load

rate increases the resource utilization also increases All theschemes have similar trends however due to our approachrsquosflexibility adaptability and responsiveness to current SFIoTsystem conditions the F-AP resource in our proposed schemecan be used more efficiently under widely different anddiversified application load situations

Through simulation experiments the obtained resultsin Figures 1ndash3 clearly demonstrate that the suitability andfeasibility of our proposed scheme while revealing enhancedperspectives regarding QoE throughput and resource uti-lization in the SFIoT system In particular the BVG and NBSprocesses work together to reach an agreement that givesmutual advantages for mobile devices and F-APs two gamemodels are sophisticatedly combined and dependent on eachotherTherefore they act cooperatively to satisfy the differentperformance requirements

4 Summary and Conclusions

The past decade has witnessed an explosive growth of IoTwhere a variety of smart mobile devices offer a plethora ofdifferent applications Under the vision of future networksFC and social network principles are integrated into the IoTparadigm in order to enforce the distributed self-cooperationand self-management of mobile devices In this paper wepresent a novel resource allocation scheme for the SFIoTsystem According to the BVG and NBS approaches ourproposed scheme is designed as two phases (i) mobiledevices take votes to decide the resource allocation methodsbased on the weighted ternary BVG model and (ii) the par-titioned computation capacity is fair-efficiently distributed toeach application by using the NBS These two game modelscontinue iteratively in a step-by-step manner in accordancewith the current SFIoT system informationThe experimentalsimulation results prove that our proposed protocol canefficiently improve theQoE system throughput and resourceutilization compared with the existing SDFC SSEC andCFIC schemes As directions for future research we aim atinvestigating the privacy and energy issues for the SFIoTsystemWe especially plan to add a blockchain-based securitylayer to the system In addition we also plan to develop anoncooperative gamemodelwith theoretical analysis andwillfocus on the addition of more features to increase the SFIoTsystem reliability in an unconstrained environment Lastbut not least machine learning techniques can be appliedto complement our proposed scheme It will be a potentialdirection and another possible extension to this work

Data Availability

The data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

The author declares that there are no conflicts of interestregarding the publication of this paper

Wireless Communications and Mobile Computing 11

Authorsrsquo Contributions

The sole author contributes to all research work

Acknowledgments

This research was supported by the MSIT (Ministry ofScience and ICT) Korea under the ITRC (InformationTechnology Research Center) support program (IITP-2018-2018-0-01799) supervised by the IITP (Institute for Infor-mation amp communications Technology Promotion) and byInstitute for Information amp communications TechnologyPromotion (IITP) grant funded by the Korea government(MSIT) (No2017-0-00498 A Development of Deidentifica-tion Technique based on Differential privacy)

References

[1] A M Ortiz D Hussein S Park S N Han and N Crespi ldquoThecluster between internet of things and social networks reviewand research challengesrdquo IEEE Internet of ings Journal vol 1no 3 pp 206ndash215 2014

[2] Z Li R Chen L Liu andGMin ldquoDynamic resource discoverybased on preference andmovement pattern similarity for large-scale social internet of thingsrdquo IEEE Internet of ings Journalvol 3 no 4 pp 581ndash589 2016

[3] H Yang and L Zhang ldquoResponse time prediction of IoT servicebased on time similarityrdquo Journal of Computing Science andEngineering vol 11 no 3 pp 100ndash108 2017

[4] E Baccarelli M Scarpiniti P G Naranjo and L Vaca-Cardenas ldquoFog of social IoT when the fog becomes socialrdquoIEEE Network vol 32 no 4 pp 68ndash80 2018

[5] H Shah-Mansouri and V W S Wong ldquoHierarchical fog-cloudcomputing for iot systems a computation offloading gamerdquoIEEE Internet ofings Journal vol 5 no 4 pp 3246ndash3257 2018

[6] K Lee and I Shin ldquoUser mobility model based computationoffloading decision for mobile cloudrdquo Journal of ComputingScience and Engineering vol 9 no 3 pp 155ndash162 2015

[7] Y Liu M J Lee and Y Zheng ldquoAdaptive multi-resourceallocation for cloudlet-based mobile cloud computing systemrdquoIEEE Transactions on Mobile Computing vol 15 no 10 pp2398ndash2410 2016

[8] K Kim S Uno and M Kim ldquoAdaptive QoS mechanism forwireless mobile networkrdquo Journal of Computing Science andEngineering vol 4 no 2 pp 153ndash172 2010

[9] A Pande V Ramamurthi and P Mohapatra ldquoQuality-orientedvideo delivery over LTErdquo Journal of Computing Science andEngineering vol 7 no 3 pp 168ndash176 2013

[10] S Kim and P K Varshney ldquoAn integrated adaptive bandwidth-management framework for QoS-sensitive multimedia cellularnetworksrdquo IEEE Transactions on Vehicular Technology vol 53no 3 pp 835ndash846 2004

[11] K SungwookGameeory Applications in Network Design IGIGlobal Hershey PA USA 2014

[12] J M Bilbao J R Fernandez N Jimenez and J J LopezldquoBiprobabilistic values for bicooperative gamesrdquo DiscreteApplied Mathematics vol 156 pp 2698ndash2711 2008

[13] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheshapley value for bicooperative gamesrdquo Annals of OperationsResearch vol 158 pp 99ndash115 2008

[14] L Liu Z Chang and X Guo ldquoSocially-aware dynamic compu-tation offloading scheme for fog computing system with energyharvesting devicesrdquo IEEE Internet of ings Journal vol 5 no3 pp 1869ndash1879 2018

[15] R Dautov S Distefano D Bruneo F Longo G Merlino andA Puliafito ldquoData processing in cyber-physical-social systemsthrough edge computingrdquo IEEE Access vol 6 pp 29822ndash298352018

[16] Y-C Chen Y-C Chang C-H Chen Y-S Lin J-L Chen andY-Y Chang ldquoCloud-fog computing for information-centricinternet-of-things applicationsrdquo in Proceedings of the 2017 IEEEInternational Conference on Applied System Innovation ICASI2017 pp 637ndash640 May 2017

[17] C Jiang L Kuang Z Han Y Ren and L Hanzo ldquoInformationcredibility modeling in cooperative networks equilibrium andmechanism designrdquo IEEE Journal on Selected Areas in Commu-nications vol 35 no 2 pp 432ndash448 2017

[18] S Kim ldquoFog radio access network system control scheme basedon the embedded game modelrdquo EURASIP Journal on WirelessCommunications and Networking vol 2017 no 113 pp 1ndash142017

[19] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheBanzhaf power index for ternary bicooperative gamesrdquoDiscreteApplied Mathematics vol 158 pp 967ndash980 2010

[20] M TsurumiM Inuiguchi and A Nishimura ldquoPseudo-banzhafvalues in bicooperative gamesrdquo in Proceedings of the IEEEInternational Conference on Systems Man and Cybernetics pp1138ndash1143 2005

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Active and Passive Electronic Components

VLSI Design

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Shock and Vibration

Hindawiwwwhindawicom Volume 2018

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

SensorsJournal of

Hindawiwwwhindawicom Volume 2018

International Journal of

RotatingMachinery

Hindawiwwwhindawicom Volume 2018

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Navigation and Observation

International Journal of

Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom

8 Wireless Communications and Mobile Computing

Table 2 Application service type and requirements and system parameters

Type Application Bandwidthrequirement

Minimumrequirement (m)

Computationrequirement Duration (Avesec)

Class I Application I 119860=512 Mbps 119860=256 Mbps 98 MHz 180 secApplication II 119860=640 Mbps 119860=320 Mbps 256MHz 120 sec

Class II

Application III 119860=768 Mbps 119860=384 Mbps 128 MHz 240 secApplication IV 119860=896 Mbps 119860=458 Mbps 256MHz 300 secApplication V 119860=128 Gbps 119860=640 Mbps 384 MHz 360 secApplication VI 119860=152 Gbps 119860=860 Mbps 512 MHz 480 sec

Parameter Value DescriptionΓ 5 the maximum outcome of a user to normalize the social power120583 09 a parameter to controls the speed of social power saturation120572 075 a factor to decide the quota for the decision-taking120573 025 a factor to decide the quota for the decision- blocking120578119868 09 a control parameter to adjust the payoff of class I service120578119868119868 07 a control parameter to adjust the payoff of class II service120576 10 a control parameter for the sharing activity between social friends

the systemperformance To implement this process we adoptthe fundamental notion of the Nash Bargaining Solution(NBS) This is captured by the following optimization prob-lem

max[119860119898119894 ]

prod119898119894isin119878

119880119868119898119894 (119860119898119894) + max[119860119898119895 ]

prod119898119895isin119879

119880119868119868119898119895 (119860119898119895) (20)

st 119880119868119898119894 (119860119898119894) = 120578119868 times log2 (1 + 119860119898119894119877119898119894 )

119880119868119868119898119895 (119860119898119895) = 120578119868119868 times (1 minus exp(minus119860119898119894119877119898119894 ))sum119898119894isin119878

119860119898119894 le (120572119868 timesCF119860119875119896)

and sum119898119895isin119879

119860119898119895 le ((1 minus 120572119868) timesCF119860119875119896)

(21)

where 120578119868 (or 120578119868119868) is a control parameter to adjust the payoffof 119898119894 (or 119898119895) and CF119860119875

119896is the total computation capacity of

F119860119875119896

24 Main Steps of Proposed SFIoT Resource AllocationScheme Under the vision of the future internet the SIoTparadigm aims to connect anything while relying on the self-establishment of inter-thing social relationships In additionthe emerging paradigm of FC exhibits the right features forcoping with the aforementioned technological 5G networkissues In order to address this challenge the SFIoT paradigmhas been gaining momentum To effectively operate theSFIoT system the interactive relationship between F-APs andmobile devices is an important research topic and should beconsidered to design the resource allocation algorithms Inthis study we provide the communication and computationresource allocation scheme which is formulated according

to the BVG and NBS Owing to our two-phase game modelwe can get the most fair-efficient system performance bycombining both game approaches The main steps of theproposed scheme are described as follows

Step 1 Application types requirements and system controlparameters are determined by the simulation scenario (seesimulation assumptions in Section 3 and Table 2)

Step 2 Each individual F-AP monitors its covering area andresponds to its corresponding mobile devices Mobile devicesgenerate their application service requests in a distributedfashion and ask FC services to their corresponding F-AP

Step 3 Each mobile device has its own social relationshipthrough a social network close friends can probabilisticallyshare the outcome of FC services

Step 4 To reduce computation complexity individual F-APsexecute separately their two-phase games in parallel

Step 5 During the first game process each mobile devicetakes two votes one is to decide the communication resourceallocation way and the other is to partition the F-APrsquoscomputation capacity

Step 6 According to (14) the F-AP decides its communi-cation allocation way Based on our weighted ternary BVGmodel one of the PA CA and EA ways is adaptivelyselected And then the F-AP estimates the division ratio ofcomputation capacity using (19)

Step 7 During the second game process each F-AP inde-pendently distributes the partitioned computation capacity tocorresponding type applications According to (20) differentapplications can retain their FC services while maximizingthe system performance

Wireless Communications and Mobile Computing 9

Step 8 Based on the two-phase game model F-APs andmobile devices are hierarchically interconnected and inter-acting with one another to operate the SFIoT system

Step 9 Constantly each F-AP is self-monitoring its coveringarea When the game players are changed due to the mobiledevice migration the proposed scheme proceeds to Step 5 forthe next two-phase game procedure

3 Simulation Results and Discussion

In this section we evaluate the effectiveness of our proposedprotocol and analyze the simulation result while compar-ing it with that of the SDFC [14] SSEC [15] and CFIC[16] schemes To ensure a fair comparison the experi-mental setup consists of the following specifications andscenario

(i) Simulated SFIoT system consists of 25 F-APs and 100mobile devices F-APs are laid out in regular patternandmobile is randomly located in the F-APsrsquo coveringareas

(ii) For the FC services total computation capacity ofeach F-AP is 30 GHz and total communicationresource of each F-AP is 50 Gbps

(iii) Each mobile device individually generates servicerequests Service type of applications is randomlydecided and service request rate is Poisson process(120588) The offered rate range is varied from 0 to 30

(iv) The social friend number (|S119898| gt 0) of each mobiledevice is a random variable which has the maximumprobability at 3 while following a normal distribution

(v) Each social tie (0 le 119890119898119898 le 1) between mobiledevices is also a random variable which has themaximum probability at 05 while following a normaldistribution

(vi) The probability (P119904) that social friends119898119894 and119898119895 canshare the service outcome is defined asP119904 = 119890119898119894119898119895120576where 120576 is a control parameter for the sharing activity

(vii) We assume that there are no physical obstacles in theexperiments and each mobile device can contact itscorresponding F-AP for FC services

(viii) For calculation simplicity the communicationresource is specified in terms of basic units (BUs)where one BU is the minimum amount (eg 128Mbps in our system) of resource management

(ix) Using the iterative water-filling method the value ofNBS in (20) is obtained

(x) Network performancemeasures obtained on the basisof 100 simulation runs are plotted as functions of theoffered service request rate (120588)

(xi) Six different applications are assumed based on thedifferent computation and communication require-ment they are generated with equal probability

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

05 1 15 2 25 30Offered Application Load (Task Request Rate)

0

05

1

15

Qua

lity

of E

xper

ienc

e (Q

oE)

Figure 1 SFIoT quality of experience (QoE)

(xii) Performance criteria obtained through simulation areQoE resource usability and task failure probabilitythese simulation metrics are evaluated mainly todemonstrate the validity of our proposed method

To facilitate the development and implementation of oursimulator Table 2 lists application service type requirementsand system control parameters

Figure 1 shows the QoE of each scheme in differentapplication load scenarios For this evaluationQoE is definedas the normalized resource loss ratio than the originallyrequested amount From the viewpoint of mobile users QoEis a key factor to evaluate their service satisfactions In theproposed scheme the system resource is differently allocatedbased on the voting approach According to the desirablefeatures of ternary voting approach the limited resourcecan be assigned while effectively reflecting the preferenceinclination of users Our two-phase game approach can wellorchestrate the SFIoT system infrastructure to maximize theQoE of mobile users The result of Figure 1 proves that ourtwo-step game method is a socially beneficial way to achievea better QoE than other existing schemes

The obtained results of task failure probabilities aredepicted in Figure 2 This performance criterion is stronglyrelated to the SFIoT system throughput which is estimatedbased on the ratio of successfully completed applications Asexpected the application load increases the available systemresource is exhausted to support new application servicesTherefore task failure probabilities increase proportionatelywith the application request rate The resulting curves inFigure 2 allow us to see that our proposed scheme has gaineda lower task failure probability than other existing schemes Itis therefore worth to say that under different application loadconditions our proposed scheme can perform excellentlyto maintain a superior system throughput based on the

10 Wireless Communications and Mobile Computing

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Task

Fai

lure

Pro

babi

lity

05 1 15 2 25 30Offered Application Load (Task Request Rate)

Figure 2 SFIoT task failure probability

0 05 1 15 2 25 3Offered Application Load (Task Request Rate)

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Syste

m R

esou

rce U

tiliz

atio

n

Figure 3 Resource utilization

self-controlled management policies It is a highly desirableproperty for the SFIoT system operations

Figure 3 illustrates the resource utilization of F-APs Wemeasure this performance metric to show and determinewhether our game-based approach can provide the systemefficiency As it was explained control decisions in theproposed scheme are made in an interactive two-phasegame model More specifically the proposed approach inthis paper can adapt the current SFIoT system conditionwhile dynamically adjusting the control decisions Whenthe application load is low (below 05) the performance ofeach scheme is not much difference As the application load

rate increases the resource utilization also increases All theschemes have similar trends however due to our approachrsquosflexibility adaptability and responsiveness to current SFIoTsystem conditions the F-AP resource in our proposed schemecan be used more efficiently under widely different anddiversified application load situations

Through simulation experiments the obtained resultsin Figures 1ndash3 clearly demonstrate that the suitability andfeasibility of our proposed scheme while revealing enhancedperspectives regarding QoE throughput and resource uti-lization in the SFIoT system In particular the BVG and NBSprocesses work together to reach an agreement that givesmutual advantages for mobile devices and F-APs two gamemodels are sophisticatedly combined and dependent on eachotherTherefore they act cooperatively to satisfy the differentperformance requirements

4 Summary and Conclusions

The past decade has witnessed an explosive growth of IoTwhere a variety of smart mobile devices offer a plethora ofdifferent applications Under the vision of future networksFC and social network principles are integrated into the IoTparadigm in order to enforce the distributed self-cooperationand self-management of mobile devices In this paper wepresent a novel resource allocation scheme for the SFIoTsystem According to the BVG and NBS approaches ourproposed scheme is designed as two phases (i) mobiledevices take votes to decide the resource allocation methodsbased on the weighted ternary BVG model and (ii) the par-titioned computation capacity is fair-efficiently distributed toeach application by using the NBS These two game modelscontinue iteratively in a step-by-step manner in accordancewith the current SFIoT system informationThe experimentalsimulation results prove that our proposed protocol canefficiently improve theQoE system throughput and resourceutilization compared with the existing SDFC SSEC andCFIC schemes As directions for future research we aim atinvestigating the privacy and energy issues for the SFIoTsystemWe especially plan to add a blockchain-based securitylayer to the system In addition we also plan to develop anoncooperative gamemodelwith theoretical analysis andwillfocus on the addition of more features to increase the SFIoTsystem reliability in an unconstrained environment Lastbut not least machine learning techniques can be appliedto complement our proposed scheme It will be a potentialdirection and another possible extension to this work

Data Availability

The data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

The author declares that there are no conflicts of interestregarding the publication of this paper

Wireless Communications and Mobile Computing 11

Authorsrsquo Contributions

The sole author contributes to all research work

Acknowledgments

This research was supported by the MSIT (Ministry ofScience and ICT) Korea under the ITRC (InformationTechnology Research Center) support program (IITP-2018-2018-0-01799) supervised by the IITP (Institute for Infor-mation amp communications Technology Promotion) and byInstitute for Information amp communications TechnologyPromotion (IITP) grant funded by the Korea government(MSIT) (No2017-0-00498 A Development of Deidentifica-tion Technique based on Differential privacy)

References

[1] A M Ortiz D Hussein S Park S N Han and N Crespi ldquoThecluster between internet of things and social networks reviewand research challengesrdquo IEEE Internet of ings Journal vol 1no 3 pp 206ndash215 2014

[2] Z Li R Chen L Liu andGMin ldquoDynamic resource discoverybased on preference andmovement pattern similarity for large-scale social internet of thingsrdquo IEEE Internet of ings Journalvol 3 no 4 pp 581ndash589 2016

[3] H Yang and L Zhang ldquoResponse time prediction of IoT servicebased on time similarityrdquo Journal of Computing Science andEngineering vol 11 no 3 pp 100ndash108 2017

[4] E Baccarelli M Scarpiniti P G Naranjo and L Vaca-Cardenas ldquoFog of social IoT when the fog becomes socialrdquoIEEE Network vol 32 no 4 pp 68ndash80 2018

[5] H Shah-Mansouri and V W S Wong ldquoHierarchical fog-cloudcomputing for iot systems a computation offloading gamerdquoIEEE Internet ofings Journal vol 5 no 4 pp 3246ndash3257 2018

[6] K Lee and I Shin ldquoUser mobility model based computationoffloading decision for mobile cloudrdquo Journal of ComputingScience and Engineering vol 9 no 3 pp 155ndash162 2015

[7] Y Liu M J Lee and Y Zheng ldquoAdaptive multi-resourceallocation for cloudlet-based mobile cloud computing systemrdquoIEEE Transactions on Mobile Computing vol 15 no 10 pp2398ndash2410 2016

[8] K Kim S Uno and M Kim ldquoAdaptive QoS mechanism forwireless mobile networkrdquo Journal of Computing Science andEngineering vol 4 no 2 pp 153ndash172 2010

[9] A Pande V Ramamurthi and P Mohapatra ldquoQuality-orientedvideo delivery over LTErdquo Journal of Computing Science andEngineering vol 7 no 3 pp 168ndash176 2013

[10] S Kim and P K Varshney ldquoAn integrated adaptive bandwidth-management framework for QoS-sensitive multimedia cellularnetworksrdquo IEEE Transactions on Vehicular Technology vol 53no 3 pp 835ndash846 2004

[11] K SungwookGameeory Applications in Network Design IGIGlobal Hershey PA USA 2014

[12] J M Bilbao J R Fernandez N Jimenez and J J LopezldquoBiprobabilistic values for bicooperative gamesrdquo DiscreteApplied Mathematics vol 156 pp 2698ndash2711 2008

[13] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheshapley value for bicooperative gamesrdquo Annals of OperationsResearch vol 158 pp 99ndash115 2008

[14] L Liu Z Chang and X Guo ldquoSocially-aware dynamic compu-tation offloading scheme for fog computing system with energyharvesting devicesrdquo IEEE Internet of ings Journal vol 5 no3 pp 1869ndash1879 2018

[15] R Dautov S Distefano D Bruneo F Longo G Merlino andA Puliafito ldquoData processing in cyber-physical-social systemsthrough edge computingrdquo IEEE Access vol 6 pp 29822ndash298352018

[16] Y-C Chen Y-C Chang C-H Chen Y-S Lin J-L Chen andY-Y Chang ldquoCloud-fog computing for information-centricinternet-of-things applicationsrdquo in Proceedings of the 2017 IEEEInternational Conference on Applied System Innovation ICASI2017 pp 637ndash640 May 2017

[17] C Jiang L Kuang Z Han Y Ren and L Hanzo ldquoInformationcredibility modeling in cooperative networks equilibrium andmechanism designrdquo IEEE Journal on Selected Areas in Commu-nications vol 35 no 2 pp 432ndash448 2017

[18] S Kim ldquoFog radio access network system control scheme basedon the embedded game modelrdquo EURASIP Journal on WirelessCommunications and Networking vol 2017 no 113 pp 1ndash142017

[19] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheBanzhaf power index for ternary bicooperative gamesrdquoDiscreteApplied Mathematics vol 158 pp 967ndash980 2010

[20] M TsurumiM Inuiguchi and A Nishimura ldquoPseudo-banzhafvalues in bicooperative gamesrdquo in Proceedings of the IEEEInternational Conference on Systems Man and Cybernetics pp1138ndash1143 2005

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Active and Passive Electronic Components

VLSI Design

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Shock and Vibration

Hindawiwwwhindawicom Volume 2018

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

SensorsJournal of

Hindawiwwwhindawicom Volume 2018

International Journal of

RotatingMachinery

Hindawiwwwhindawicom Volume 2018

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Navigation and Observation

International Journal of

Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom

Wireless Communications and Mobile Computing 9

Step 8 Based on the two-phase game model F-APs andmobile devices are hierarchically interconnected and inter-acting with one another to operate the SFIoT system

Step 9 Constantly each F-AP is self-monitoring its coveringarea When the game players are changed due to the mobiledevice migration the proposed scheme proceeds to Step 5 forthe next two-phase game procedure

3 Simulation Results and Discussion

In this section we evaluate the effectiveness of our proposedprotocol and analyze the simulation result while compar-ing it with that of the SDFC [14] SSEC [15] and CFIC[16] schemes To ensure a fair comparison the experi-mental setup consists of the following specifications andscenario

(i) Simulated SFIoT system consists of 25 F-APs and 100mobile devices F-APs are laid out in regular patternandmobile is randomly located in the F-APsrsquo coveringareas

(ii) For the FC services total computation capacity ofeach F-AP is 30 GHz and total communicationresource of each F-AP is 50 Gbps

(iii) Each mobile device individually generates servicerequests Service type of applications is randomlydecided and service request rate is Poisson process(120588) The offered rate range is varied from 0 to 30

(iv) The social friend number (|S119898| gt 0) of each mobiledevice is a random variable which has the maximumprobability at 3 while following a normal distribution

(v) Each social tie (0 le 119890119898119898 le 1) between mobiledevices is also a random variable which has themaximum probability at 05 while following a normaldistribution

(vi) The probability (P119904) that social friends119898119894 and119898119895 canshare the service outcome is defined asP119904 = 119890119898119894119898119895120576where 120576 is a control parameter for the sharing activity

(vii) We assume that there are no physical obstacles in theexperiments and each mobile device can contact itscorresponding F-AP for FC services

(viii) For calculation simplicity the communicationresource is specified in terms of basic units (BUs)where one BU is the minimum amount (eg 128Mbps in our system) of resource management

(ix) Using the iterative water-filling method the value ofNBS in (20) is obtained

(x) Network performancemeasures obtained on the basisof 100 simulation runs are plotted as functions of theoffered service request rate (120588)

(xi) Six different applications are assumed based on thedifferent computation and communication require-ment they are generated with equal probability

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

05 1 15 2 25 30Offered Application Load (Task Request Rate)

0

05

1

15

Qua

lity

of E

xper

ienc

e (Q

oE)

Figure 1 SFIoT quality of experience (QoE)

(xii) Performance criteria obtained through simulation areQoE resource usability and task failure probabilitythese simulation metrics are evaluated mainly todemonstrate the validity of our proposed method

To facilitate the development and implementation of oursimulator Table 2 lists application service type requirementsand system control parameters

Figure 1 shows the QoE of each scheme in differentapplication load scenarios For this evaluationQoE is definedas the normalized resource loss ratio than the originallyrequested amount From the viewpoint of mobile users QoEis a key factor to evaluate their service satisfactions In theproposed scheme the system resource is differently allocatedbased on the voting approach According to the desirablefeatures of ternary voting approach the limited resourcecan be assigned while effectively reflecting the preferenceinclination of users Our two-phase game approach can wellorchestrate the SFIoT system infrastructure to maximize theQoE of mobile users The result of Figure 1 proves that ourtwo-step game method is a socially beneficial way to achievea better QoE than other existing schemes

The obtained results of task failure probabilities aredepicted in Figure 2 This performance criterion is stronglyrelated to the SFIoT system throughput which is estimatedbased on the ratio of successfully completed applications Asexpected the application load increases the available systemresource is exhausted to support new application servicesTherefore task failure probabilities increase proportionatelywith the application request rate The resulting curves inFigure 2 allow us to see that our proposed scheme has gaineda lower task failure probability than other existing schemes Itis therefore worth to say that under different application loadconditions our proposed scheme can perform excellentlyto maintain a superior system throughput based on the

10 Wireless Communications and Mobile Computing

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Task

Fai

lure

Pro

babi

lity

05 1 15 2 25 30Offered Application Load (Task Request Rate)

Figure 2 SFIoT task failure probability

0 05 1 15 2 25 3Offered Application Load (Task Request Rate)

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Syste

m R

esou

rce U

tiliz

atio

n

Figure 3 Resource utilization

self-controlled management policies It is a highly desirableproperty for the SFIoT system operations

Figure 3 illustrates the resource utilization of F-APs Wemeasure this performance metric to show and determinewhether our game-based approach can provide the systemefficiency As it was explained control decisions in theproposed scheme are made in an interactive two-phasegame model More specifically the proposed approach inthis paper can adapt the current SFIoT system conditionwhile dynamically adjusting the control decisions Whenthe application load is low (below 05) the performance ofeach scheme is not much difference As the application load

rate increases the resource utilization also increases All theschemes have similar trends however due to our approachrsquosflexibility adaptability and responsiveness to current SFIoTsystem conditions the F-AP resource in our proposed schemecan be used more efficiently under widely different anddiversified application load situations

Through simulation experiments the obtained resultsin Figures 1ndash3 clearly demonstrate that the suitability andfeasibility of our proposed scheme while revealing enhancedperspectives regarding QoE throughput and resource uti-lization in the SFIoT system In particular the BVG and NBSprocesses work together to reach an agreement that givesmutual advantages for mobile devices and F-APs two gamemodels are sophisticatedly combined and dependent on eachotherTherefore they act cooperatively to satisfy the differentperformance requirements

4 Summary and Conclusions

The past decade has witnessed an explosive growth of IoTwhere a variety of smart mobile devices offer a plethora ofdifferent applications Under the vision of future networksFC and social network principles are integrated into the IoTparadigm in order to enforce the distributed self-cooperationand self-management of mobile devices In this paper wepresent a novel resource allocation scheme for the SFIoTsystem According to the BVG and NBS approaches ourproposed scheme is designed as two phases (i) mobiledevices take votes to decide the resource allocation methodsbased on the weighted ternary BVG model and (ii) the par-titioned computation capacity is fair-efficiently distributed toeach application by using the NBS These two game modelscontinue iteratively in a step-by-step manner in accordancewith the current SFIoT system informationThe experimentalsimulation results prove that our proposed protocol canefficiently improve theQoE system throughput and resourceutilization compared with the existing SDFC SSEC andCFIC schemes As directions for future research we aim atinvestigating the privacy and energy issues for the SFIoTsystemWe especially plan to add a blockchain-based securitylayer to the system In addition we also plan to develop anoncooperative gamemodelwith theoretical analysis andwillfocus on the addition of more features to increase the SFIoTsystem reliability in an unconstrained environment Lastbut not least machine learning techniques can be appliedto complement our proposed scheme It will be a potentialdirection and another possible extension to this work

Data Availability

The data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

The author declares that there are no conflicts of interestregarding the publication of this paper

Wireless Communications and Mobile Computing 11

Authorsrsquo Contributions

The sole author contributes to all research work

Acknowledgments

This research was supported by the MSIT (Ministry ofScience and ICT) Korea under the ITRC (InformationTechnology Research Center) support program (IITP-2018-2018-0-01799) supervised by the IITP (Institute for Infor-mation amp communications Technology Promotion) and byInstitute for Information amp communications TechnologyPromotion (IITP) grant funded by the Korea government(MSIT) (No2017-0-00498 A Development of Deidentifica-tion Technique based on Differential privacy)

References

[1] A M Ortiz D Hussein S Park S N Han and N Crespi ldquoThecluster between internet of things and social networks reviewand research challengesrdquo IEEE Internet of ings Journal vol 1no 3 pp 206ndash215 2014

[2] Z Li R Chen L Liu andGMin ldquoDynamic resource discoverybased on preference andmovement pattern similarity for large-scale social internet of thingsrdquo IEEE Internet of ings Journalvol 3 no 4 pp 581ndash589 2016

[3] H Yang and L Zhang ldquoResponse time prediction of IoT servicebased on time similarityrdquo Journal of Computing Science andEngineering vol 11 no 3 pp 100ndash108 2017

[4] E Baccarelli M Scarpiniti P G Naranjo and L Vaca-Cardenas ldquoFog of social IoT when the fog becomes socialrdquoIEEE Network vol 32 no 4 pp 68ndash80 2018

[5] H Shah-Mansouri and V W S Wong ldquoHierarchical fog-cloudcomputing for iot systems a computation offloading gamerdquoIEEE Internet ofings Journal vol 5 no 4 pp 3246ndash3257 2018

[6] K Lee and I Shin ldquoUser mobility model based computationoffloading decision for mobile cloudrdquo Journal of ComputingScience and Engineering vol 9 no 3 pp 155ndash162 2015

[7] Y Liu M J Lee and Y Zheng ldquoAdaptive multi-resourceallocation for cloudlet-based mobile cloud computing systemrdquoIEEE Transactions on Mobile Computing vol 15 no 10 pp2398ndash2410 2016

[8] K Kim S Uno and M Kim ldquoAdaptive QoS mechanism forwireless mobile networkrdquo Journal of Computing Science andEngineering vol 4 no 2 pp 153ndash172 2010

[9] A Pande V Ramamurthi and P Mohapatra ldquoQuality-orientedvideo delivery over LTErdquo Journal of Computing Science andEngineering vol 7 no 3 pp 168ndash176 2013

[10] S Kim and P K Varshney ldquoAn integrated adaptive bandwidth-management framework for QoS-sensitive multimedia cellularnetworksrdquo IEEE Transactions on Vehicular Technology vol 53no 3 pp 835ndash846 2004

[11] K SungwookGameeory Applications in Network Design IGIGlobal Hershey PA USA 2014

[12] J M Bilbao J R Fernandez N Jimenez and J J LopezldquoBiprobabilistic values for bicooperative gamesrdquo DiscreteApplied Mathematics vol 156 pp 2698ndash2711 2008

[13] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheshapley value for bicooperative gamesrdquo Annals of OperationsResearch vol 158 pp 99ndash115 2008

[14] L Liu Z Chang and X Guo ldquoSocially-aware dynamic compu-tation offloading scheme for fog computing system with energyharvesting devicesrdquo IEEE Internet of ings Journal vol 5 no3 pp 1869ndash1879 2018

[15] R Dautov S Distefano D Bruneo F Longo G Merlino andA Puliafito ldquoData processing in cyber-physical-social systemsthrough edge computingrdquo IEEE Access vol 6 pp 29822ndash298352018

[16] Y-C Chen Y-C Chang C-H Chen Y-S Lin J-L Chen andY-Y Chang ldquoCloud-fog computing for information-centricinternet-of-things applicationsrdquo in Proceedings of the 2017 IEEEInternational Conference on Applied System Innovation ICASI2017 pp 637ndash640 May 2017

[17] C Jiang L Kuang Z Han Y Ren and L Hanzo ldquoInformationcredibility modeling in cooperative networks equilibrium andmechanism designrdquo IEEE Journal on Selected Areas in Commu-nications vol 35 no 2 pp 432ndash448 2017

[18] S Kim ldquoFog radio access network system control scheme basedon the embedded game modelrdquo EURASIP Journal on WirelessCommunications and Networking vol 2017 no 113 pp 1ndash142017

[19] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheBanzhaf power index for ternary bicooperative gamesrdquoDiscreteApplied Mathematics vol 158 pp 967ndash980 2010

[20] M TsurumiM Inuiguchi and A Nishimura ldquoPseudo-banzhafvalues in bicooperative gamesrdquo in Proceedings of the IEEEInternational Conference on Systems Man and Cybernetics pp1138ndash1143 2005

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Active and Passive Electronic Components

VLSI Design

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Shock and Vibration

Hindawiwwwhindawicom Volume 2018

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

SensorsJournal of

Hindawiwwwhindawicom Volume 2018

International Journal of

RotatingMachinery

Hindawiwwwhindawicom Volume 2018

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Navigation and Observation

International Journal of

Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom

10 Wireless Communications and Mobile Computing

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Task

Fai

lure

Pro

babi

lity

05 1 15 2 25 30Offered Application Load (Task Request Rate)

Figure 2 SFIoT task failure probability

0 05 1 15 2 25 3Offered Application Load (Task Request Rate)

e Proposed Schemee SDFC schemee SSEC schemee CFIC scheme

0

01

02

03

04

05

06

07

08

09

1

Syste

m R

esou

rce U

tiliz

atio

n

Figure 3 Resource utilization

self-controlled management policies It is a highly desirableproperty for the SFIoT system operations

Figure 3 illustrates the resource utilization of F-APs Wemeasure this performance metric to show and determinewhether our game-based approach can provide the systemefficiency As it was explained control decisions in theproposed scheme are made in an interactive two-phasegame model More specifically the proposed approach inthis paper can adapt the current SFIoT system conditionwhile dynamically adjusting the control decisions Whenthe application load is low (below 05) the performance ofeach scheme is not much difference As the application load

rate increases the resource utilization also increases All theschemes have similar trends however due to our approachrsquosflexibility adaptability and responsiveness to current SFIoTsystem conditions the F-AP resource in our proposed schemecan be used more efficiently under widely different anddiversified application load situations

Through simulation experiments the obtained resultsin Figures 1ndash3 clearly demonstrate that the suitability andfeasibility of our proposed scheme while revealing enhancedperspectives regarding QoE throughput and resource uti-lization in the SFIoT system In particular the BVG and NBSprocesses work together to reach an agreement that givesmutual advantages for mobile devices and F-APs two gamemodels are sophisticatedly combined and dependent on eachotherTherefore they act cooperatively to satisfy the differentperformance requirements

4 Summary and Conclusions

The past decade has witnessed an explosive growth of IoTwhere a variety of smart mobile devices offer a plethora ofdifferent applications Under the vision of future networksFC and social network principles are integrated into the IoTparadigm in order to enforce the distributed self-cooperationand self-management of mobile devices In this paper wepresent a novel resource allocation scheme for the SFIoTsystem According to the BVG and NBS approaches ourproposed scheme is designed as two phases (i) mobiledevices take votes to decide the resource allocation methodsbased on the weighted ternary BVG model and (ii) the par-titioned computation capacity is fair-efficiently distributed toeach application by using the NBS These two game modelscontinue iteratively in a step-by-step manner in accordancewith the current SFIoT system informationThe experimentalsimulation results prove that our proposed protocol canefficiently improve theQoE system throughput and resourceutilization compared with the existing SDFC SSEC andCFIC schemes As directions for future research we aim atinvestigating the privacy and energy issues for the SFIoTsystemWe especially plan to add a blockchain-based securitylayer to the system In addition we also plan to develop anoncooperative gamemodelwith theoretical analysis andwillfocus on the addition of more features to increase the SFIoTsystem reliability in an unconstrained environment Lastbut not least machine learning techniques can be appliedto complement our proposed scheme It will be a potentialdirection and another possible extension to this work

Data Availability

The data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

The author declares that there are no conflicts of interestregarding the publication of this paper

Wireless Communications and Mobile Computing 11

Authorsrsquo Contributions

The sole author contributes to all research work

Acknowledgments

This research was supported by the MSIT (Ministry ofScience and ICT) Korea under the ITRC (InformationTechnology Research Center) support program (IITP-2018-2018-0-01799) supervised by the IITP (Institute for Infor-mation amp communications Technology Promotion) and byInstitute for Information amp communications TechnologyPromotion (IITP) grant funded by the Korea government(MSIT) (No2017-0-00498 A Development of Deidentifica-tion Technique based on Differential privacy)

References

[1] A M Ortiz D Hussein S Park S N Han and N Crespi ldquoThecluster between internet of things and social networks reviewand research challengesrdquo IEEE Internet of ings Journal vol 1no 3 pp 206ndash215 2014

[2] Z Li R Chen L Liu andGMin ldquoDynamic resource discoverybased on preference andmovement pattern similarity for large-scale social internet of thingsrdquo IEEE Internet of ings Journalvol 3 no 4 pp 581ndash589 2016

[3] H Yang and L Zhang ldquoResponse time prediction of IoT servicebased on time similarityrdquo Journal of Computing Science andEngineering vol 11 no 3 pp 100ndash108 2017

[4] E Baccarelli M Scarpiniti P G Naranjo and L Vaca-Cardenas ldquoFog of social IoT when the fog becomes socialrdquoIEEE Network vol 32 no 4 pp 68ndash80 2018

[5] H Shah-Mansouri and V W S Wong ldquoHierarchical fog-cloudcomputing for iot systems a computation offloading gamerdquoIEEE Internet ofings Journal vol 5 no 4 pp 3246ndash3257 2018

[6] K Lee and I Shin ldquoUser mobility model based computationoffloading decision for mobile cloudrdquo Journal of ComputingScience and Engineering vol 9 no 3 pp 155ndash162 2015

[7] Y Liu M J Lee and Y Zheng ldquoAdaptive multi-resourceallocation for cloudlet-based mobile cloud computing systemrdquoIEEE Transactions on Mobile Computing vol 15 no 10 pp2398ndash2410 2016

[8] K Kim S Uno and M Kim ldquoAdaptive QoS mechanism forwireless mobile networkrdquo Journal of Computing Science andEngineering vol 4 no 2 pp 153ndash172 2010

[9] A Pande V Ramamurthi and P Mohapatra ldquoQuality-orientedvideo delivery over LTErdquo Journal of Computing Science andEngineering vol 7 no 3 pp 168ndash176 2013

[10] S Kim and P K Varshney ldquoAn integrated adaptive bandwidth-management framework for QoS-sensitive multimedia cellularnetworksrdquo IEEE Transactions on Vehicular Technology vol 53no 3 pp 835ndash846 2004

[11] K SungwookGameeory Applications in Network Design IGIGlobal Hershey PA USA 2014

[12] J M Bilbao J R Fernandez N Jimenez and J J LopezldquoBiprobabilistic values for bicooperative gamesrdquo DiscreteApplied Mathematics vol 156 pp 2698ndash2711 2008

[13] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheshapley value for bicooperative gamesrdquo Annals of OperationsResearch vol 158 pp 99ndash115 2008

[14] L Liu Z Chang and X Guo ldquoSocially-aware dynamic compu-tation offloading scheme for fog computing system with energyharvesting devicesrdquo IEEE Internet of ings Journal vol 5 no3 pp 1869ndash1879 2018

[15] R Dautov S Distefano D Bruneo F Longo G Merlino andA Puliafito ldquoData processing in cyber-physical-social systemsthrough edge computingrdquo IEEE Access vol 6 pp 29822ndash298352018

[16] Y-C Chen Y-C Chang C-H Chen Y-S Lin J-L Chen andY-Y Chang ldquoCloud-fog computing for information-centricinternet-of-things applicationsrdquo in Proceedings of the 2017 IEEEInternational Conference on Applied System Innovation ICASI2017 pp 637ndash640 May 2017

[17] C Jiang L Kuang Z Han Y Ren and L Hanzo ldquoInformationcredibility modeling in cooperative networks equilibrium andmechanism designrdquo IEEE Journal on Selected Areas in Commu-nications vol 35 no 2 pp 432ndash448 2017

[18] S Kim ldquoFog radio access network system control scheme basedon the embedded game modelrdquo EURASIP Journal on WirelessCommunications and Networking vol 2017 no 113 pp 1ndash142017

[19] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheBanzhaf power index for ternary bicooperative gamesrdquoDiscreteApplied Mathematics vol 158 pp 967ndash980 2010

[20] M TsurumiM Inuiguchi and A Nishimura ldquoPseudo-banzhafvalues in bicooperative gamesrdquo in Proceedings of the IEEEInternational Conference on Systems Man and Cybernetics pp1138ndash1143 2005

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Active and Passive Electronic Components

VLSI Design

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Shock and Vibration

Hindawiwwwhindawicom Volume 2018

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

SensorsJournal of

Hindawiwwwhindawicom Volume 2018

International Journal of

RotatingMachinery

Hindawiwwwhindawicom Volume 2018

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Navigation and Observation

International Journal of

Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom

Wireless Communications and Mobile Computing 11

Authorsrsquo Contributions

The sole author contributes to all research work

Acknowledgments

This research was supported by the MSIT (Ministry ofScience and ICT) Korea under the ITRC (InformationTechnology Research Center) support program (IITP-2018-2018-0-01799) supervised by the IITP (Institute for Infor-mation amp communications Technology Promotion) and byInstitute for Information amp communications TechnologyPromotion (IITP) grant funded by the Korea government(MSIT) (No2017-0-00498 A Development of Deidentifica-tion Technique based on Differential privacy)

References

[1] A M Ortiz D Hussein S Park S N Han and N Crespi ldquoThecluster between internet of things and social networks reviewand research challengesrdquo IEEE Internet of ings Journal vol 1no 3 pp 206ndash215 2014

[2] Z Li R Chen L Liu andGMin ldquoDynamic resource discoverybased on preference andmovement pattern similarity for large-scale social internet of thingsrdquo IEEE Internet of ings Journalvol 3 no 4 pp 581ndash589 2016

[3] H Yang and L Zhang ldquoResponse time prediction of IoT servicebased on time similarityrdquo Journal of Computing Science andEngineering vol 11 no 3 pp 100ndash108 2017

[4] E Baccarelli M Scarpiniti P G Naranjo and L Vaca-Cardenas ldquoFog of social IoT when the fog becomes socialrdquoIEEE Network vol 32 no 4 pp 68ndash80 2018

[5] H Shah-Mansouri and V W S Wong ldquoHierarchical fog-cloudcomputing for iot systems a computation offloading gamerdquoIEEE Internet ofings Journal vol 5 no 4 pp 3246ndash3257 2018

[6] K Lee and I Shin ldquoUser mobility model based computationoffloading decision for mobile cloudrdquo Journal of ComputingScience and Engineering vol 9 no 3 pp 155ndash162 2015

[7] Y Liu M J Lee and Y Zheng ldquoAdaptive multi-resourceallocation for cloudlet-based mobile cloud computing systemrdquoIEEE Transactions on Mobile Computing vol 15 no 10 pp2398ndash2410 2016

[8] K Kim S Uno and M Kim ldquoAdaptive QoS mechanism forwireless mobile networkrdquo Journal of Computing Science andEngineering vol 4 no 2 pp 153ndash172 2010

[9] A Pande V Ramamurthi and P Mohapatra ldquoQuality-orientedvideo delivery over LTErdquo Journal of Computing Science andEngineering vol 7 no 3 pp 168ndash176 2013

[10] S Kim and P K Varshney ldquoAn integrated adaptive bandwidth-management framework for QoS-sensitive multimedia cellularnetworksrdquo IEEE Transactions on Vehicular Technology vol 53no 3 pp 835ndash846 2004

[11] K SungwookGameeory Applications in Network Design IGIGlobal Hershey PA USA 2014

[12] J M Bilbao J R Fernandez N Jimenez and J J LopezldquoBiprobabilistic values for bicooperative gamesrdquo DiscreteApplied Mathematics vol 156 pp 2698ndash2711 2008

[13] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheshapley value for bicooperative gamesrdquo Annals of OperationsResearch vol 158 pp 99ndash115 2008

[14] L Liu Z Chang and X Guo ldquoSocially-aware dynamic compu-tation offloading scheme for fog computing system with energyharvesting devicesrdquo IEEE Internet of ings Journal vol 5 no3 pp 1869ndash1879 2018

[15] R Dautov S Distefano D Bruneo F Longo G Merlino andA Puliafito ldquoData processing in cyber-physical-social systemsthrough edge computingrdquo IEEE Access vol 6 pp 29822ndash298352018

[16] Y-C Chen Y-C Chang C-H Chen Y-S Lin J-L Chen andY-Y Chang ldquoCloud-fog computing for information-centricinternet-of-things applicationsrdquo in Proceedings of the 2017 IEEEInternational Conference on Applied System Innovation ICASI2017 pp 637ndash640 May 2017

[17] C Jiang L Kuang Z Han Y Ren and L Hanzo ldquoInformationcredibility modeling in cooperative networks equilibrium andmechanism designrdquo IEEE Journal on Selected Areas in Commu-nications vol 35 no 2 pp 432ndash448 2017

[18] S Kim ldquoFog radio access network system control scheme basedon the embedded game modelrdquo EURASIP Journal on WirelessCommunications and Networking vol 2017 no 113 pp 1ndash142017

[19] J M Bilbao J R Fernandez N Jimenez and J J Lopez ldquoTheBanzhaf power index for ternary bicooperative gamesrdquoDiscreteApplied Mathematics vol 158 pp 967ndash980 2010

[20] M TsurumiM Inuiguchi and A Nishimura ldquoPseudo-banzhafvalues in bicooperative gamesrdquo in Proceedings of the IEEEInternational Conference on Systems Man and Cybernetics pp1138ndash1143 2005

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Active and Passive Electronic Components

VLSI Design

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Shock and Vibration

Hindawiwwwhindawicom Volume 2018

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

SensorsJournal of

Hindawiwwwhindawicom Volume 2018

International Journal of

RotatingMachinery

Hindawiwwwhindawicom Volume 2018

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Navigation and Observation

International Journal of

Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Active and Passive Electronic Components

VLSI Design

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Shock and Vibration

Hindawiwwwhindawicom Volume 2018

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

SensorsJournal of

Hindawiwwwhindawicom Volume 2018

International Journal of

RotatingMachinery

Hindawiwwwhindawicom Volume 2018

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Navigation and Observation

International Journal of

Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom