Research Article Location Prediction-Based Data ...downloads.hindawi.com/journals/mpe/2014/453564.pdf · Research Article Location Prediction-Based Data Dissemination Using Swarm

Research ArticleLocation Prediction-Based Data Dissemination Using SwarmIntelligence in Opportunistic Cognitive Networks

Jie Li12 Xingwei Wang3 Jie Jia3 Pengfei Wang4 Yan Zhou4 and Zhijie Zhao5

1 Computing Center Northeastern University Shenyang 110819 China2 Key Laboratory of Networked Control System The Chinese Academy of Sciences Shenyang 110016 China3 College of Information Science and Engineering Northeastern University Shenyang 110819 China4 Software College Northeastern University Shenyang 110819 China5 Information and Technology Center of China Mobile Group Liaoning Co Ltd Liaoning 110179 China

Correspondence should be addressed to Jie Li lijiemailneueducn

Received 8 June 2014 Revised 8 August 2014 Accepted 11 August 2014 Published 25 September 2014

Academic Editor Baozhen Yao

Copyright copy 2014 Jie Li et al This is an open access article distributed under the Creative Commons Attribution License whichpermits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Swarm intelligence is widely used in the application of communication networks In this paper we adopt a biologically inspiredstrategy to investigate the data dissemination problem in the opportunistic cognitive networks (OCNs) We model the system asa centralized and distributed hybrid system including a location prediction server and a pervasive environment deploying thelarge-scale human-centric devices To exploit such environment data gathering and dissemination are fundamentally based onthe contact opportunities To tackle the lack of contemporaneous end-to-end connectivity in opportunistic networks we applyant colony optimization as a cognitive heuristic technology to formulate a self-adaptive dissemination-based routing scheme inopportunistic cognitive networksThis routing strategy has attempted to find themost appropriate nodes conveyingmessages to thedestination node based on the location prediction information and intimacy between nodes which uses the online unsupervisedlearning on geographical locations and the biologically inspired algorithm on the relationship of nodes to estimate the deliveryprobability Extensive simulation is carried out on the real-world traces to evaluate the accuracy of the location prediction and theproposed scheme in terms of transmission cost delivery ratio average hops and delivery latency which achieves better routingperformances compared to the typical routing schemes in OCNs

1 Introduction

Cognitive networks [1] have already been prototyped formany commercial and civilian applications Combined withthe social intelligence cognitive networks promise to supportservices like citizen journalism mobile social networkingenvironmental monitoring and traffic monitoring by inte-grating ubiquitous sensing large-scale data collection andcloud computing Opportunistic network [2] provides anideal solution for promoting the evolution of communicationamong human beings and machines Cognitive networktechnology can be applied to the communication systemof opportunistic networks to provide heuristic schemes onthe algorithm design and system implementation Smarthandheld devices (eg smart phone or PDA) carried by

a large number of participants contribute to opportunisticcognitive networks with their sensing and communicationparts [3 4]

Moreover by including people in the loop it is nowpossible to design applications that can dramatically improvedaily lives of individuals and communities The inherentmobility of participants provides unprecedented spatiotem-poral coverage and also makes it possible to observe variousevents CarTel [5] is a mobile sensor computing systemdesigned to collect process deliver and visualize data fromsensors located on mobile units such as automobiles PEIR[6] is an application that uses location data sampled fromeveryday mobile phones to calculate personalized estimatesof environmental impact and exposure

Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2014 Article ID 453564 15 pageshttpdxdoiorg1011552014453564

2 Mathematical Problems in Engineering

The opportunistic cognitive network is a kind of delaytolerant networks (DTNs) and it utilizes communicationopportunities obtained from node movement to relay pack-ets Mobile nodes are enabled to communicate with eachother even if a route connecting them never exists The basicrouting strategy of opportunistic networks is ldquostore-carry-forwardrdquo Messages are routed between the sender and thedestination(s) and any possible node can be used as a relayprovided that it is likely to bring the message closer to thefinal destination Message transmission is mainly dependenton intermediate relays so relay node selection is criticalfor efficient data dissemination in opportunistic cognitivenetworks

Since human mobility is mostly unpredictable con-ventional routing algorithms for mobile ad hoc networks(MANETs) no longer perform well in the opportunisticcognitive networks Therefore new algorithms are requiredto overcome intermittent connectivity in the opportunisticnetworks

Stochastic routing protocols such as Epidemic [7] FirstContact (FC) [8] and Direct Delivery (DD) [9] broadcastmessages to any encountered node in order to increase thedelivery ratio Epidemic routing diffuses messages similarto the way viruses or bacteria spread in biology Wheneverencountering another node a node replicates and transfersmessages After receivingmessages a node will move to otherplaces and continuously replicate and deliver the messagesto other encountered ones First Contact sends messages tothe first encountered node without copying the messages toother nodes Direct Delivery comprises the trade-off betweenEpidemic and First Contact by finding an optimal number ofmessage copies

Creating more copies of a message increases the messagedelivery but decreases the network lifetime These stochasticrouting approaches usually consider the destinations of mes-sages as nodes rather than locations

Different from stochastic routing current contact-basedrouting chooses the most appropriate nodes relaying mes-sages to the destinations based on historical contact infor-mation such as contact times contact duration and contactcycle

Probabilistic Routing Protocol using History of Encoun-ters and Transitivity (PRoPHET) [10] is a context-based rout-ing protocol based on the history of encounters PRoPHETestimates the delivery predictability for each known desti-nation at each node before passing a message The estima-tion is based on the history of encounters between nodesSimBet [11] uses historical contacts to calculate two metricssimilarity and betweenness The similarity is calculated byhow frequently a node and its destination have met Thebetweenness is calculated by how many nodes which a nodehas met However if the utility metrics are equal SimBetwill prevent its forwarding behavior To improve this flawBUBBLE [12] adds the knowledge of community structure toensure message dissemination In addition the betweennessmay be useless if the message is near its destination Sim-BetAge [13] an improved version of SimBet was proposedto address these shortcomings Spray and Wait [14] thatuses only a handful of copies per message can achieve

comparable delays to an oracle-based optimal scheme thatminimizes delay while using the lowest possible number oftransmissions

The work in [15] does routing by simply calculatingthe delivery probability for a node to be at a location inMobySpace which is a high-dimensional Euclidean spacebased on the preknown mobility model However therequired assumption of that each node has the knowledgeabout mobility patterns of other nodes in the network makesthis work unpractical in realistic scenarios

For the concerns of optimality robustness and flexibilitysome routing protocols are developed by inspiration frombiology [16ndash21] Such routing schemes are also well appliedin the transportation domain [22ndash26] which are basedon prediction mechanism to improve the performance oftransportation systems

BeeHive is developed with inspiration of the foragingprinciples of honeybees [27] Communication between realbees is modeled by designing intelligent bee agents whichare able to make routing decisions in large and complextopologies The simulation results conclude that bee agentsoccupy smaller bandwidth and require less processor timeArtificial Bee Colony algorithm [17] is developed with scan-ning strategy for periodic vehicle routing problemThe workin [28] presents a biologically inspired discrete-event mod-eling approach for simulating alternative computer networkprotocols Adaptation and probabilistic specifications areintroduced into honeybee (BEE) and Routing InformationProtocol (RIP) routing algorithms

The ant colony optimization (ACO) methods have beeninspired by operating principles of ants [29] which empowera colony of ants to perform complex tasks such as nestbuilding and foraging [30]

Schoonderwoerd [31] first developed ACO-basedapproach for routing in telecommunication networks Thebasic principle in the approach is about the use of stigmergyin multiagents interaction Randomized ants traverse thenetwork nodes probabilistically and select the highestprobability path The approach is shown to be sufficientlyadaptive and demonstrates robustness across difficultnetwork conditions Yao et al [19 20] improved ACO fordelivery routing problem and PROMETShop schedulingproblems

The aforementioned routing algorithms show good per-formance on message delivery However geographic coordi-nates of nodes have little or no correlation with their contacttimes so the algorithms do not perform well when eachdevice frequently appears at different regions as most peopledaily do

In this paper we propose the location-prediction andswarm-intelligence-based data dissemination (LOPSI) algo-rithm for opportunistic cognitive networks The LOPSI algo-rithm is a probabilistic routing protocol combining locationprediction and the ant colony optimization It firstly predictspossible locations of relay nodes and destination(s) in succes-sive time seriesThemobile nodes calculate intimacy (contactfrequency) with potential relay nodes using ACO and thenmake a forwarding decision based on node intimacy andprobabilities of node mobility

Mathematical Problems in Engineering 3

predictionserver

Location prediction packetData packet

AP1

AP2

AP3

Nsource

Ndestination

Nk

Nk

Nj

Nj

Trajectory-

Figure 1 Network model of data dissemination in opportunisticcognitive networks

The remainder of this paper is organized as follows Thesystem model is illustrated in Section 2 Section 3 specifiesthe data dissemination algorithm Extensive simulations havebeen done for performance evaluation in Section 4 Section 5concludes the paper

2 Network Model

As mentioned in Section 1 the location-prediction andswarm-intelligence-based data dissemination (LOPSI) algo-rithm is a probabilistic routing protocol using location andintimacy information of potential relay nodes and destinationnodes The network is a mixture of an opportunistic networkand a centralized infrastructure as shown in Figure 1 Thecentralized infrastructure consists of a number of wirelessaccess points (APs) and a backbone connecting the APs

Mobile nodes (carrying smart devices) can only accessto the network when they are walking into the transmissionrange of any AP Each AP periodically uploads connectionrecords of mobile nodes to the location prediction server(LPS) which will eventually maintain a mobility database ofall nodes

The data exchanged in this scenario is assumed tobe of high magnitude and data transmission can onlyoccurs between peer counterparts as in normal opportunisticnetworks The LPS is merely responsible for storing nodemobility records and predicting node location upon receivingquery from mobile nodes via one of the APs The accuracyof the localization prediction algorithm can increase thedelivery ratio of the proposed approach

In the proposed system there exist two major entitiesthe mobile nodes and the LPS The mobile nodes work in adecentralizedmannerwhile the LPS is a powerful central unitwhich collects the trajectories of mobile users and performs

complex computation tasks to provide location predictionservice to the request mobile users

The mobile nodes only store the encountering infor-mation of the contact nodes in the local buffer Based onthe encountering information each mobile node obtains theintimacy between the contact nodes and the destination nodeand then refines the forwarding probability to the destinationnode The mobile nodes can forward the data message tothe next hop by the optimal wireless channel accordingto its own system state The mobile nodes can solve thedata dissemination locally without the need of coordinationwith a central server or the other clients Thus the datadissemination process is decentralized

The LPS is applied to track the movement trajectories ofall mobile nodes thus it needs to collect global movementinformation predicting the encounter opportunities Specifi-cally it uses the long-term trajectory information to constructthe Markov chain of a mobile user and to determine theprobable mobility trajectory based on partial encounteringhistory The computations in the server side and in the clientside are independent and their optimization results will notaffect each other Therefore we have designed three datadissemination algorithms to adapt different infrastructure ofthe network environment If the mobile nodes can connect tothe LPS it can obtain the location prediction information tooptimize the accuracy of forwarding node selection whereasthe mobile nodes can use the local information that isintimacy to estimate the delivery probability of the contactnodes

The APs are deemed as living and working locationsin the system Nodes can communicate with each otherthrough short-range communication media for instanceZigBee Bluetooth NFC or WiFi Direct Each node migratesfrom one location to another according to its own mobilitymodel When encountering other nodes the mobile nodedynamically calculates intimacy with them using ACO Theintimacy truly reflects historical contact information Itaccumulates upon each contact and on the other hand decaysover time

Once a mobile node would like to send data to a destina-tion node it firstly consults a nearest AP for its neighborsrsquotrend of movement The LPS performs location predictionusingMarkov process inference and returns back to the querynode an ordered list indicating probabilities of the neighbornodes meeting the destination node in successive time series(usually more than one time slot)

The mobile node will compute forwarding probabilityof its neighbors by considering their intimacy with thedestination node and then make a forwarding decision

The mixture network model enhances traditional oppor-tunistic networks with the centralized infrastructure whichtakes good advantage of existing AP assets but not burdenscurrent network The data forwarding is a probabilisticscheme guided by location prediction rather than stochasticor trivially probabilistic

By this means data will be delivered to the destinationwith higher probability and hence the network efficiency isimproved


3 Algorithm Design

In this section we formulate a data dissemination problemin the network environment lacking contemporaneous end-to-end connectivity To tackle the problem we propose acontact-based probability routing algorithm LOPSI whichimplements the data dissemination by calculating the for-warding probability based on location prediction schemeand a swarm intelligence heuristic methodTherefore LOPSIis fundamentally based on two routing schemes that areLocation Prediction-Based Data Dissemination (LOPDAD)and Ant Colony Optimization- (ACO-) Based Data Dis-semination (ACODAD) LOPDAD uses location predictioninformation to calculate the maximum probability of thelocation where the forwarding nodes and the destinationnode encounter which is suitable for the opportunistic envi-ronment deploying a centralized infrastructure ACODADuses the swarm intelligence mechanism ACO to select theforwarding nodes according to the intimacy between theforwarding nodes and the destination node which is suitablefor the fully distributed data dissemination in opportunisticnetworks LOPSI combines the merits of the two afore-mentioned algorithms Depending on the requirements ofapplications researchers can select the suitable algorithm toapply

31 Location Prediction-Based Data Dissemination (LOP-DAD) Studies on human mobility patterns have shown thatpeople daily activities exist in a high degree of repeatabilityPeople usually visit several fixed places regularly in each dayand do activities in a relatively fixed period According tomobile trajectories and regular behavior pattern of mobilenodes it can be used tomodel the scene based on the locationof mobile nodes and use relevant algorithms to predict theprobability of the node arriving at a certain position and toestimate the location of the mobile node

Markov chain algorithm is currently themost widely usedin location prediction algorithm with high accuracy Herewe use the second-order Markov chain model to predictthe location of the mobile node which has higher accuracythan the first-order Markov chain model according to thesimulation

It can use Markov model to describe the applicationscenario such as Campus where it is assumed that there are119898 locations Location 119894 is the 119894th status119883

119894of Markov process

and the state space is 119864 = 1198831 1198832 119883

119898 Thus scene

mobility model is defined as 119883 119879 and 119879 is time seriesFor each application scenarioMarkov chainmodel can be

used to predict the future location state of each mobile nodeSpecific modeling and forecasting process is as follows

311 Preparation Process Preparation before prediction pro-cess includes the following steps

(1) Determination of State Set According to the collection ofmobile nodes trajectories from the system server the locationelements in the collecting data are counted which is denotedas set 119871 As set 119871 contains a number of location elements

the locations of higher visiting frequency are chosen as statespace of the system denoted as set 119864 119864 sub 119871

(2) Discretization of Data Set Statistical data of all usersrelated to state set 119864 is made Then the dataset of each useris processed to be discrete set of the fixed time period so theset after discretization is denoted as follows

(119905119896 119883119894) 119896 = 1 2 3 119894 isin 1 2 3 119898 (1)

312 Location Prediction AlgorithmBased onO2MM Order-1 Markov chain model (O1MM) uses the state transitionmatrix and the initial distribution to predict which is simpleand intuitive [16] However as the indeterminacy of the statetransition probabilities is unscientific division of the initialstate of the system the prediction result of this method tendsto produce larger errors

Unlike the first order a Markov chain of higher order isa Markov model with memory that is a Markov chain thatdepends on not only the current state but also on 119899 minus 1 statesbefore where 119899 is the order and 119899 is finite [17] The chainis dependent on where it is right now and also where it wasin the last occasion Order-2 Markov chain model (O2MM)is used to improve the accuracy of prediction methodCompared with the prediction based on order-1 Markovchain order-2 Markov chain model can be more completeand rational use of information and effectively integratedwithcorrelation analysis so as to improve prediction accuracyO2MM depends on the current state and also the just visitedstate

The finite state space of O2MM is

119864 = 1198831 1198832 119883

119903119883119894 119883119895 119883

119898 119894 = 1 2 119898

119903 = 1 2 119898 119895 = 1 2 119898

(2)

and if the conditional probability is

119875 119883 (119905119899) = 119883

119895

= 119875 119883 (119905119899) = 119883

119895| 119883 (119905119899minus1) = 119883

119894 119883 (119905119899minus2) = 119883

119903

(3)

the transition probability of the node located at 119883119895at time

slice 119905119899under the condition that the node is located at 119883

119894at

time slice 119905119899minus1

and119883119903at time slice 119905

119899minus2is

119875119903119894119895119883 (119905119899) = 119883

119895

=

119898

sum

119894=1119903=1

119875 119883 (119905119899) = 119883

119895| 119883 (119905119899minus1) = 119883

119894 119883 (119905119899minus2) = 119883

119903

(4)

Formula (4) approximately equals the frequency that thenode visits the location 119883

119895when the state space tends to

infinity [18]

119901119903119894119895=

119903=119898119894=119898

sum

119903=1119894=1

119888119903119894119895

sum119898

119896=1119888119903119894119896

(5)


(1) input State Space Set 119864 = 119883119894 119894 isin 1 2 3 119898 Nodes Set119873 = 119873

119895 119895 isin 1 2 3 119899 the initial

probability distribution is 119875 (119899 minus 1 119899 minus 2) = 119901119903119894 119903 119894 isin 1 2 3 119898

(2) Discretization of data set Statistical data of all users related to state set E is made Then the data set of eachuser is processed to be discrete set of the fixed time slice so the set after discretization is denoted asfollow (119905

119896 119883119894) 119896 isin 119873+ 119894 isin 1 2 3 119898

(3) calculate the probability of the node to visit location119883119895according to (5) where the location state of the node

at current time slice and also the just visited state is respectively 119883119894and119883

119903

(4) Calculate one step transition probability matrix according to (6)(5) Calculate the probability of each state at time slice 119905

119899

(a) 119875 (119899) = 119875 (119899 minus 1 119899 minus 2) 119875(6) the location state at time slice 119905

119899is

(b)119883119895= argmax 119875(119899)

119895

(7) return 119883119895

Algorithm 1 119871 Markov(119873119888 119905119899) location prediction based on O2MM

where 119888119903119894119895

is the number of times that the observation nodevisits location 119883

119895by records statistics sum119898

119896=1119888119903119894119896

is the totalnumber of times that the node visits all the locations in 119864and then the probability of the node to visit location 119883

119895and

the location state of the node at current time slice and also thejust visited state is respectively119883

119894and119883

119903

If there are 119898 location states in state space set one-steptransition probability matrix is a119898 times 1198982 matrix

119875 =

119898⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞

[

[

[

[

[

[

[

[

[

[

[

[

119901111

119901112

sdot sdot sdot 11990111(119898minus1)

11990111119898

119901211

119901212


11990121119898

d

11990111989811

11990111989812


1199011198981119898

11990111989821

11990111989822


1199011198982119898

d

1199011198981198981

1199011198981198982


119901119898119898119898

]

]

]

]

]

]

]

]

]

]

]

]

1198982

(6)

The description of the location prediction algorithmbased onO2MM is as shown in Algorithm 1

According to Algorithm 1 given the initial probabilitydistribution of the node at time slice 119899 minus 1 119899 minus 2 we canrecurrence the probability distribution at time slice 119905

119899by

O2MM The state set 119864 = 1198831 1198832 119898 = 2 the initial state is

(119905119899minus2 1198831) (119905119899minus1 1198832) and the initial probability distribution

is

119875 (119899 minus 1 119899 minus 2)

= (119901(119899minus1119899minus2)

11 119901(119899minus1119899minus2)

12 119901(119899minus1119899minus2)

21 119901(119899minus1119899minus2)

22) = (0 1 0 0)

(7)

The transition matrix for Markov chain of order two is

119875 =

[

[

[

[

119901111

119901112

119901121

119901122

119901211

119901212

119901221

119901222

]

]

]

]

(8)

Here the first two numbers of the index representing thecurrent state and the last number represent the next stateAnd the probability distribution at time slice 119899 is

119875 (119899) = 119875 (119899 minus 1 119899 minus 2) 119875 = 119901(119899)

1 119901(119899)

2 (9)

The location state is obtained by the location predictionserver at time slice 119899 is

119883119895= argmax 119901(119899)

1 119901(119899)

2 (10)

In our system the location prediction server gathers thelocation data of the mobile nodes by the APs The serverexecutes Algorithm 1 to predict the trajectory of each nodeaccording to the discrete time slice

313 Transmission Probability of LOPDAD The data dis-semination mechanism can use the result of the locationprediction algorithm using O2MM At a certain time slicethe data forwarding probability equals the probability of theforwarding node visiting the location where a destinationnode is which is given by

119901119889

119871= 119901(119899)

119883119889

119895

(11)

where119883119889119895is the location state of destination node119873

119889at time

slice 119905119899 and 119901(119899)

119883119889

119895

is the probability that node 119873119888visits the

location119883119889119895at time slice 119905

119899

Algorithm 2 describes the process of the location-baseddata dissemination algorithm which obtained the forward-ing probability 119901119889

119871of119873119888to the destination node119873

119889

314 The Execution of Location Prediction The LPS predictsthe locations where the destination node and the forwardingnodes will encounter at the future time slices The thresholdof time slices is 119897 At time slice 119905

119899 when the LPS receives the

service request information REQ(119862119873119891119873119889)which includes

the destination node119873119889and the encounter nodes set 119862119873

119891

from the data carrier node 119873119904 the LPS calculates the


Input State Space Set 119864 = 119883119894 119894 isin 1 2 3 119898119873

119889119873119888 the initial probability

distribution is 119875 (119899 minus 1 119899 minus 2) = 119901119903119894 119903 119894 isin 1 2 3 119898

Output 119901119889

119871

(1) 119883119889

119895= 119871 Markov(119873

119889 119905119899)

(2ndash5) Algorithm 1 steps 1ndash5(6) calculation of forwarding probability according to (11)(7) return 119901119889

119871

Algorithm 2 119875 Markov(119873119888 119905119899) data dissemination probability based on location prediction

(1) 119873119904rarr LPS REQ(119862119873

119891119873119889) 119894 = 0 119895 = 0 119896 = 0 119897 = 3

(2) for all 119899 isin [1 119897] do(3) 119883(119873

119889 119905119899) = 119871 Markov(119873

119889 119905119899)

(4) for all 119873119891isin 119862119873

119891 do

(5) 119883(119873119891 119905119899) = 119871 Markov(119873

119891 119905119899) calculate the location state of119873

119891at 119905119899

(6) if (119883(119873119891 119905119899) == 119883

119894(119873119889 119905119899)) then 119873

119891and119873

119889encounter at 119905

119899

(7) 119865119873[119895++] = 119873119891 119873119891is stored as the forwarding node in set 119865119873

(8) 119865119875[119894++] = 119875 Markov(119873119891 119905119899) The probability of119873

119891meets119873

119889is stored in set 119865119875

(9) 119865119879[119896++] = 119899 the encounter time slice of119873119891

(10) end if(11) end for(12) end for(13) LPS rarr 119873

119904 SEI(119865(119873

119891 119875 119879) 119883(119873

119889 119905119899))

Algorithm 3 LoP Service(119862119873119891119873119889) the location prediction service in the LPS

location state set 119883 = 119883(119873119889 119905119897) of 119873

119889in the time slice

119905119899+1 119905119899+2 119905

119899+119897

119883 (119873119889 119905119897) = (119905

1 119883119894) (1199052 119883119895) (119905

119897 119883119901) sube (119905

119896 119883119894)

119896 isin 119873+ 119894 119895 119901 isin 1 2 3 119898

(12)

And the LPS also calculates the forwarding probability FP of each node 119873

119891in the set 119862119873

119891 The forwarding nodes

set FN in which the node 119873119891will visit the location where

the destination node 119873119889locates at the same time slice and

the encounter time slice is recorded in FT Finally the serversends the service information SEI(119865(119873

119891 119875 119879) 119883(119873

119889 119905119899))

to the node 119873119904 The execution process is described in

Algorithm 3

315 The Selection Mechanism of the Forwarding Nodes SetWhen the data carrier119873

119904receives the SEI from the predictive

server considering the cache management and the load ofthe network the number of copies of the forwarding data is afixed value COPY which is decided by the average buffer sizeof each node and the current load of the network If119873

119904finds

that the location of119873119889is the same as119873

119904during the threshold

time slice it only transmits to the nodes which can encounter119873119889at earlier time slice than 119873

119904 Otherwise if the number

of nodes in the forwarding nodes set is less than COPY 119873119904

transmits the data to the nodes in the forwarding nodes setand delete the data in its own buffer If the number of nodes

in the forwarding nodes set is more than COPY119873119904transmits

the data only to the nodes having maximum probability nomore than COPY according to the following

119875119889

119871119904119891=

119875 Markov (119873119891 119905119899)

119899

119873119891isin 119865 119905

119899isin 119879 119899 isin [1 119897]

(13)

where 119875 Markov(119873119891 119905119899) is the forwarding probability which

is equal to the probability that 119873119891encounters 119873

119889at time

slice 119905119899 where 119899 is the number of time slice intervals when

119873119891meets 119873

119889 The larger the number of time slice intervals

is the lower the probability of forwarding node is And theforwarding nodes set 1198651015840 which is selected by119873

119904

1198651015840sube 119865 119865

1015840119873119891 = argmax119891leCOPY

119875119889

119871119904119891 (14)

32 Swarm Intelligence Heuristic Data Dissemination (ACO-DAD) The inspiring source of ACO is the pheromonetrail laying and following behavior of real ants which usepheromones as a communication medium Artificial antsused in ACO are stochastic solution construction proceduresthat probabilistically build a solution by iteratively addingsolution components to partial solutions by taking intoaccount (i) heuristic information on the problem instancebeing solved if available and (ii) (artificial) pheromonetrails which change dynamically at run-time to reflect theagentsrsquo acquired search experience [19] In our algorithm


ACO is improved to be applied in data dissemination inopportunistic cognitive networks

321 Ant Colony Optimization- (ACO-) Based Data Dissem-ination in OCN (ACODAD) The pheromone in ACODADis the intimacy between two nodes The more frequency andcontinuous the contact between two nodes is the higher thevalue of intimacy is It means that the higher the pheromoneis The data carrier node tends to choose the node which hashigh intimacy value with the destination nodes to forwardthe data It means that the data forwarding probability of thenode having high intimacy with the destination node is highThe comparison of the characteristics betweenACODADandACO is shown in Table 1

322 Intimacy between Two Nodes in OCN Each node inopportunistic cognitive networks maintains a relationshiptable with other nodes using the value of intimacy

For example at time slice 119905119904 the data carrier node

119873119904arrives at location 119883

119894and senses all the other nodes

within the communication range via the communicationchannels (such as ZigBee Bluetooth NFC and other short-range communication protocols) Those nodes are added inencounter nodes set 119862119873

119891 of 119873

119904and are recorded in the

relationship table with the calculation results of intimacyThe intimacy between two nodes 119873

119894and 119873

119895depends

on the frequency of two nodes in connection based oncontact times 119899

119894119895 lasting time of one connection Δ119863

119888

119894119895

and the encounter intervals between two contacts Δ119868119888119894119895 The

mathematical description of intimacy is as follows where119877119894119895(119905) is the intimacy of119873

119894and119873

119895at time slice 119905

119877119894119895 (119905)

=

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

forallΔ119868119888

119894119895lt 119896

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

times (1 minus 120588)119890

existΔ119868119888

119894119895gt 119896

0 Δ119868119888

119894119895gt 119879 over the system running time

(15)

where120588 is the evaporation rate 119896 is the threshold of encountertime intervals of two nodes and 119890 is the times when the timeintervals is more than 119896

Although the intimacy is defined in consideration ofencounter time intervals if the two nodes are not in con-nection for a long time (more than a certain time thresholdvalue 119896) it enables the intimacy evaporation mechanism toensure the most frequently contact nodes with a high degreeof intimacy If the encounter time intervals are beyond thesystem running time 119879 the value of intimacy is zero

When two nodes encounter they respectively computethe intimacy based on records of encounter time 119905

119904and the

departure time 119905119890at the contact times 119888 Each node records

the time according to its own time clock When 119888 = 1 Δ119868119888119894119895

equals the encounter time of the first contact The algorithmof intimacy is as shown in Algorithm 4

Algorithm 4 captures the essence of (15) The intimacybetween two nodes is used to compute the forwardingprobability in ACODAD

The intimacy updates during the time duration Δ119905 thevariation of the intimacy is given by

119877119894119895 (119905 + Δ119905) = 119877119894119895 (

119905) + Δ119877119894119895 (119905)

Δ119877119894119895 (119905)

=

(119899119894119895+ Δ119899119894119895) times sum

(119899119894119895+Δ119899119894119895)

119888=1Δ119863119888

119894119895

sum

(119899119894119895+Δ119899119894119895)

119888=1Δ119868119888

119894119895

minus

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

Δ119899119894119895gt 0 forallΔ119868

119888

119894119895lt 119896

minus120588 times

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

Δ119899119894119895= 0 Δ119905 gt 119896 forallΔ119868

119888

119894119895lt 119896

0 otherwise(16)

323 Transmission Probability of ACODAD When the datacarrier node 119873

119904encounters the nodes in set 119862119873

119891 and

chooses the forwarding nodes from 119862119873119891 to the destination

119873119889based on the intimacy between the node in 119862119873

119891 and

119873119889 the forwarding probability is defined as the following

equation

119901119889

119877119904119891=

1 119895 = 119889

0 Intimacy (119873119891 119873119889) lt Intimacy (119873

119894 119873119889)

[119877119891119889 (119905)]

120572

sdot [120578119891119889 (119905)]

120573

sum119873119891isinallowd119889 119877

120572

119895119889sdot 120578120573

119895119889

119873119891isin allowd

119889

(17)

where allowd119889is given by

allowd119905119889= 119865 119873

119891| (119877119891119889 (119905) gt 119877119894119889 (

119905))

allowd119905119889isin 119865 119873

119891 allowd

119889= allowd119905

119889minus tabu

119889

(18)

where tabu119889is the node set including those nodes which have

already carried the transmission data copy and also the nodehad the copy before Such nodes will not be selected as theforwarding nodes

The parameters 120572 and120573 control the relative importance ofthe pheromone versus the heuristic information 120578

119891119889 which

is given by

120578119891119889=

1

119871119891119889

(19)

where 119871119891119889

is the time slice intervals that 119873119891and 119873

119889will

encounter which is an estimation value given by the recordsin the contact vector

A heuristic value 120578 respectively represents a prioriinformation about the problem instance definition or run-time information provided by a source different from the ants


Table 1 Comparison of characteristics between ACODAD and ACO

Characteristics ACODAD ACOTransmission process ofdata

Data dissemination based on the encounter of twonodes

Artificial ants move from one location to theneighbor one

Transition probability The probability that the forwarding nodeencounters the destination node

The state transition probabilities that are from onelocation to the next

Pheromone Intimacy between two nodes Pheromone trail laying by ants

Path length The interval of time slices that two nodesencounter The distance between two locations

Pheromone evaporation No contact between the two nodes duringthreshold time As time goes by the pheromone evaporates

Input 119873119894119873119895

Output 119877119894119895(119879)

(1) define 119879 the system running time(2) 119871 encounter time interval threshold(3) Contact vector array 997888997888rarr119872

119894119895(119888 119905119904 119905119890)

(4) initialization 997888997888rarr119872119894119895[0] = (0 0 0) 119877

119894119895(0) = 0 119896 = 119871 119891119897119886119892 = 1

(5) forall 119905 = 0 119879(6) if 119873

119894recieve HELLO from119873

119895

(7) 119899119894119895= ++119888

(8) for all 119888 = 1 119899119894119895

(9) Δ119863119888

119894119895[119888] = 119905

119890minus 119905119904

(10) 119863119888

119894119895+= Δ119863119888

119894119895

(11) Δ119868119888

119894119895[119888] = 119905

119904minus 119905

(12) 119868119888

119894119895+= Δ119868119888

119894119895

(13) 119905 = 119905119890

(14) if (Δ119868119888119894119895[119888] gt 119896)

(15) 119890++(16) 119891119897119886119892 = 0(17) end if(18) end for(19) end if(20) end for(21) if (119891119897119886119892 == 0)(22) return 119877

119894119895(119879) = ((119899

119894119895times 119863119888

119894119895) 119868119888

119894119895) times (1 minus 120588)

119890

(23) else(24) return 119877

119894119895(119879) = (119899

119894119895times 119863119888

119894119895) 119868119888

119894119895

(25) end if

Algorithm 4 Intimacy(119873119894119873119895) computes the intimacy value between two nodes at encounter time T

(1) Initialization node set 119873 = 119873119894 119894 = 1 2 119896 119896 isin 119873

+ state space 119864 = 119883

119895 119895 = 1 2 119898119898 isin 119873

+ time

slice set 119879 = 119905119899 119899 = 0 1 2 119897 119897 isin 119873

+Maximum data copies COPY

(2) for 119899 = 0 119897 the system running time the function executes the one hop data dissemination to the forwarding node set

(3) LOPSI Sec(119873119904)

(4) for forall119873119891isin 1198651015840119873119891

(5) execution of LOPSI Sec(119873119891)

(6) end for(7) end for

Algorithm 5 LOPSI(119873119904 119873119889)


In many cases 120578 is the cost or an estimate of the cost ofextending the current stateThese values are used by the antsrsquoheuristic rule tomake probabilistic decisions on how tomoveon the graph [32]

In our algorithm the heuristic value is defined as thedistance between 119873

119891and 119873

119889 As the location state of nodes

is discretized by the time slice And the delay of dataforwarding between the two nodes is not decided by theabsolute path length or the distance between the two nodesbut by calculating the number of time slice intervals betweenforwarding node and the destination node to meet with eachother

The data carrier node 119873119904chooses the forwarding nodes

from set 119862119873119891 according to the forwarding probability

based on intimacy Considering the cache management ouralgorithms set the maximum copy quantity of a unique datamessage If the data forwarding operation is executed119873

119904will

transmit the data to the nodes in set 1198621015840119873119891 which is given

by (20) and the quantity of nodes denoted by 119891 is no morethan COPY Finally119873

119904will delete the data stored in its buffer

1198621015840sube 119862 119873

119891 119862

1015840119873119891 = argmax119891leCOPY

119875119889

119877119904119891 (20)

33 Location Prediction Based Data Dissemination UsingSwarm Intelligence (LOPSI) The data dissemination algo-rithm LOPSI considers not only the intimacy between theforwarding node and the destination node but also thelocation where the two nodes may encounter Accordingto the location prediction algorithm the set of forwardingnodes 119865119873

119891 which visit the location where the destination

nodewill be during the threshold time slices can be obtainedAnd the data carrier node 119873

119904only sends message to nodes

in 119865119873119891 and compares the intimacy between 119873

119889and 119873

119891isin

119865119873119891 Combined with the prediction results from LOPDAD

and ACODAD the transmission probability from node119873119904to

the forwarding node119873119891is calculated by the weight formula

119901119889

119904119891= 120574 lowast 119901

119889

119877119904119891+ 120575 lowast 119901

119889

119871119904119891 120574 + 120575 = 1 (21)

where 119901119889119877119904119891

is the forwarding probability obtained fromACODAD 119901119889

119871119904119891is the forwarding probability obtained from

LOPDAD and parameters 120574 and 120575 are the weights of the twoprobabilities

In our campus environment the mobile node can com-municate to the location prediction server anywhere andanytime The prediction server provides only location pre-diction service and no other services Algorithm 5 describesthe whole execution process of LOPSI LOSI Sec is partof Algorithm 5 which is a function that the data messagetransfers from the current node to the forwarding node setWhen the source node119873

119904wants to send the data message to

the destination node 119873119889 119873119904will choose the potential nodes

by location prediction schemes and then obtain the intimacybetween the potential nodes and the destination node Thusthe forwarding node set will be determined The executionprocess of LOPSI is described in Algorithm 5

Figure 2 The simulation based on a realistic campus scenario

Table 2 The accuracy of location prediction algorithms based onO1MM and O2MM

O1MM O2MMPrediction accuracy 05610 08030Time complexity 119874(119873) 119874(119873

2)

Storage space 119874(1198732) 119874(119873

3)

For any node 119873119894carrying data message LOPSI Sec(119873

119894)

will determine the forwarding node set The description ofLOPSI Sec(119873

119894) is given as shown in Algorithm 6

Algorithm 5 describes the essence of LOPSI Each datahas a TTL which indicates how long the data can live inthe network It is set by the provider at the time of datageneration In Algorithm 5 the length of time slices 119897 equalsTTL The data carried by each node within TTL durationcannot be forwarded and then be automatically discarded

4 Performance Evaluation

In this section we present the simulation results to demon-strate the performance of proposed data dissemination algo-rithms Note that the recent work studying the nature ofhuman mobility has proved that suitable movement modelscan sufficiently present the behavior of human mobility [33]The model of mobility pattern deployed in our platform isSPMBMmodel [34] which is amobilitymodel that integratestemporal and spatial relationships and selects the shortestpath for the node randomly walking in the map area

41 Simulation Settings Thedata sets to evaluate the locationprediction algorithm based on O2MM are obtained fromwireless topology discovery (WTD) [35] which are employedin our simulation The accuracy of the location predictionalgorithms based on O1MM and O2MM can be obtainedfrom our previous work [36] which is given in Table 2

The simulation is based on a realistic campus scenarioshown in Figure 2 There are 40 locations and each oneinstalled a WiFi access point which can cover the campus


(1) 119873119894senses the contact nodes set 119862119873

119891

(2) for forall119873119891isin 119862119873

119891

(3) update intimacy(119873119894 119873119891)

(4) end forlowast the source node deliver the data directly to the destination node if the destination node is in the contactnodes set 119862119873

119891 and update the node set tabu

119889

lowast

(5) if 119873119889isin 119862119873

119891

(6) 119873119894transmits data to119873

119889

(7) tabu119889larr 119873

119894 119873119889

(8) break the Algorithm 5(9) else(10) 119873

119904sends REQ(119862119873

119891119873119889) to the server

(11) the server executes LoP Service(119862119873119891119873119889)

(12) 119873119904receives SEI

(13) 119873119904sends119873

119889to the nodes in 119865119873

119891

(14) for forall119873119891isin 119865119873

119891

(15) Send intimacy(119873119891 119873119889) to119873

119904

(16) end for(17) 119873

119904calculates 119901119889

119904119891according to (21)

(18) if (119891 le COPY) 119891 is the number of forwarding nodes(19) 119865

1015840119873119891 = 119865 119873

119891

(20) else(21) 119865

1015840119873119891 = argmax

119891leCOPY (119901119889

119904119891)

(22) end if(23) 119873

119894sends data to 1198651015840119873

119891

(24) tabu119889larr 119873

119894 1198651015840119873119891

(25) 119873119894delete the data copy in its buffer

(26) end if

Algorithm 6 LOPSI Sec(119873119894) the function executes one-hop data dissemination to the forwarding node set

Table 3 Simulation parameters

Parameter Value

Scene features

Simulation time 12 hField area 4500m lowast 3400mScene NEU CampusAPs 40

Node features

Mobility model SPMBMMovement speed for cars 27ndash139msMovement speed for pedestrians 05ndash15msTransmission rate 250KBsMaximum transmission range 10mTransmission mode BroadcastCache size 10MB1G

Message featuresPacket size 500KBndash1MB at randomFrequency of creating packets From 25 s to 35 s at randomNumber of copies 8TTL 5 hours

area The mobile nodes can be cars and pedestrians withsmart phones The location prediction server can communi-cate with mobile nodes via WiFi and only provide locationprediction service and no other services The only way toexchange and obtain data is through the contact of two nodesIf the communication range increases and is out of the range

of APs the mobile devices can exchange data by the ACO-DAD without using location prediction scheme by differentwireless communication techniques including WiFi DirectBluetooth and ZigBee In order to evaluate the performanceof the proposed data dissemination algorithms we conducta series of experiments under the parameters in Table 3 The


First In First Out is applied on buffer management In orderto avoid the heavy traffic load and cache load each data isset a TTL and the maximum copies of each data is a fixedconstant in the system

With the above settings the three data disseminationalgorithms proposed in our work LOPDAD ACODADand LOPSI are evaluated and compared with well-knownopportunistic routing protocols PRoPHET and Spray andWait

42 EvaluationMetrics Fourmetrics are used to evaluate theperformance requirements of the aforementioned data dis-semination algorithms average hops delivery ratio averagelatency and transmission cost

AverageHopHThis hop-countmetric is to assess the deliverycost in time and in cache N denotes the total number offorwarding nodes of every transmission of data includingboth successful and failure delivery Y is the total number ofcreated unique data messages H is given by

119867 =

119873

119884

(22)

Delivery Ratio R This metric is to evaluate the effectivenessand utility of the algorithm S is the total number of success-fully delivered unique data messages R is given by

119877 =

119878

119884

(23)

Average Latency L [35] The average latency of a uniquemessage is calculated by the following equation

119871 =

1

119884

119884

sum

119894=1

(119879119889119894minus 119879119904119894) (24)

where 119879119904119894is the moment that a unique data message 119894 is

originated and 119879119889119894

is the time when the first replicate ofunique message 119894 arrives at the destination The replicateis a copy of a unique message The number of replicatesdepends on the methodology of the routing algorithm singleor multiple copies [35]

Transmission Cost C It is a key metric to characterize theresource consumption and evaluate the data disseminationalgorithms in OCNs The total number of copies of uniquedata messages to deliver in the system denoted by Q isdivided by the number of copies of successfully deliveredmessages denoted by 119875

119862 =

119876

119875

(25)

43 Influence on Performance with the Variation ofTotal Number of Nodes

Average Hops As shown in Figure 3(a) with the increasingof numbers of nodes due to the hop limitation of Spray andWait the average hops are relatively small PRoPHET has norestrictions in this respect the frequency of nodes encoun-tering increases resulting in an increase in the average hopsOur algorithms are proposed to consider the managementof cache space and limit the maximum number of copiesof the nodes in the network which constrains the hops ofdata messages transmission LOPDAD and LOPSI especiallyset a time threshold on executing location prediction whichguaranteed the data message delivery to the destination nodewithin the time threshold The location state is discrete bythe time slices so threshold of time slices corresponds tothe number of hops Therefore the average hops of ouralgorithms are relatively small

Delivery Ratio As shown in Figure 3(b) whether for Sprayand Wait PRoPHET and our data distribution algorithmsthe delivery ratio significantly increases with the increaseof the nodes LOPDAD ACODAD LOPSI and Spray andWait have constrained the number of copies of the datamessages transmitted in the network Even if the numberof nodes and the amount of data messages increases thestorage space and the network overhead maintain a goodstatus avoiding data transmission failure by the heavy loadof cache and network resources depletionTherefore deliveryratio is better than the PRoPHET transmission LOPDADselects forwarding nodes which are most likely to completethe task based on location prediction ACODAD selects thebest forwarding nodes by high intimacy LOPSI tends tomakemore ldquoassertiverdquo options to select the forwarding nodes basedon the two aforementioned factors so the delivery ratio hasbeen significantly improved

Average Latency As shown in Figure 3(c) the average latencyis reduced with the increase of the nodes indicating thatour data distribution algorithms are assertive to select theforwarding nodes which are more likely to contact with thedestination node The data transfer of LOPDAD occurs atthe location that can connect with AP points not at anyother encounter places so the average delay is longer thanthat of ACODADand LOPSIThe forwarding nodes selectionmechanism of LOPSI is better than that of ACODAD whichhas less transmission operation but high delivery ratio so theaverage latency is lower than that of ACODAD

Transmission Cost Figure 3(d) shows the transmission costof our proposed algorithm and some existing algorithmsLOPSI has the lowest transmission cost since it only transfersmessages to the nodes with the highest forwarding proba-bility to the destination and the quantity of copies of themessage is a constant which equals the hops estimated by thelocation prediction algorithm With the increase of numberof nodes the opportunity of forwarding messages to thepotential nodes increases which leads to increase of the


PRoPHETSpray and WaitACODAD

LOPDADLOPSI

100 150 200 250 300

The number of nodes

50

45

40

35

30

25

20

Aver

age h

ops

(a)

02

03

04

05

06

07

Deli

very

ratio

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI

PRoPHETSpray and Wait

(b)

2500

3000

3500

4000

4500

5000

5500

6000

Aver

age l

aten

cy

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(c)

30

35

40

45

50

55

60

Tran

smiss

ion

cost

The number of nodes100 150 200 250 300

ACODADLOPDADLOPSI

Spray and Wait

(d)

Figure 3 (a) Average hops variation with the increase of the number of nodes (b) Delivery ratio variations with the increase of the numberof nodes (c) Average latency variations with the increase of the number of nodes (d) Transmission cost variations with the increase of thenumber of nodes

delivery ratio and also the average hops decrease (as shownin Figure 3(a)) PRoPHET has the highest transmission costsince it has no consideration in the cache management Thetransmission cost is at the value of more than 1100 muchmore than the transmission cost of the other four algorithmsThus the curve of transmission cost of PRoPHET cannot bedrawn in the scale of the graphThe transmission costs of theother three algorithms are lower than that of PRoPHET sincethey constrained the quantity of copies but not better thanLOPSI because the selection schemes of potential forwardingnodes are no better than that of LOPSI

44 Influence on Performance with the Variation of TTL Inthis scenario in order to illustrate the influence on perfor-mance by the variation of TTL the simulation parameter ofthe number of nodes is set to be 200

Average Hops As shown in Figure 4(a) with the increaseof TTL the data messages live long in the network whichwill increase the load of cache and the network HoweverLOPDAD and LOPSI have little change in the average hopssince the most data messages are successfully delivered to


55

50

45

40

35

30

25

20

Aver

age h

ops

4 5 6 7 8 9 10 11 12 13TTL (h)

ACODAD

LOPDADLOPSI


(a)

4 5 6 7 8 9 10 11 12 13TTL (h)

07

06

05

04

03

02

Del

iver

y ra

tio

ACODAD

LOPDADLOPSI


(b)

20003000400050006000700080009000

10000110001200013000

Aver

age l

aten

cy (s

)

4 5 6 7 8 9 10 11 12 13

ACODAD

LOPDADLOPSI

TTL (h)PRoPHETSpray and Wait

(c)

25

30

35

40

45

50

55

60

Tran

smiss

ion

cost

4 6 8 10 12 14TTL (h)

ACODADLOPDADLOPSI

Spray and Wait

(d)

Figure 4 (a) Average hops variations with the increase of TTL (b) Delivery ratio variations with the increase of TTL (c) Average latencyvariations with the increase of TTL (d) Transmission cost variations with the increase of TTL

the destination in the time threshold according to theaccuracy of O2MM being 80 During the TTL if the datadissemination is failure in the first time threshold periodthen start the second time threshold to deliver the datamessage the average hops may be doubled however theprobability of that condition is less than 20 The averagehops of ACODAD are more than LOPDAD and LOPIS as ithas no consideration of location where the destination nodewill be only by random encounter of two nodes With theincrease of the network load the average hops will increase

Delivery Ratio As shown in Figure 4(b) with the increase ofTTL the delivery ratio of LOPDAD changes little According

to the characteristic of Markov chain prediction the futurestatus of long-time prediction tends to be stabilized whichmeans that the prediction accuracy is reduced If the timethreshold is set too big then delivery ratio will decreaseThe delivery ratio of ACODAD and LOPIS increase since thecopy of data message in the network is not changed but theopportunistic of encounter is increased

Average Latency As shown in Figure 4(c) with the increase ofTTL the average latency of those five algorithmswill increasesince the network load is heavy and the total number of datamessages increases which makes the buffer overcrowdedSince the algorithms except PRoPHET constrain the copiesof messages to avoid the traffic loads the latency is lower


than that of PRoPHET which transfers the message to anypotential nodes without copies constraint

Transmission Cost As shown in Figure 4(d) with the increaseof TTL the transmission cost of LOPSI and LOPDADdecreases since the delivery ratio is higher (as shown inFigure 4(b)) and the number of copies of a unique messageis slightly changed It results in that LOPSI and LOPDADestimate the optimal path and then determine the number ofcopies of a unique message Spray and Wait and ACODADconstrain the number of copies so the variation of transmis-sion cost is slight As regards PRoPHET since it does notconstrain the copies of a unique message the transmissioncost is much higher than the transmission cost of the otherfour algorithms and at a more than thousand value Thus thecurve of transmission cost of PRoPHET cannot be drawn inthe scale of the graph

5 Conclusion and Future Work

In this paper we consider the efficient data disseminationmechanism in opportunistic cognitive networksWe proposea swarm intelligence heuristic data dissemination algorithmbased on location prediction The algorithm can select theefficient forwarding nodes with the maximum probability toencounter the destination node at the location where theymost likely to encounter and a high value of intimacy withthe destination node Moreover the algorithm considers thecache management and has good performances in the trans-mission cost and delivery ratio and simultaneously decreasesthe average hops and delivery delay The algorithm has goodscalability which can consider credibility incentives energycontrol and buffer management mechanism in the futurework Furthermore the swarm intelligence scheme can alsobe used in group construction of the mobile nodes whichcan promote the application performance in opportunisticcognitive networks

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgment

The authors would like to thank all the editors of thispaper They read the manuscript very carefully and providedvaluable feedbacks which are helpful to improve the qualityof the draft

References

[1] Q H Mahmoud Cognitive Networks Towards Self-Aware Net-works John Wiley amp Sons London UK 2007

[2] L Pelusi A Passarella and M Conti ldquoOpportunistic network-ing data forwarding in disconnected mobile ad hoc networksrdquoIEEE Communications Magazine vol 44 no 11 pp 134ndash1412006

[3] R Yu P Wang and Z Zhao ldquoNDI node-dependence-baseddynamic gaming Incentive algorithm in opportunistic net-worksrdquo in Proceedings of the 23rd International Conference onComputer Communications andNetworks ( ICCCN rsquo14) pp 581ndash588 Shanghai China 2014

[4] R Yu R Liu X Wang and J Cao ldquoImproving data qualitywith an accumulated reputation model in participatory sensingsystemsrdquo Sensors vol 3 pp 5573ndash5594 2014

[5] B Hull V Bychkovsky K Chen et al ldquoCarTel a distributedmobile sensor computing systemrdquo in Proceedings of the 4thACM International Conference on Embedded Networked SensorSystems pp 125ndash138 November 2006

[6] M Mun S Reddy K Shilton et al ldquoPEIR the personalenvironmental impact report as a platform for participatorysensing systems researchrdquo in Proceedings of the 7th ACMInternational Conference on Mobile Systems Applications andServices (MobiSys rsquo09) pp 55ndash68 June 2009

[7] A Vahdat and D Becker ldquoEpidemic routing for partially con-nected ad hoc networksrdquo Tech Rep Department of ComputerScience Duke Univeristy Durham NC USA 2000

[8] S Jain K Fall and R Patra ldquoRouting in a delay tolerantnetworkrdquo in Proceeing of the Conference on Computer Commu-nications (ACM SIGCOMM rsquo04) pp 145ndash158 New York NYUSA September 2004

[9] T Spyropoulos K Psounis and C S Raghavendra ldquoSingle-copy routing in intermittently connected mobile networksrdquo inProceedings of the 1st Annual IEEECommunications SocietyCon-ference on Sensor and Ad Hoc Communications and Networks(SECON rsquo04) pp 235ndash244 October 2004

[10] A Lindgren and A Droia ldquoProbabilistic routing protocolfor intermittently connected networksrdquo Internet Draft draft-lindgren-dtnrg-prophet-02 Work in Progress 2006

[11] E M Daly and M Haahr ldquoSocial network analysis for routingin disconnected delay-tolerantmanetsrdquo inProceedings of the 8thACM International Symposium on Mobile Ad Hoc Networkingand Computing (MobiHoc rsquo07) pp 32ndash40 ACM New York NYUSA 2007

[12] P Hui J Crowcroft and E Yoneki ldquoBUBBLE rap social-basedforwarding in delay tolerant networksrdquo in Proceedings of the 9thACM International Symposium on Mobile Ad Hoc Networkingand Computing (MobiHoc rsquo08) pp 241ndash250 May 2008

[13] J A B Link N Viol A Goliath and K Wehrle ldquoSimBe-tAge utilizing temporal changes in social networks for pocketswitched networksrdquo in Proceedings of the 1st ACM Workshopon User-Provided Networking Challenges and Opportunities (U-NET rsquo09) pp 13ndash18ACMNewYorkNYUSADecember 2009

[14] T Spyropoulos K Psounis and C S Raghavendra ldquoSpray andwait an efficient routing scheme for intermittently connectedmobile networksrdquo in Proceedings of the ACM SIGCOMMWorkshops Conference on Computer Communications pp 252ndash259 August 2005

[15] J Leguay T Friedman and V Conan ldquoDTN routing in amobility pattern spacerdquo in Proceedings of the ACM SIGCOMMWorkshops Conference on Computer Communications pp 276ndash283 August 2005

[16] B Z Yao P Hu X H Lu J J Gao and M H Zhang ldquoTransitnetwork design based on travel time reliabilityrdquo TransportationResearch C vol 43 pp 233ndash248 2014

[17] B Yao P Hu M Zhang and S Wang ldquoArtificial bee colonyalgorithm with scanning strategy for the periodic vehiclerouting problemrdquo Simulation vol 89 no 6 pp 762ndash770 2013


[18] B-Z Yao C-Y Yang and J-B Yao ldquoTunnel surroundingrock displacement prediction using support vector machinerdquoInternational Journal of Computational Intelligence Systems vol3 no 6 pp 843ndash852 2010

[19] B Yao P Hu M Zhang and X Tian ldquoImproved ant colonyoptimization for seafood product delivery routing problemrdquoPROMETmdashTrafficampTransportation vol 26 no 1 pp 1ndash10 2014

[20] B Yao C Yang J Hu J Yao and J Sun ldquoAn improved antcolony optimization for flexible job shop scheduling problemsrdquoAdvanced Science Letters vol 4 no 6-7 pp 2127ndash2131 2011

[21] B Z Yao J B Yao and M H Zhang ldquoImproved supportvector machine regression in multi-step-ahead prediction forrock displacement surrounding a tunnelrdquo Scientia Iranica Inpress

[22] B Yu Z Z Yang and K Chen ldquoHybrid model for predictionof bus arrival times at next stationrdquo Journal of AdvancedTransportation vol 44 no 3 pp 193ndash204 2010

[23] Y Bin Y Zhongzhen and Y Baozhen ldquoBus arrival timeprediction using support vector machinesrdquo Journal of IntelligentTransportation Systems vol 10 no 4 pp 151ndash158 2006

[24] B Yu Z Z Yang and B Z Yao ldquoA hybrid algorithm forvehicle routing problem with time windowsrdquo Expert Systemswith Applications vol 38 no 1 pp 435ndash441 2011

[25] B Yu H B Zhu W J Cai N Ma and B Z Yao ldquoTwo-phaseoptimization approach to transit Hub locationmdashthe case ofDalianrdquo Journal of Transport Geography vol 33 pp 62ndash71 2013

[26] B Yu Z Yang and J Yao ldquoGenetic algorithm for bus frequencyoptimizationrdquo Journal of Transportation Engineering vol 136no 6 pp 576ndash583 2010

[27] M Farooq Bee-Inspired Protocol Engineering From Nature toNetworks Springer New York NY USA 2009

[28] A Zengin H Sarjoughian and H Ekiz ldquoDiscrete event mod-eling of swarm intelligence based routing in network systemsrdquoInformation Sciences vol 222 pp 81ndash98 2013

[29] E Bonabeau M Dorigo and G Theraulaz Swarm IntelligenceFrom Natural to Artificial Systems Oxford University Press1999

[30] MDorigo VManiezzo andA Colorni ldquoAnt system optimiza-tion by a colony of cooperating agentsrdquo IEEE Transactions onSystems Man and Cybernetics B vol 26 no 1 pp 29ndash41 1996

[31] R Schoonderwoerd Collective intelligence for network control[MS thesis] Faculty of Technical Informatics Delft Universityof Technology 1996

[32] M Dorigo and T Stutzle ldquoThe ant colony optimization meta-heuristic algorithms applications and advancesrdquo inHandbookof Metaheuristics pp 251ndash285 Springer 2002

[33] V D Le H Scholten P J M Havinga and H Ngo ldquoLocation-based data dissemination with human mobility using onlinedensity estimationrdquo in Proceedings of the 11th Annual IEEEConsumer Communications amp Networking Conference pp 747ndash754 Las Vegas Nev USA November 2014

[34] A Ahmed and K Abu Bakar ldquoA simulation based study ofwell known routing protocols for delay tolerant networkrdquoWorldApplied Sciences Journal vol 28 no 3 pp 353ndash360 2013

[35] M McNett and G M Voelker UCSD Wireless TopologyDiscovery Project [EBOL] 2013 httpwwwsysnetucsdeduwtdwtdhtml

[36] J Li X Xing R Yu XWang and Y Zhou ldquoSocial relationship-based mobile node location prediction algorithm in oppor-tunistic cognitive networksrdquo WIT Transactions on Informationand Communication Technologies vol 59 pp 113ndash119 2014

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MathematicsJournal of


Mathematical Problems in Engineering

Hindawi Publishing Corporationhttpwwwhindawicom

Differential EquationsInternational Journal of

Volume 2014

Applied MathematicsJournal of


Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of


Mathematical PhysicsAdvances in

Complex AnalysisJournal of


OptimizationJournal of


CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of


Operations ResearchAdvances in

Journal of


Function Spaces

Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of Mathematics and Mathematical Sciences


The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Algebra

Discrete Dynamics in Nature and Society



Decision SciencesAdvances in

Discrete MathematicsJournal of


Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Stochastic AnalysisInternational Journal of


The opportunistic cognitive network is a kind of delaytolerant networks (DTNs) and it utilizes communicationopportunities obtained from node movement to relay pack-ets Mobile nodes are enabled to communicate with eachother even if a route connecting them never exists The basicrouting strategy of opportunistic networks is ldquostore-carry-forwardrdquo Messages are routed between the sender and thedestination(s) and any possible node can be used as a relayprovided that it is likely to bring the message closer to thefinal destination Message transmission is mainly dependenton intermediate relays so relay node selection is criticalfor efficient data dissemination in opportunistic cognitivenetworks

Since human mobility is mostly unpredictable con-ventional routing algorithms for mobile ad hoc networks(MANETs) no longer perform well in the opportunisticcognitive networks Therefore new algorithms are requiredto overcome intermittent connectivity in the opportunisticnetworks

Stochastic routing protocols such as Epidemic [7] FirstContact (FC) [8] and Direct Delivery (DD) [9] broadcastmessages to any encountered node in order to increase thedelivery ratio Epidemic routing diffuses messages similarto the way viruses or bacteria spread in biology Wheneverencountering another node a node replicates and transfersmessages After receivingmessages a node will move to otherplaces and continuously replicate and deliver the messagesto other encountered ones First Contact sends messages tothe first encountered node without copying the messages toother nodes Direct Delivery comprises the trade-off betweenEpidemic and First Contact by finding an optimal number ofmessage copies

Creating more copies of a message increases the messagedelivery but decreases the network lifetime These stochasticrouting approaches usually consider the destinations of mes-sages as nodes rather than locations

Different from stochastic routing current contact-basedrouting chooses the most appropriate nodes relaying mes-sages to the destinations based on historical contact infor-mation such as contact times contact duration and contactcycle

Probabilistic Routing Protocol using History of Encoun-ters and Transitivity (PRoPHET) [10] is a context-based rout-ing protocol based on the history of encounters PRoPHETestimates the delivery predictability for each known desti-nation at each node before passing a message The estima-tion is based on the history of encounters between nodesSimBet [11] uses historical contacts to calculate two metricssimilarity and betweenness The similarity is calculated byhow frequently a node and its destination have met Thebetweenness is calculated by how many nodes which a nodehas met However if the utility metrics are equal SimBetwill prevent its forwarding behavior To improve this flawBUBBLE [12] adds the knowledge of community structure toensure message dissemination In addition the betweennessmay be useless if the message is near its destination Sim-BetAge [13] an improved version of SimBet was proposedto address these shortcomings Spray and Wait [14] thatuses only a handful of copies per message can achieve

comparable delays to an oracle-based optimal scheme thatminimizes delay while using the lowest possible number oftransmissions

The work in [15] does routing by simply calculatingthe delivery probability for a node to be at a location inMobySpace which is a high-dimensional Euclidean spacebased on the preknown mobility model However therequired assumption of that each node has the knowledgeabout mobility patterns of other nodes in the network makesthis work unpractical in realistic scenarios

For the concerns of optimality robustness and flexibilitysome routing protocols are developed by inspiration frombiology [16ndash21] Such routing schemes are also well appliedin the transportation domain [22ndash26] which are basedon prediction mechanism to improve the performance oftransportation systems

BeeHive is developed with inspiration of the foragingprinciples of honeybees [27] Communication between realbees is modeled by designing intelligent bee agents whichare able to make routing decisions in large and complextopologies The simulation results conclude that bee agentsoccupy smaller bandwidth and require less processor timeArtificial Bee Colony algorithm [17] is developed with scan-ning strategy for periodic vehicle routing problemThe workin [28] presents a biologically inspired discrete-event mod-eling approach for simulating alternative computer networkprotocols Adaptation and probabilistic specifications areintroduced into honeybee (BEE) and Routing InformationProtocol (RIP) routing algorithms

The ant colony optimization (ACO) methods have beeninspired by operating principles of ants [29] which empowera colony of ants to perform complex tasks such as nestbuilding and foraging [30]

Schoonderwoerd [31] first developed ACO-basedapproach for routing in telecommunication networks Thebasic principle in the approach is about the use of stigmergyin multiagents interaction Randomized ants traverse thenetwork nodes probabilistically and select the highestprobability path The approach is shown to be sufficientlyadaptive and demonstrates robustness across difficultnetwork conditions Yao et al [19 20] improved ACO fordelivery routing problem and PROMETShop schedulingproblems

The aforementioned routing algorithms show good per-formance on message delivery However geographic coordi-nates of nodes have little or no correlation with their contacttimes so the algorithms do not perform well when eachdevice frequently appears at different regions as most peopledaily do

In this paper we propose the location-prediction andswarm-intelligence-based data dissemination (LOPSI) algo-rithm for opportunistic cognitive networks The LOPSI algo-rithm is a probabilistic routing protocol combining locationprediction and the ant colony optimization It firstly predictspossible locations of relay nodes and destination(s) in succes-sive time seriesThemobile nodes calculate intimacy (contactfrequency) with potential relay nodes using ACO and thenmake a forwarding decision based on node intimacy andprobabilities of node mobility


predictionserver


AP1

AP2

AP3

Nsource

Ndestination

Nk

Nk

Nj

Nj

Trajectory-



2 Network Model














3 Algorithm Design







119898 Thus scene







(119905119896 119883119894) 119896 = 1 2 3 119894 isin 1 2 3 119898 (1)




119864 = 1198831 1198832 119883

119903119883119894 119883119895 119883

119898 119894 = 1 2 119898

119903 = 1 2 119898 119895 = 1 2 119898

(2)


119875 119883 (119905119899) = 119883

119895

= 119875 119883 (119905119899) = 119883

119895| 119883 (119905119899minus1) = 119883

119894 119883 (119905119899minus2) = 119883

119903

(3)



119894at



119899minus2is

119875119903119894119895119883 (119905119899) = 119883

119895

=

119898

sum

119894=1119903=1

119875 119883 (119905119899) = 119883

119895| 119883 (119905119899minus1) = 119883

119894 119883 (119905119899minus2) = 119883

119903

(4)



infinity [18]

119901119903119894119895=

119903=119898119894=119898

sum

119903=1119894=1

119888119903119894119895

sum119898

119896=1119888119903119894119896

(5)



119895 119895 isin 1 2 3 119899 the initial



119896 119883119894) 119896 isin 119873+ 119894 isin 1 2 3 119898



119903


119899


119899is

(b)119883119895= argmax 119875(119899)

119895

(7) return 119883119895


where 119888119903119894119895



119896=1119888119903119894119896


119895and


119894and119883

119903


119875 =

119898⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞

[

[

[

[

[

[

[

[

[

[

[

[

119901111

119901112


11990111119898

119901211

119901212


11990121119898

d

11990111989811

11990111989812


1199011198981119898

11990111989821

11990111989822


1199011198982119898

d

1199011198981198981

1199011198981198982


119901119898119898119898

]

]

]

]

]

]

]

]

]

]

]

]

1198982

(6)



119899by



is

119875 (119899 minus 1 119899 minus 2)

= (119901(119899minus1119899minus2)

11 119901(119899minus1119899minus2)

12 119901(119899minus1119899minus2)

21 119901(119899minus1119899minus2)

22) = (0 1 0 0)

(7)


119875 =

[

[

[

[

119901111

119901112

119901121

119901122

119901211

119901212

119901221

119901222

]

]

]

]

(8)


119875 (119899) = 119875 (119899 minus 1 119899 minus 2) 119875 = 119901(119899)

1 119901(119899)

2 (9)


119883119895= argmax 119901(119899)

1 119901(119899)

2 (10)



119901119889

119871= 119901(119899)

119883119889

119895

(11)


119889at time

slice 119905119899 and 119901(119899)

119883119889

119895



119899



119889





119891






Output 119901119889

119871

(1) 119883119889

119895= 119871 Markov(119873

119889 119905119899)


119871


(1) 119873119904rarr LPS REQ(119862119873

119891119873119889) 119894 = 0 119895 = 0 119896 = 0 119897 = 3

(2) for all 119899 isin [1 119897] do(3) 119883(119873

119889 119905119899) = 119871 Markov(119873

119889 119905119899)

(4) for all 119873119891isin 119862119873

119891 do

(5) 119883(119873119891 119905119899) = 119871 Markov(119873


119891at 119905119899

(6) if (119883(119873119891 119905119899) == 119883

119894(119873119889 119905119899)) then 119873

119891and119873


119899



119891meets119873

119889is stored in set 119865119875



119904 SEI(119865(119873

119891 119875 119879) 119883(119873

119889 119905119899))




119905119899+1 119905119899+2 119905

119899+119897

119883 (119873119889 119905119897) = (119905

1 119883119894) (1199052 119883119895) (119905

119897 119883119901) sube (119905

119896 119883119894)

119896 isin 119873+ 119894 119895 119901 isin 1 2 3 119898

(12)


119891in the set 119862119873





119891 119875 119879) 119883(119873

119889 119905119899))


Algorithm 3




119904finds









119875119889

119871119904119891=

119875 Markov (119873119891 119905119899)

119899

119873119891isin 119865 119905

119899isin 119879 119899 isin [1 119897]

(13)



119889at time


119873119891meets 119873



119904

1198651015840sube 119865 119865

1015840119873119891 = argmax119891leCOPY

119875119889

119871119904119891 (14)










119891 of 119873



119894and 119873

119895depends



119888

119894119895



119894and119873


119877119894119895 (119905)

=

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

forallΔ119868119888

119894119895lt 119896

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

times (1 minus 120588)119890

existΔ119868119888

119894119895gt 119896

0 Δ119868119888


(15)




119904and the






119877119894119895 (119905 + Δ119905) = 119877119894119895 (

119905) + Δ119877119894119895 (119905)

Δ119877119894119895 (119905)

=

(119899119894119895+ Δ119899119894119895) times sum

(119899119894119895+Δ119899119894119895)

119888=1Δ119863119888

119894119895

sum

(119899119894119895+Δ119899119894119895)

119888=1Δ119868119888

119894119895

minus

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

Δ119899119894119895gt 0 forallΔ119868

119888

119894119895lt 119896

minus120588 times

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

Δ119899119894119895= 0 Δ119905 gt 119896 forallΔ119868

119888

119894119895lt 119896

0 otherwise(16)



119891 and



119891 and


equation

119901119889

119877119904119891=

1 119895 = 119889


119894 119873119889)

[119877119891119889 (119905)]

120572

sdot [120578119891119889 (119905)]

120573

sum119873119891isinallowd119889 119877

120572

119895119889sdot 120578120573

119895119889


119889

(17)


allowd119905119889= 119865 119873

119891| (119877119891119889 (119905) gt 119877119894119889 (

119905))

allowd119905119889isin 119865 119873

119891 allowd

119889= allowd119905

119889minus tabu

119889

(18)




119891119889 which

is given by

120578119891119889=

1

119871119891119889

(19)

where 119871119891119889


119889will













Input 119873119894119873119895

Output 119877119894119895(119879)


119894119895(119888 119905119904 119905119890)


119894119895(0) = 0 119896 = 119871 119891119897119886119892 = 1

(5) forall 119905 = 0 119879(6) if 119873


119895

(7) 119899119894119895= ++119888

(8) for all 119888 = 1 119899119894119895

(9) Δ119863119888

119894119895[119888] = 119905

119890minus 119905119904

(10) 119863119888

119894119895+= Δ119863119888

119894119895

(11) Δ119868119888

119894119895[119888] = 119905

119904minus 119905

(12) 119868119888

119894119895+= Δ119868119888

119894119895

(13) 119905 = 119905119890

(14) if (Δ119868119888119894119895[119888] gt 119896)


119894119895(119879) = ((119899

119894119895times 119863119888

119894119895) 119868119888

119894119895) times (1 minus 120588)

119890

(23) else(24) return 119877

119894119895(119879) = (119899

119894119895times 119863119888

119894119895) 119868119888

119894119895

(25) end if



+ state space 119864 = 119883

119895 119895 = 1 2 119898119898 isin 119873

+ time

slice set 119879 = 119905119899 119899 = 0 1 2 119897 119897 isin 119873



(3) LOPSI Sec(119873119904)

(4) for forall119873119891isin 1198651015840119873119891



Algorithm 5 LOPSI(119873119904 119873119889)




119891and 119873






119904will




1198621015840sube 119862 119873

119891 119862

1015840119873119891 = argmax119891leCOPY

119875119889

119877119904119891 (20)






119889and 119873

119891isin




119901119889

119904119891= 120574 lowast 119901

119889

119877119904119891+ 120575 lowast 119901

119889

119871119904119891 120574 + 120575 = 1 (21)

where 119901119889119877119904119891











2)

Storage space 119874(1198732) 119874(119873

3)


119894)










119891

(2) for forall119873119891isin 119862119873

119891

(3) update intimacy(119873119894 119873119891)



119889

lowast

(5) if 119873119889isin 119862119873

119891


119889

(7) tabu119889larr 119873

119894 119873119889


119904sends REQ(119862119873

119891119873119889) to the server


(12) 119873119904receives SEI

(13) 119873119904sends119873

119889to the nodes in 119865119873

119891

(14) for forall119873119891isin 119865119873

119891

(15) Send intimacy(119873119891 119873119889) to119873

119904

(16) end for(17) 119873

119904calculates 119901119889

119904119891according to (21)


1015840119873119891 = 119865 119873

119891

(20) else(21) 119865

1015840119873119891 = argmax

119891leCOPY (119901119889

119904119891)

(22) end if(23) 119873

119894sends data to 1198651015840119873

119891

(24) tabu119889larr 119873

119894 1198651015840119873119891


(26) end if



Parameter Value

Scene features


Node features










119867 =

119873

119884

(22)


119877 =

119878

119884

(23)


119871 =

1

119884

119884

sum

119894=1

(119879119889119894minus 119879119904119894) (24)


originated and 119879119889119894



119862 =

119876

119875

(25)








LOPDADLOPSI

100 150 200 250 300

The number of nodes

50

45

40

35

30

25

20

Aver

age h

ops

(a)

02

03

04

05

06

07

Deli

very

ratio

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(b)

2500

3000

3500

4000

4500

5000

5500

6000

Aver

age l

aten

cy

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(c)

30

35

40

45

50

55

60

Tran

smiss

ion

cost


ACODADLOPDADLOPSI

Spray and Wait

(d)






55

50

45

40

35

30

25

20

Aver

age h

ops

4 5 6 7 8 9 10 11 12 13TTL (h)

ACODAD

LOPDADLOPSI


(a)

4 5 6 7 8 9 10 11 12 13TTL (h)

07

06

05

04

03

02

Del

iver

y ra

tio

ACODAD

LOPDADLOPSI


(b)

20003000400050006000700080009000

10000110001200013000

Aver

age l

aten

cy (s

)

4 5 6 7 8 9 10 11 12 13

ACODAD

LOPDADLOPSI


(c)

25

30

35

40

45

50

55

60

Tran

smiss

ion

cost

4 6 8 10 12 14TTL (h)

ACODADLOPDADLOPSI

Spray and Wait

(d)













Acknowledgment


References













































Volume 2014




Journal of











Journal of


Function Spaces






Algebra










predictionserver


AP1

AP2

AP3

Nsource

Ndestination

Nk

Nk

Nj

Nj

Trajectory-



2 Network Model














3 Algorithm Design







119898 Thus scene







(119905119896 119883119894) 119896 = 1 2 3 119894 isin 1 2 3 119898 (1)




119864 = 1198831 1198832 119883

119903119883119894 119883119895 119883

119898 119894 = 1 2 119898

119903 = 1 2 119898 119895 = 1 2 119898

(2)


119875 119883 (119905119899) = 119883

119895

= 119875 119883 (119905119899) = 119883

119895| 119883 (119905119899minus1) = 119883

119894 119883 (119905119899minus2) = 119883

119903

(3)



119894at



119899minus2is

119875119903119894119895119883 (119905119899) = 119883

119895

=

119898

sum

119894=1119903=1

119875 119883 (119905119899) = 119883

119895| 119883 (119905119899minus1) = 119883

119894 119883 (119905119899minus2) = 119883

119903

(4)



infinity [18]

119901119903119894119895=

119903=119898119894=119898

sum

119903=1119894=1

119888119903119894119895

sum119898

119896=1119888119903119894119896

(5)



119895 119895 isin 1 2 3 119899 the initial



119896 119883119894) 119896 isin 119873+ 119894 isin 1 2 3 119898



119903


119899


119899is

(b)119883119895= argmax 119875(119899)

119895

(7) return 119883119895


where 119888119903119894119895



119896=1119888119903119894119896


119895and


119894and119883

119903


119875 =

119898⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞

[

[

[

[

[

[

[

[

[

[

[

[

119901111

119901112


11990111119898

119901211

119901212


11990121119898

d

11990111989811

11990111989812


1199011198981119898

11990111989821

11990111989822


1199011198982119898

d

1199011198981198981

1199011198981198982


119901119898119898119898

]

]

]

]

]

]

]

]

]

]

]

]

1198982

(6)



119899by



is

119875 (119899 minus 1 119899 minus 2)

= (119901(119899minus1119899minus2)

11 119901(119899minus1119899minus2)

12 119901(119899minus1119899minus2)

21 119901(119899minus1119899minus2)

22) = (0 1 0 0)

(7)


119875 =

[

[

[

[

119901111

119901112

119901121

119901122

119901211

119901212

119901221

119901222

]

]

]

]

(8)


119875 (119899) = 119875 (119899 minus 1 119899 minus 2) 119875 = 119901(119899)

1 119901(119899)

2 (9)


119883119895= argmax 119901(119899)

1 119901(119899)

2 (10)



119901119889

119871= 119901(119899)

119883119889

119895

(11)


119889at time

slice 119905119899 and 119901(119899)

119883119889

119895



119899



119889





119891






Output 119901119889

119871

(1) 119883119889

119895= 119871 Markov(119873

119889 119905119899)


119871


(1) 119873119904rarr LPS REQ(119862119873

119891119873119889) 119894 = 0 119895 = 0 119896 = 0 119897 = 3

(2) for all 119899 isin [1 119897] do(3) 119883(119873

119889 119905119899) = 119871 Markov(119873

119889 119905119899)

(4) for all 119873119891isin 119862119873

119891 do

(5) 119883(119873119891 119905119899) = 119871 Markov(119873


119891at 119905119899

(6) if (119883(119873119891 119905119899) == 119883

119894(119873119889 119905119899)) then 119873

119891and119873


119899



119891meets119873

119889is stored in set 119865119875



119904 SEI(119865(119873

119891 119875 119879) 119883(119873

119889 119905119899))




119905119899+1 119905119899+2 119905

119899+119897

119883 (119873119889 119905119897) = (119905

1 119883119894) (1199052 119883119895) (119905

119897 119883119901) sube (119905

119896 119883119894)

119896 isin 119873+ 119894 119895 119901 isin 1 2 3 119898

(12)


119891in the set 119862119873





119891 119875 119879) 119883(119873

119889 119905119899))


Algorithm 3




119904finds









119875119889

119871119904119891=

119875 Markov (119873119891 119905119899)

119899

119873119891isin 119865 119905

119899isin 119879 119899 isin [1 119897]

(13)



119889at time


119873119891meets 119873



119904

1198651015840sube 119865 119865

1015840119873119891 = argmax119891leCOPY

119875119889

119871119904119891 (14)










119891 of 119873



119894and 119873

119895depends



119888

119894119895



119894and119873


119877119894119895 (119905)

=

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

forallΔ119868119888

119894119895lt 119896

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

times (1 minus 120588)119890

existΔ119868119888

119894119895gt 119896

0 Δ119868119888


(15)




119904and the






119877119894119895 (119905 + Δ119905) = 119877119894119895 (

119905) + Δ119877119894119895 (119905)

Δ119877119894119895 (119905)

=

(119899119894119895+ Δ119899119894119895) times sum

(119899119894119895+Δ119899119894119895)

119888=1Δ119863119888

119894119895

sum

(119899119894119895+Δ119899119894119895)

119888=1Δ119868119888

119894119895

minus

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

Δ119899119894119895gt 0 forallΔ119868

119888

119894119895lt 119896

minus120588 times

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

Δ119899119894119895= 0 Δ119905 gt 119896 forallΔ119868

119888

119894119895lt 119896

0 otherwise(16)



119891 and



119891 and


equation

119901119889

119877119904119891=

1 119895 = 119889


119894 119873119889)

[119877119891119889 (119905)]

120572

sdot [120578119891119889 (119905)]

120573

sum119873119891isinallowd119889 119877

120572

119895119889sdot 120578120573

119895119889


119889

(17)


allowd119905119889= 119865 119873

119891| (119877119891119889 (119905) gt 119877119894119889 (

119905))

allowd119905119889isin 119865 119873

119891 allowd

119889= allowd119905

119889minus tabu

119889

(18)




119891119889 which

is given by

120578119891119889=

1

119871119891119889

(19)

where 119871119891119889


119889will













Input 119873119894119873119895

Output 119877119894119895(119879)


119894119895(119888 119905119904 119905119890)


119894119895(0) = 0 119896 = 119871 119891119897119886119892 = 1

(5) forall 119905 = 0 119879(6) if 119873


119895

(7) 119899119894119895= ++119888

(8) for all 119888 = 1 119899119894119895

(9) Δ119863119888

119894119895[119888] = 119905

119890minus 119905119904

(10) 119863119888

119894119895+= Δ119863119888

119894119895

(11) Δ119868119888

119894119895[119888] = 119905

119904minus 119905

(12) 119868119888

119894119895+= Δ119868119888

119894119895

(13) 119905 = 119905119890

(14) if (Δ119868119888119894119895[119888] gt 119896)


119894119895(119879) = ((119899

119894119895times 119863119888

119894119895) 119868119888

119894119895) times (1 minus 120588)

119890

(23) else(24) return 119877

119894119895(119879) = (119899

119894119895times 119863119888

119894119895) 119868119888

119894119895

(25) end if



+ state space 119864 = 119883

119895 119895 = 1 2 119898119898 isin 119873

+ time

slice set 119879 = 119905119899 119899 = 0 1 2 119897 119897 isin 119873



(3) LOPSI Sec(119873119904)

(4) for forall119873119891isin 1198651015840119873119891



Algorithm 5 LOPSI(119873119904 119873119889)




119891and 119873






119904will




1198621015840sube 119862 119873

119891 119862

1015840119873119891 = argmax119891leCOPY

119875119889

119877119904119891 (20)






119889and 119873

119891isin




119901119889

119904119891= 120574 lowast 119901

119889

119877119904119891+ 120575 lowast 119901

119889

119871119904119891 120574 + 120575 = 1 (21)

where 119901119889119877119904119891











2)

Storage space 119874(1198732) 119874(119873

3)


119894)










119891

(2) for forall119873119891isin 119862119873

119891

(3) update intimacy(119873119894 119873119891)



119889

lowast

(5) if 119873119889isin 119862119873

119891


119889

(7) tabu119889larr 119873

119894 119873119889


119904sends REQ(119862119873

119891119873119889) to the server


(12) 119873119904receives SEI

(13) 119873119904sends119873

119889to the nodes in 119865119873

119891

(14) for forall119873119891isin 119865119873

119891

(15) Send intimacy(119873119891 119873119889) to119873

119904

(16) end for(17) 119873

119904calculates 119901119889

119904119891according to (21)


1015840119873119891 = 119865 119873

119891

(20) else(21) 119865

1015840119873119891 = argmax

119891leCOPY (119901119889

119904119891)

(22) end if(23) 119873

119894sends data to 1198651015840119873

119891

(24) tabu119889larr 119873

119894 1198651015840119873119891


(26) end if



Parameter Value

Scene features


Node features










119867 =

119873

119884

(22)


119877 =

119878

119884

(23)


119871 =

1

119884

119884

sum

119894=1

(119879119889119894minus 119879119904119894) (24)


originated and 119879119889119894



119862 =

119876

119875

(25)








LOPDADLOPSI

100 150 200 250 300

The number of nodes

50

45

40

35

30

25

20

Aver

age h

ops

(a)

02

03

04

05

06

07

Deli

very

ratio

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(b)

2500

3000

3500

4000

4500

5000

5500

6000

Aver

age l

aten

cy

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(c)

30

35

40

45

50

55

60

Tran

smiss

ion

cost


ACODADLOPDADLOPSI

Spray and Wait

(d)






55

50

45

40

35

30

25

20

Aver

age h

ops

4 5 6 7 8 9 10 11 12 13TTL (h)

ACODAD

LOPDADLOPSI


(a)

4 5 6 7 8 9 10 11 12 13TTL (h)

07

06

05

04

03

02

Del

iver

y ra

tio

ACODAD

LOPDADLOPSI


(b)

20003000400050006000700080009000

10000110001200013000

Aver

age l

aten

cy (s

)

4 5 6 7 8 9 10 11 12 13

ACODAD

LOPDADLOPSI


(c)

25

30

35

40

45

50

55

60

Tran

smiss

ion

cost

4 6 8 10 12 14TTL (h)

ACODADLOPDADLOPSI

Spray and Wait

(d)













Acknowledgment


References













































Volume 2014




Journal of











Journal of


Function Spaces






Algebra










3 Algorithm Design







119898 Thus scene







(119905119896 119883119894) 119896 = 1 2 3 119894 isin 1 2 3 119898 (1)




119864 = 1198831 1198832 119883

119903119883119894 119883119895 119883

119898 119894 = 1 2 119898

119903 = 1 2 119898 119895 = 1 2 119898

(2)


119875 119883 (119905119899) = 119883

119895

= 119875 119883 (119905119899) = 119883

119895| 119883 (119905119899minus1) = 119883

119894 119883 (119905119899minus2) = 119883

119903

(3)



119894at



119899minus2is

119875119903119894119895119883 (119905119899) = 119883

119895

=

119898

sum

119894=1119903=1

119875 119883 (119905119899) = 119883

119895| 119883 (119905119899minus1) = 119883

119894 119883 (119905119899minus2) = 119883

119903

(4)



infinity [18]

119901119903119894119895=

119903=119898119894=119898

sum

119903=1119894=1

119888119903119894119895

sum119898

119896=1119888119903119894119896

(5)



119895 119895 isin 1 2 3 119899 the initial



119896 119883119894) 119896 isin 119873+ 119894 isin 1 2 3 119898



119903


119899


119899is

(b)119883119895= argmax 119875(119899)

119895

(7) return 119883119895


where 119888119903119894119895



119896=1119888119903119894119896


119895and


119894and119883

119903


119875 =

119898⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞

[

[

[

[

[

[

[

[

[

[

[

[

119901111

119901112


11990111119898

119901211

119901212


11990121119898

d

11990111989811

11990111989812


1199011198981119898

11990111989821

11990111989822


1199011198982119898

d

1199011198981198981

1199011198981198982


119901119898119898119898

]

]

]

]

]

]

]

]

]

]

]

]

1198982

(6)



119899by



is

119875 (119899 minus 1 119899 minus 2)

= (119901(119899minus1119899minus2)

11 119901(119899minus1119899minus2)

12 119901(119899minus1119899minus2)

21 119901(119899minus1119899minus2)

22) = (0 1 0 0)

(7)


119875 =

[

[

[

[

119901111

119901112

119901121

119901122

119901211

119901212

119901221

119901222

]

]

]

]

(8)


119875 (119899) = 119875 (119899 minus 1 119899 minus 2) 119875 = 119901(119899)

1 119901(119899)

2 (9)


119883119895= argmax 119901(119899)

1 119901(119899)

2 (10)



119901119889

119871= 119901(119899)

119883119889

119895

(11)


119889at time

slice 119905119899 and 119901(119899)

119883119889

119895



119899



119889





119891






Output 119901119889

119871

(1) 119883119889

119895= 119871 Markov(119873

119889 119905119899)


119871


(1) 119873119904rarr LPS REQ(119862119873

119891119873119889) 119894 = 0 119895 = 0 119896 = 0 119897 = 3

(2) for all 119899 isin [1 119897] do(3) 119883(119873

119889 119905119899) = 119871 Markov(119873

119889 119905119899)

(4) for all 119873119891isin 119862119873

119891 do

(5) 119883(119873119891 119905119899) = 119871 Markov(119873


119891at 119905119899

(6) if (119883(119873119891 119905119899) == 119883

119894(119873119889 119905119899)) then 119873

119891and119873


119899



119891meets119873

119889is stored in set 119865119875



119904 SEI(119865(119873

119891 119875 119879) 119883(119873

119889 119905119899))




119905119899+1 119905119899+2 119905

119899+119897

119883 (119873119889 119905119897) = (119905

1 119883119894) (1199052 119883119895) (119905

119897 119883119901) sube (119905

119896 119883119894)

119896 isin 119873+ 119894 119895 119901 isin 1 2 3 119898

(12)


119891in the set 119862119873





119891 119875 119879) 119883(119873

119889 119905119899))


Algorithm 3




119904finds









119875119889

119871119904119891=

119875 Markov (119873119891 119905119899)

119899

119873119891isin 119865 119905

119899isin 119879 119899 isin [1 119897]

(13)



119889at time


119873119891meets 119873



119904

1198651015840sube 119865 119865

1015840119873119891 = argmax119891leCOPY

119875119889

119871119904119891 (14)










119891 of 119873



119894and 119873

119895depends



119888

119894119895



119894and119873


119877119894119895 (119905)

=

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

forallΔ119868119888

119894119895lt 119896

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

times (1 minus 120588)119890

existΔ119868119888

119894119895gt 119896

0 Δ119868119888


(15)




119904and the






119877119894119895 (119905 + Δ119905) = 119877119894119895 (

119905) + Δ119877119894119895 (119905)

Δ119877119894119895 (119905)

=

(119899119894119895+ Δ119899119894119895) times sum

(119899119894119895+Δ119899119894119895)

119888=1Δ119863119888

119894119895

sum

(119899119894119895+Δ119899119894119895)

119888=1Δ119868119888

119894119895

minus

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

Δ119899119894119895gt 0 forallΔ119868

119888

119894119895lt 119896

minus120588 times

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

Δ119899119894119895= 0 Δ119905 gt 119896 forallΔ119868

119888

119894119895lt 119896

0 otherwise(16)



119891 and



119891 and


equation

119901119889

119877119904119891=

1 119895 = 119889


119894 119873119889)

[119877119891119889 (119905)]

120572

sdot [120578119891119889 (119905)]

120573

sum119873119891isinallowd119889 119877

120572

119895119889sdot 120578120573

119895119889


119889

(17)


allowd119905119889= 119865 119873

119891| (119877119891119889 (119905) gt 119877119894119889 (

119905))

allowd119905119889isin 119865 119873

119891 allowd

119889= allowd119905

119889minus tabu

119889

(18)




119891119889 which

is given by

120578119891119889=

1

119871119891119889

(19)

where 119871119891119889


119889will













Input 119873119894119873119895

Output 119877119894119895(119879)


119894119895(119888 119905119904 119905119890)


119894119895(0) = 0 119896 = 119871 119891119897119886119892 = 1

(5) forall 119905 = 0 119879(6) if 119873


119895

(7) 119899119894119895= ++119888

(8) for all 119888 = 1 119899119894119895

(9) Δ119863119888

119894119895[119888] = 119905

119890minus 119905119904

(10) 119863119888

119894119895+= Δ119863119888

119894119895

(11) Δ119868119888

119894119895[119888] = 119905

119904minus 119905

(12) 119868119888

119894119895+= Δ119868119888

119894119895

(13) 119905 = 119905119890

(14) if (Δ119868119888119894119895[119888] gt 119896)


119894119895(119879) = ((119899

119894119895times 119863119888

119894119895) 119868119888

119894119895) times (1 minus 120588)

119890

(23) else(24) return 119877

119894119895(119879) = (119899

119894119895times 119863119888

119894119895) 119868119888

119894119895

(25) end if



+ state space 119864 = 119883

119895 119895 = 1 2 119898119898 isin 119873

+ time

slice set 119879 = 119905119899 119899 = 0 1 2 119897 119897 isin 119873



(3) LOPSI Sec(119873119904)

(4) for forall119873119891isin 1198651015840119873119891



Algorithm 5 LOPSI(119873119904 119873119889)




119891and 119873






119904will




1198621015840sube 119862 119873

119891 119862

1015840119873119891 = argmax119891leCOPY

119875119889

119877119904119891 (20)






119889and 119873

119891isin




119901119889

119904119891= 120574 lowast 119901

119889

119877119904119891+ 120575 lowast 119901

119889

119871119904119891 120574 + 120575 = 1 (21)

where 119901119889119877119904119891











2)

Storage space 119874(1198732) 119874(119873

3)


119894)










119891

(2) for forall119873119891isin 119862119873

119891

(3) update intimacy(119873119894 119873119891)



119889

lowast

(5) if 119873119889isin 119862119873

119891


119889

(7) tabu119889larr 119873

119894 119873119889


119904sends REQ(119862119873

119891119873119889) to the server


(12) 119873119904receives SEI

(13) 119873119904sends119873

119889to the nodes in 119865119873

119891

(14) for forall119873119891isin 119865119873

119891

(15) Send intimacy(119873119891 119873119889) to119873

119904

(16) end for(17) 119873

119904calculates 119901119889

119904119891according to (21)


1015840119873119891 = 119865 119873

119891

(20) else(21) 119865

1015840119873119891 = argmax

119891leCOPY (119901119889

119904119891)

(22) end if(23) 119873

119894sends data to 1198651015840119873

119891

(24) tabu119889larr 119873

119894 1198651015840119873119891


(26) end if



Parameter Value

Scene features


Node features










119867 =

119873

119884

(22)


119877 =

119878

119884

(23)


119871 =

1

119884

119884

sum

119894=1

(119879119889119894minus 119879119904119894) (24)


originated and 119879119889119894



119862 =

119876

119875

(25)








LOPDADLOPSI

100 150 200 250 300

The number of nodes

50

45

40

35

30

25

20

Aver

age h

ops

(a)

02

03

04

05

06

07

Deli

very

ratio

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(b)

2500

3000

3500

4000

4500

5000

5500

6000

Aver

age l

aten

cy

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(c)

30

35

40

45

50

55

60

Tran

smiss

ion

cost


ACODADLOPDADLOPSI

Spray and Wait

(d)






55

50

45

40

35

30

25

20

Aver

age h

ops

4 5 6 7 8 9 10 11 12 13TTL (h)

ACODAD

LOPDADLOPSI


(a)

4 5 6 7 8 9 10 11 12 13TTL (h)

07

06

05

04

03

02

Del

iver

y ra

tio

ACODAD

LOPDADLOPSI


(b)

20003000400050006000700080009000

10000110001200013000

Aver

age l

aten

cy (s

)

4 5 6 7 8 9 10 11 12 13

ACODAD

LOPDADLOPSI


(c)

25

30

35

40

45

50

55

60

Tran

smiss

ion

cost

4 6 8 10 12 14TTL (h)

ACODADLOPDADLOPSI

Spray and Wait

(d)













Acknowledgment


References













































Volume 2014




Journal of











Journal of


Function Spaces






Algebra











119895 119895 isin 1 2 3 119899 the initial



119896 119883119894) 119896 isin 119873+ 119894 isin 1 2 3 119898



119903


119899


119899is

(b)119883119895= argmax 119875(119899)

119895

(7) return 119883119895


where 119888119903119894119895



119896=1119888119903119894119896


119895and


119894and119883

119903


119875 =

119898⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞

[

[

[

[

[

[

[

[

[

[

[

[

119901111

119901112


11990111119898

119901211

119901212


11990121119898

d

11990111989811

11990111989812


1199011198981119898

11990111989821

11990111989822


1199011198982119898

d

1199011198981198981

1199011198981198982


119901119898119898119898

]

]

]

]

]

]

]

]

]

]

]

]

1198982

(6)



119899by



is

119875 (119899 minus 1 119899 minus 2)

= (119901(119899minus1119899minus2)

11 119901(119899minus1119899minus2)

12 119901(119899minus1119899minus2)

21 119901(119899minus1119899minus2)

22) = (0 1 0 0)

(7)


119875 =

[

[

[

[

119901111

119901112

119901121

119901122

119901211

119901212

119901221

119901222

]

]

]

]

(8)


119875 (119899) = 119875 (119899 minus 1 119899 minus 2) 119875 = 119901(119899)

1 119901(119899)

2 (9)


119883119895= argmax 119901(119899)

1 119901(119899)

2 (10)



119901119889

119871= 119901(119899)

119883119889

119895

(11)


119889at time

slice 119905119899 and 119901(119899)

119883119889

119895



119899



119889





119891






Output 119901119889

119871

(1) 119883119889

119895= 119871 Markov(119873

119889 119905119899)


119871


(1) 119873119904rarr LPS REQ(119862119873

119891119873119889) 119894 = 0 119895 = 0 119896 = 0 119897 = 3

(2) for all 119899 isin [1 119897] do(3) 119883(119873

119889 119905119899) = 119871 Markov(119873

119889 119905119899)

(4) for all 119873119891isin 119862119873

119891 do

(5) 119883(119873119891 119905119899) = 119871 Markov(119873


119891at 119905119899

(6) if (119883(119873119891 119905119899) == 119883

119894(119873119889 119905119899)) then 119873

119891and119873


119899



119891meets119873

119889is stored in set 119865119875



119904 SEI(119865(119873

119891 119875 119879) 119883(119873

119889 119905119899))




119905119899+1 119905119899+2 119905

119899+119897

119883 (119873119889 119905119897) = (119905

1 119883119894) (1199052 119883119895) (119905

119897 119883119901) sube (119905

119896 119883119894)

119896 isin 119873+ 119894 119895 119901 isin 1 2 3 119898

(12)


119891in the set 119862119873





119891 119875 119879) 119883(119873

119889 119905119899))


Algorithm 3




119904finds









119875119889

119871119904119891=

119875 Markov (119873119891 119905119899)

119899

119873119891isin 119865 119905

119899isin 119879 119899 isin [1 119897]

(13)



119889at time


119873119891meets 119873



119904

1198651015840sube 119865 119865

1015840119873119891 = argmax119891leCOPY

119875119889

119871119904119891 (14)










119891 of 119873



119894and 119873

119895depends



119888

119894119895



119894and119873


119877119894119895 (119905)

=

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

forallΔ119868119888

119894119895lt 119896

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

times (1 minus 120588)119890

existΔ119868119888

119894119895gt 119896

0 Δ119868119888


(15)




119904and the






119877119894119895 (119905 + Δ119905) = 119877119894119895 (

119905) + Δ119877119894119895 (119905)

Δ119877119894119895 (119905)

=

(119899119894119895+ Δ119899119894119895) times sum

(119899119894119895+Δ119899119894119895)

119888=1Δ119863119888

119894119895

sum

(119899119894119895+Δ119899119894119895)

119888=1Δ119868119888

119894119895

minus

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

Δ119899119894119895gt 0 forallΔ119868

119888

119894119895lt 119896

minus120588 times

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

Δ119899119894119895= 0 Δ119905 gt 119896 forallΔ119868

119888

119894119895lt 119896

0 otherwise(16)



119891 and



119891 and


equation

119901119889

119877119904119891=

1 119895 = 119889


119894 119873119889)

[119877119891119889 (119905)]

120572

sdot [120578119891119889 (119905)]

120573

sum119873119891isinallowd119889 119877

120572

119895119889sdot 120578120573

119895119889


119889

(17)


allowd119905119889= 119865 119873

119891| (119877119891119889 (119905) gt 119877119894119889 (

119905))

allowd119905119889isin 119865 119873

119891 allowd

119889= allowd119905

119889minus tabu

119889

(18)




119891119889 which

is given by

120578119891119889=

1

119871119891119889

(19)

where 119871119891119889


119889will













Input 119873119894119873119895

Output 119877119894119895(119879)


119894119895(119888 119905119904 119905119890)


119894119895(0) = 0 119896 = 119871 119891119897119886119892 = 1

(5) forall 119905 = 0 119879(6) if 119873


119895

(7) 119899119894119895= ++119888

(8) for all 119888 = 1 119899119894119895

(9) Δ119863119888

119894119895[119888] = 119905

119890minus 119905119904

(10) 119863119888

119894119895+= Δ119863119888

119894119895

(11) Δ119868119888

119894119895[119888] = 119905

119904minus 119905

(12) 119868119888

119894119895+= Δ119868119888

119894119895

(13) 119905 = 119905119890

(14) if (Δ119868119888119894119895[119888] gt 119896)


119894119895(119879) = ((119899

119894119895times 119863119888

119894119895) 119868119888

119894119895) times (1 minus 120588)

119890

(23) else(24) return 119877

119894119895(119879) = (119899

119894119895times 119863119888

119894119895) 119868119888

119894119895

(25) end if



+ state space 119864 = 119883

119895 119895 = 1 2 119898119898 isin 119873

+ time

slice set 119879 = 119905119899 119899 = 0 1 2 119897 119897 isin 119873



(3) LOPSI Sec(119873119904)

(4) for forall119873119891isin 1198651015840119873119891



Algorithm 5 LOPSI(119873119904 119873119889)




119891and 119873






119904will




1198621015840sube 119862 119873

119891 119862

1015840119873119891 = argmax119891leCOPY

119875119889

119877119904119891 (20)






119889and 119873

119891isin




119901119889

119904119891= 120574 lowast 119901

119889

119877119904119891+ 120575 lowast 119901

119889

119871119904119891 120574 + 120575 = 1 (21)

where 119901119889119877119904119891











2)

Storage space 119874(1198732) 119874(119873

3)


119894)










119891

(2) for forall119873119891isin 119862119873

119891

(3) update intimacy(119873119894 119873119891)



119889

lowast

(5) if 119873119889isin 119862119873

119891


119889

(7) tabu119889larr 119873

119894 119873119889


119904sends REQ(119862119873

119891119873119889) to the server


(12) 119873119904receives SEI

(13) 119873119904sends119873

119889to the nodes in 119865119873

119891

(14) for forall119873119891isin 119865119873

119891

(15) Send intimacy(119873119891 119873119889) to119873

119904

(16) end for(17) 119873

119904calculates 119901119889

119904119891according to (21)


1015840119873119891 = 119865 119873

119891

(20) else(21) 119865

1015840119873119891 = argmax

119891leCOPY (119901119889

119904119891)

(22) end if(23) 119873

119894sends data to 1198651015840119873

119891

(24) tabu119889larr 119873

119894 1198651015840119873119891


(26) end if



Parameter Value

Scene features


Node features










119867 =

119873

119884

(22)


119877 =

119878

119884

(23)


119871 =

1

119884

119884

sum

119894=1

(119879119889119894minus 119879119904119894) (24)


originated and 119879119889119894



119862 =

119876

119875

(25)








LOPDADLOPSI

100 150 200 250 300

The number of nodes

50

45

40

35

30

25

20

Aver

age h

ops

(a)

02

03

04

05

06

07

Deli

very

ratio

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(b)

2500

3000

3500

4000

4500

5000

5500

6000

Aver

age l

aten

cy

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(c)

30

35

40

45

50

55

60

Tran

smiss

ion

cost


ACODADLOPDADLOPSI

Spray and Wait

(d)






55

50

45

40

35

30

25

20

Aver

age h

ops

4 5 6 7 8 9 10 11 12 13TTL (h)

ACODAD

LOPDADLOPSI


(a)

4 5 6 7 8 9 10 11 12 13TTL (h)

07

06

05

04

03

02

Del

iver

y ra

tio

ACODAD

LOPDADLOPSI


(b)

20003000400050006000700080009000

10000110001200013000

Aver

age l

aten

cy (s

)

4 5 6 7 8 9 10 11 12 13

ACODAD

LOPDADLOPSI


(c)

25

30

35

40

45

50

55

60

Tran

smiss

ion

cost

4 6 8 10 12 14TTL (h)

ACODADLOPDADLOPSI

Spray and Wait

(d)













Acknowledgment


References













































Volume 2014




Journal of











Journal of


Function Spaces






Algebra













Output 119901119889

119871

(1) 119883119889

119895= 119871 Markov(119873

119889 119905119899)


119871


(1) 119873119904rarr LPS REQ(119862119873

119891119873119889) 119894 = 0 119895 = 0 119896 = 0 119897 = 3

(2) for all 119899 isin [1 119897] do(3) 119883(119873

119889 119905119899) = 119871 Markov(119873

119889 119905119899)

(4) for all 119873119891isin 119862119873

119891 do

(5) 119883(119873119891 119905119899) = 119871 Markov(119873


119891at 119905119899

(6) if (119883(119873119891 119905119899) == 119883

119894(119873119889 119905119899)) then 119873

119891and119873


119899



119891meets119873

119889is stored in set 119865119875



119904 SEI(119865(119873

119891 119875 119879) 119883(119873

119889 119905119899))




119905119899+1 119905119899+2 119905

119899+119897

119883 (119873119889 119905119897) = (119905

1 119883119894) (1199052 119883119895) (119905

119897 119883119901) sube (119905

119896 119883119894)

119896 isin 119873+ 119894 119895 119901 isin 1 2 3 119898

(12)


119891in the set 119862119873





119891 119875 119879) 119883(119873

119889 119905119899))


Algorithm 3




119904finds









119875119889

119871119904119891=

119875 Markov (119873119891 119905119899)

119899

119873119891isin 119865 119905

119899isin 119879 119899 isin [1 119897]

(13)



119889at time


119873119891meets 119873



119904

1198651015840sube 119865 119865

1015840119873119891 = argmax119891leCOPY

119875119889

119871119904119891 (14)










119891 of 119873



119894and 119873

119895depends



119888

119894119895



119894and119873


119877119894119895 (119905)

=

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

forallΔ119868119888

119894119895lt 119896

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

times (1 minus 120588)119890

existΔ119868119888

119894119895gt 119896

0 Δ119868119888


(15)




119904and the






119877119894119895 (119905 + Δ119905) = 119877119894119895 (

119905) + Δ119877119894119895 (119905)

Δ119877119894119895 (119905)

=

(119899119894119895+ Δ119899119894119895) times sum

(119899119894119895+Δ119899119894119895)

119888=1Δ119863119888

119894119895

sum

(119899119894119895+Δ119899119894119895)

119888=1Δ119868119888

119894119895

minus

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

Δ119899119894119895gt 0 forallΔ119868

119888

119894119895lt 119896

minus120588 times

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

Δ119899119894119895= 0 Δ119905 gt 119896 forallΔ119868

119888

119894119895lt 119896

0 otherwise(16)



119891 and



119891 and


equation

119901119889

119877119904119891=

1 119895 = 119889


119894 119873119889)

[119877119891119889 (119905)]

120572

sdot [120578119891119889 (119905)]

120573

sum119873119891isinallowd119889 119877

120572

119895119889sdot 120578120573

119895119889


119889

(17)


allowd119905119889= 119865 119873

119891| (119877119891119889 (119905) gt 119877119894119889 (

119905))

allowd119905119889isin 119865 119873

119891 allowd

119889= allowd119905

119889minus tabu

119889

(18)




119891119889 which

is given by

120578119891119889=

1

119871119891119889

(19)

where 119871119891119889


119889will













Input 119873119894119873119895

Output 119877119894119895(119879)


119894119895(119888 119905119904 119905119890)


119894119895(0) = 0 119896 = 119871 119891119897119886119892 = 1

(5) forall 119905 = 0 119879(6) if 119873


119895

(7) 119899119894119895= ++119888

(8) for all 119888 = 1 119899119894119895

(9) Δ119863119888

119894119895[119888] = 119905

119890minus 119905119904

(10) 119863119888

119894119895+= Δ119863119888

119894119895

(11) Δ119868119888

119894119895[119888] = 119905

119904minus 119905

(12) 119868119888

119894119895+= Δ119868119888

119894119895

(13) 119905 = 119905119890

(14) if (Δ119868119888119894119895[119888] gt 119896)


119894119895(119879) = ((119899

119894119895times 119863119888

119894119895) 119868119888

119894119895) times (1 minus 120588)

119890

(23) else(24) return 119877

119894119895(119879) = (119899

119894119895times 119863119888

119894119895) 119868119888

119894119895

(25) end if



+ state space 119864 = 119883

119895 119895 = 1 2 119898119898 isin 119873

+ time

slice set 119879 = 119905119899 119899 = 0 1 2 119897 119897 isin 119873



(3) LOPSI Sec(119873119904)

(4) for forall119873119891isin 1198651015840119873119891



Algorithm 5 LOPSI(119873119904 119873119889)




119891and 119873






119904will




1198621015840sube 119862 119873

119891 119862

1015840119873119891 = argmax119891leCOPY

119875119889

119877119904119891 (20)






119889and 119873

119891isin




119901119889

119904119891= 120574 lowast 119901

119889

119877119904119891+ 120575 lowast 119901

119889

119871119904119891 120574 + 120575 = 1 (21)

where 119901119889119877119904119891











2)

Storage space 119874(1198732) 119874(119873

3)


119894)










119891

(2) for forall119873119891isin 119862119873

119891

(3) update intimacy(119873119894 119873119891)



119889

lowast

(5) if 119873119889isin 119862119873

119891


119889

(7) tabu119889larr 119873

119894 119873119889


119904sends REQ(119862119873

119891119873119889) to the server


(12) 119873119904receives SEI

(13) 119873119904sends119873

119889to the nodes in 119865119873

119891

(14) for forall119873119891isin 119865119873

119891

(15) Send intimacy(119873119891 119873119889) to119873

119904

(16) end for(17) 119873

119904calculates 119901119889

119904119891according to (21)


1015840119873119891 = 119865 119873

119891

(20) else(21) 119865

1015840119873119891 = argmax

119891leCOPY (119901119889

119904119891)

(22) end if(23) 119873

119894sends data to 1198651015840119873

119891

(24) tabu119889larr 119873

119894 1198651015840119873119891


(26) end if



Parameter Value

Scene features


Node features










119867 =

119873

119884

(22)


119877 =

119878

119884

(23)


119871 =

1

119884

119884

sum

119894=1

(119879119889119894minus 119879119904119894) (24)


originated and 119879119889119894



119862 =

119876

119875

(25)








LOPDADLOPSI

100 150 200 250 300

The number of nodes

50

45

40

35

30

25

20

Aver

age h

ops

(a)

02

03

04

05

06

07

Deli

very

ratio

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(b)

2500

3000

3500

4000

4500

5000

5500

6000

Aver

age l

aten

cy

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(c)

30

35

40

45

50

55

60

Tran

smiss

ion

cost


ACODADLOPDADLOPSI

Spray and Wait

(d)






55

50

45

40

35

30

25

20

Aver

age h

ops

4 5 6 7 8 9 10 11 12 13TTL (h)

ACODAD

LOPDADLOPSI


(a)

4 5 6 7 8 9 10 11 12 13TTL (h)

07

06

05

04

03

02

Del

iver

y ra

tio

ACODAD

LOPDADLOPSI


(b)

20003000400050006000700080009000

10000110001200013000

Aver

age l

aten

cy (s

)

4 5 6 7 8 9 10 11 12 13

ACODAD

LOPDADLOPSI


(c)

25

30

35

40

45

50

55

60

Tran

smiss

ion

cost

4 6 8 10 12 14TTL (h)

ACODADLOPDADLOPSI

Spray and Wait

(d)













Acknowledgment


References













































Volume 2014




Journal of











Journal of


Function Spaces






Algebra

















119891 of 119873



119894and 119873

119895depends



119888

119894119895



119894and119873


119877119894119895 (119905)

=

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

forallΔ119868119888

119894119895lt 119896

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

times (1 minus 120588)119890

existΔ119868119888

119894119895gt 119896

0 Δ119868119888


(15)




119904and the






119877119894119895 (119905 + Δ119905) = 119877119894119895 (

119905) + Δ119877119894119895 (119905)

Δ119877119894119895 (119905)

=

(119899119894119895+ Δ119899119894119895) times sum

(119899119894119895+Δ119899119894119895)

119888=1Δ119863119888

119894119895

sum

(119899119894119895+Δ119899119894119895)

119888=1Δ119868119888

119894119895

minus

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

Δ119899119894119895gt 0 forallΔ119868

119888

119894119895lt 119896

minus120588 times

119899119894119895times sum

119899119894119895

119888=1Δ119863119888

119894119895

sum

119899119894119895

119888=1Δ119868119888

119894119895

Δ119899119894119895= 0 Δ119905 gt 119896 forallΔ119868

119888

119894119895lt 119896

0 otherwise(16)



119891 and



119891 and


equation

119901119889

119877119904119891=

1 119895 = 119889


119894 119873119889)

[119877119891119889 (119905)]

120572

sdot [120578119891119889 (119905)]

120573

sum119873119891isinallowd119889 119877

120572

119895119889sdot 120578120573

119895119889


119889

(17)


allowd119905119889= 119865 119873

119891| (119877119891119889 (119905) gt 119877119894119889 (

119905))

allowd119905119889isin 119865 119873

119891 allowd

119889= allowd119905

119889minus tabu

119889

(18)




119891119889 which

is given by

120578119891119889=

1

119871119891119889

(19)

where 119871119891119889


119889will













Input 119873119894119873119895

Output 119877119894119895(119879)


119894119895(119888 119905119904 119905119890)


119894119895(0) = 0 119896 = 119871 119891119897119886119892 = 1

(5) forall 119905 = 0 119879(6) if 119873


119895

(7) 119899119894119895= ++119888

(8) for all 119888 = 1 119899119894119895

(9) Δ119863119888

119894119895[119888] = 119905

119890minus 119905119904

(10) 119863119888

119894119895+= Δ119863119888

119894119895

(11) Δ119868119888

119894119895[119888] = 119905

119904minus 119905

(12) 119868119888

119894119895+= Δ119868119888

119894119895

(13) 119905 = 119905119890

(14) if (Δ119868119888119894119895[119888] gt 119896)


119894119895(119879) = ((119899

119894119895times 119863119888

119894119895) 119868119888

119894119895) times (1 minus 120588)

119890

(23) else(24) return 119877

119894119895(119879) = (119899

119894119895times 119863119888

119894119895) 119868119888

119894119895

(25) end if



+ state space 119864 = 119883

119895 119895 = 1 2 119898119898 isin 119873

+ time

slice set 119879 = 119905119899 119899 = 0 1 2 119897 119897 isin 119873



(3) LOPSI Sec(119873119904)

(4) for forall119873119891isin 1198651015840119873119891



Algorithm 5 LOPSI(119873119904 119873119889)




119891and 119873






119904will




1198621015840sube 119862 119873

119891 119862

1015840119873119891 = argmax119891leCOPY

119875119889

119877119904119891 (20)






119889and 119873

119891isin




119901119889

119904119891= 120574 lowast 119901

119889

119877119904119891+ 120575 lowast 119901

119889

119871119904119891 120574 + 120575 = 1 (21)

where 119901119889119877119904119891











2)

Storage space 119874(1198732) 119874(119873

3)


119894)










119891

(2) for forall119873119891isin 119862119873

119891

(3) update intimacy(119873119894 119873119891)



119889

lowast

(5) if 119873119889isin 119862119873

119891


119889

(7) tabu119889larr 119873

119894 119873119889


119904sends REQ(119862119873

119891119873119889) to the server


(12) 119873119904receives SEI

(13) 119873119904sends119873

119889to the nodes in 119865119873

119891

(14) for forall119873119891isin 119865119873

119891

(15) Send intimacy(119873119891 119873119889) to119873

119904

(16) end for(17) 119873

119904calculates 119901119889

119904119891according to (21)


1015840119873119891 = 119865 119873

119891

(20) else(21) 119865

1015840119873119891 = argmax

119891leCOPY (119901119889

119904119891)

(22) end if(23) 119873

119894sends data to 1198651015840119873

119891

(24) tabu119889larr 119873

119894 1198651015840119873119891


(26) end if



Parameter Value

Scene features


Node features










119867 =

119873

119884

(22)


119877 =

119878

119884

(23)


119871 =

1

119884

119884

sum

119894=1

(119879119889119894minus 119879119904119894) (24)


originated and 119879119889119894



119862 =

119876

119875

(25)








LOPDADLOPSI

100 150 200 250 300

The number of nodes

50

45

40

35

30

25

20

Aver

age h

ops

(a)

02

03

04

05

06

07

Deli

very

ratio

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(b)

2500

3000

3500

4000

4500

5000

5500

6000

Aver

age l

aten

cy

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(c)

30

35

40

45

50

55

60

Tran

smiss

ion

cost


ACODADLOPDADLOPSI

Spray and Wait

(d)






55

50

45

40

35

30

25

20

Aver

age h

ops

4 5 6 7 8 9 10 11 12 13TTL (h)

ACODAD

LOPDADLOPSI


(a)

4 5 6 7 8 9 10 11 12 13TTL (h)

07

06

05

04

03

02

Del

iver

y ra

tio

ACODAD

LOPDADLOPSI


(b)

20003000400050006000700080009000

10000110001200013000

Aver

age l

aten

cy (s

)

4 5 6 7 8 9 10 11 12 13

ACODAD

LOPDADLOPSI


(c)

25

30

35

40

45

50

55

60

Tran

smiss

ion

cost

4 6 8 10 12 14TTL (h)

ACODADLOPDADLOPSI

Spray and Wait

(d)













Acknowledgment


References













































Volume 2014




Journal of











Journal of


Function Spaces






Algebra



















Input 119873119894119873119895

Output 119877119894119895(119879)


119894119895(119888 119905119904 119905119890)


119894119895(0) = 0 119896 = 119871 119891119897119886119892 = 1

(5) forall 119905 = 0 119879(6) if 119873


119895

(7) 119899119894119895= ++119888

(8) for all 119888 = 1 119899119894119895

(9) Δ119863119888

119894119895[119888] = 119905

119890minus 119905119904

(10) 119863119888

119894119895+= Δ119863119888

119894119895

(11) Δ119868119888

119894119895[119888] = 119905

119904minus 119905

(12) 119868119888

119894119895+= Δ119868119888

119894119895

(13) 119905 = 119905119890

(14) if (Δ119868119888119894119895[119888] gt 119896)


119894119895(119879) = ((119899

119894119895times 119863119888

119894119895) 119868119888

119894119895) times (1 minus 120588)

119890

(23) else(24) return 119877

119894119895(119879) = (119899

119894119895times 119863119888

119894119895) 119868119888

119894119895

(25) end if



+ state space 119864 = 119883

119895 119895 = 1 2 119898119898 isin 119873

+ time

slice set 119879 = 119905119899 119899 = 0 1 2 119897 119897 isin 119873



(3) LOPSI Sec(119873119904)

(4) for forall119873119891isin 1198651015840119873119891



Algorithm 5 LOPSI(119873119904 119873119889)




119891and 119873






119904will




1198621015840sube 119862 119873

119891 119862

1015840119873119891 = argmax119891leCOPY

119875119889

119877119904119891 (20)






119889and 119873

119891isin




119901119889

119904119891= 120574 lowast 119901

119889

119877119904119891+ 120575 lowast 119901

119889

119871119904119891 120574 + 120575 = 1 (21)

where 119901119889119877119904119891











2)

Storage space 119874(1198732) 119874(119873

3)


119894)










119891

(2) for forall119873119891isin 119862119873

119891

(3) update intimacy(119873119894 119873119891)



119889

lowast

(5) if 119873119889isin 119862119873

119891


119889

(7) tabu119889larr 119873

119894 119873119889


119904sends REQ(119862119873

119891119873119889) to the server


(12) 119873119904receives SEI

(13) 119873119904sends119873

119889to the nodes in 119865119873

119891

(14) for forall119873119891isin 119865119873

119891

(15) Send intimacy(119873119891 119873119889) to119873

119904

(16) end for(17) 119873

119904calculates 119901119889

119904119891according to (21)


1015840119873119891 = 119865 119873

119891

(20) else(21) 119865

1015840119873119891 = argmax

119891leCOPY (119901119889

119904119891)

(22) end if(23) 119873

119894sends data to 1198651015840119873

119891

(24) tabu119889larr 119873

119894 1198651015840119873119891


(26) end if



Parameter Value

Scene features


Node features










119867 =

119873

119884

(22)


119877 =

119878

119884

(23)


119871 =

1

119884

119884

sum

119894=1

(119879119889119894minus 119879119904119894) (24)


originated and 119879119889119894



119862 =

119876

119875

(25)








LOPDADLOPSI

100 150 200 250 300

The number of nodes

50

45

40

35

30

25

20

Aver

age h

ops

(a)

02

03

04

05

06

07

Deli

very

ratio

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(b)

2500

3000

3500

4000

4500

5000

5500

6000

Aver

age l

aten

cy

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(c)

30

35

40

45

50

55

60

Tran

smiss

ion

cost


ACODADLOPDADLOPSI

Spray and Wait

(d)






55

50

45

40

35

30

25

20

Aver

age h

ops

4 5 6 7 8 9 10 11 12 13TTL (h)

ACODAD

LOPDADLOPSI


(a)

4 5 6 7 8 9 10 11 12 13TTL (h)

07

06

05

04

03

02

Del

iver

y ra

tio

ACODAD

LOPDADLOPSI


(b)

20003000400050006000700080009000

10000110001200013000

Aver

age l

aten

cy (s

)

4 5 6 7 8 9 10 11 12 13

ACODAD

LOPDADLOPSI


(c)

25

30

35

40

45

50

55

60

Tran

smiss

ion

cost

4 6 8 10 12 14TTL (h)

ACODADLOPDADLOPSI

Spray and Wait

(d)













Acknowledgment


References













































Volume 2014




Journal of











Journal of


Function Spaces






Algebra












119891and 119873






119904will




1198621015840sube 119862 119873

119891 119862

1015840119873119891 = argmax119891leCOPY

119875119889

119877119904119891 (20)






119889and 119873

119891isin




119901119889

119904119891= 120574 lowast 119901

119889

119877119904119891+ 120575 lowast 119901

119889

119871119904119891 120574 + 120575 = 1 (21)

where 119901119889119877119904119891











2)

Storage space 119874(1198732) 119874(119873

3)


119894)










119891

(2) for forall119873119891isin 119862119873

119891

(3) update intimacy(119873119894 119873119891)



119889

lowast

(5) if 119873119889isin 119862119873

119891


119889

(7) tabu119889larr 119873

119894 119873119889


119904sends REQ(119862119873

119891119873119889) to the server


(12) 119873119904receives SEI

(13) 119873119904sends119873

119889to the nodes in 119865119873

119891

(14) for forall119873119891isin 119865119873

119891

(15) Send intimacy(119873119891 119873119889) to119873

119904

(16) end for(17) 119873

119904calculates 119901119889

119904119891according to (21)


1015840119873119891 = 119865 119873

119891

(20) else(21) 119865

1015840119873119891 = argmax

119891leCOPY (119901119889

119904119891)

(22) end if(23) 119873

119894sends data to 1198651015840119873

119891

(24) tabu119889larr 119873

119894 1198651015840119873119891


(26) end if



Parameter Value

Scene features


Node features










119867 =

119873

119884

(22)


119877 =

119878

119884

(23)


119871 =

1

119884

119884

sum

119894=1

(119879119889119894minus 119879119904119894) (24)


originated and 119879119889119894



119862 =

119876

119875

(25)








LOPDADLOPSI

100 150 200 250 300

The number of nodes

50

45

40

35

30

25

20

Aver

age h

ops

(a)

02

03

04

05

06

07

Deli

very

ratio

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(b)

2500

3000

3500

4000

4500

5000

5500

6000

Aver

age l

aten

cy

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(c)

30

35

40

45

50

55

60

Tran

smiss

ion

cost


ACODADLOPDADLOPSI

Spray and Wait

(d)






55

50

45

40

35

30

25

20

Aver

age h

ops

4 5 6 7 8 9 10 11 12 13TTL (h)

ACODAD

LOPDADLOPSI


(a)

4 5 6 7 8 9 10 11 12 13TTL (h)

07

06

05

04

03

02

Del

iver

y ra

tio

ACODAD

LOPDADLOPSI


(b)

20003000400050006000700080009000

10000110001200013000

Aver

age l

aten

cy (s

)

4 5 6 7 8 9 10 11 12 13

ACODAD

LOPDADLOPSI


(c)

25

30

35

40

45

50

55

60

Tran

smiss

ion

cost

4 6 8 10 12 14TTL (h)

ACODADLOPDADLOPSI

Spray and Wait

(d)













Acknowledgment


References













































Volume 2014




Journal of











Journal of


Function Spaces






Algebra











119891

(2) for forall119873119891isin 119862119873

119891

(3) update intimacy(119873119894 119873119891)



119889

lowast

(5) if 119873119889isin 119862119873

119891


119889

(7) tabu119889larr 119873

119894 119873119889


119904sends REQ(119862119873

119891119873119889) to the server


(12) 119873119904receives SEI

(13) 119873119904sends119873

119889to the nodes in 119865119873

119891

(14) for forall119873119891isin 119865119873

119891

(15) Send intimacy(119873119891 119873119889) to119873

119904

(16) end for(17) 119873

119904calculates 119901119889

119904119891according to (21)


1015840119873119891 = 119865 119873

119891

(20) else(21) 119865

1015840119873119891 = argmax

119891leCOPY (119901119889

119904119891)

(22) end if(23) 119873

119894sends data to 1198651015840119873

119891

(24) tabu119889larr 119873

119894 1198651015840119873119891


(26) end if



Parameter Value

Scene features


Node features










119867 =

119873

119884

(22)


119877 =

119878

119884

(23)


119871 =

1

119884

119884

sum

119894=1

(119879119889119894minus 119879119904119894) (24)


originated and 119879119889119894



119862 =

119876

119875

(25)








LOPDADLOPSI

100 150 200 250 300

The number of nodes

50

45

40

35

30

25

20

Aver

age h

ops

(a)

02

03

04

05

06

07

Deli

very

ratio

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(b)

2500

3000

3500

4000

4500

5000

5500

6000

Aver

age l

aten

cy

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(c)

30

35

40

45

50

55

60

Tran

smiss

ion

cost


ACODADLOPDADLOPSI

Spray and Wait

(d)






55

50

45

40

35

30

25

20

Aver

age h

ops

4 5 6 7 8 9 10 11 12 13TTL (h)

ACODAD

LOPDADLOPSI


(a)

4 5 6 7 8 9 10 11 12 13TTL (h)

07

06

05

04

03

02

Del

iver

y ra

tio

ACODAD

LOPDADLOPSI


(b)

20003000400050006000700080009000

10000110001200013000

Aver

age l

aten

cy (s

)

4 5 6 7 8 9 10 11 12 13

ACODAD

LOPDADLOPSI


(c)

25

30

35

40

45

50

55

60

Tran

smiss

ion

cost

4 6 8 10 12 14TTL (h)

ACODADLOPDADLOPSI

Spray and Wait

(d)













Acknowledgment


References













































Volume 2014




Journal of











Journal of


Function Spaces






Algebra














119867 =

119873

119884

(22)


119877 =

119878

119884

(23)


119871 =

1

119884

119884

sum

119894=1

(119879119889119894minus 119879119904119894) (24)


originated and 119879119889119894



119862 =

119876

119875

(25)








LOPDADLOPSI

100 150 200 250 300

The number of nodes

50

45

40

35

30

25

20

Aver

age h

ops

(a)

02

03

04

05

06

07

Deli

very

ratio

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(b)

2500

3000

3500

4000

4500

5000

5500

6000

Aver

age l

aten

cy

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(c)

30

35

40

45

50

55

60

Tran

smiss

ion

cost


ACODADLOPDADLOPSI

Spray and Wait

(d)






55

50

45

40

35

30

25

20

Aver

age h

ops

4 5 6 7 8 9 10 11 12 13TTL (h)

ACODAD

LOPDADLOPSI


(a)

4 5 6 7 8 9 10 11 12 13TTL (h)

07

06

05

04

03

02

Del

iver

y ra

tio

ACODAD

LOPDADLOPSI


(b)

20003000400050006000700080009000

10000110001200013000

Aver

age l

aten

cy (s

)

4 5 6 7 8 9 10 11 12 13

ACODAD

LOPDADLOPSI


(c)

25

30

35

40

45

50

55

60

Tran

smiss

ion

cost

4 6 8 10 12 14TTL (h)

ACODADLOPDADLOPSI

Spray and Wait

(d)













Acknowledgment


References













































Volume 2014




Journal of











Journal of


Function Spaces






Algebra











LOPDADLOPSI

100 150 200 250 300

The number of nodes

50

45

40

35

30

25

20

Aver

age h

ops

(a)

02

03

04

05

06

07

Deli

very

ratio

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(b)

2500

3000

3500

4000

4500

5000

5500

6000

Aver

age l

aten

cy

100 150 200 250 300

The number of nodes

ACODAD

LOPDADLOPSI


(c)

30

35

40

45

50

55

60

Tran

smiss

ion

cost


ACODADLOPDADLOPSI

Spray and Wait

(d)






55

50

45

40

35

30

25

20

Aver

age h

ops

4 5 6 7 8 9 10 11 12 13TTL (h)

ACODAD

LOPDADLOPSI


(a)

4 5 6 7 8 9 10 11 12 13TTL (h)

07

06

05

04

03

02

Del

iver

y ra

tio

ACODAD

LOPDADLOPSI


(b)

20003000400050006000700080009000

10000110001200013000

Aver

age l

aten

cy (s

)

4 5 6 7 8 9 10 11 12 13

ACODAD

LOPDADLOPSI


(c)

25

30

35

40

45

50

55

60

Tran

smiss

ion

cost

4 6 8 10 12 14TTL (h)

ACODADLOPDADLOPSI

Spray and Wait

(d)













Acknowledgment


References













































Volume 2014




Journal of











Journal of


Function Spaces






Algebra










55

50

45

40

35

30

25

20

Aver

age h

ops

4 5 6 7 8 9 10 11 12 13TTL (h)

ACODAD

LOPDADLOPSI


(a)

4 5 6 7 8 9 10 11 12 13TTL (h)

07

06

05

04

03

02

Del

iver

y ra

tio

ACODAD

LOPDADLOPSI


(b)

20003000400050006000700080009000

10000110001200013000

Aver

age l

aten

cy (s

)

4 5 6 7 8 9 10 11 12 13

ACODAD

LOPDADLOPSI


(c)

25

30

35

40

45

50

55

60

Tran

smiss

ion

cost

4 6 8 10 12 14TTL (h)

ACODADLOPDADLOPSI

Spray and Wait

(d)













Acknowledgment


References













































Volume 2014




Journal of











Journal of


Function Spaces






Algebra
















Acknowledgment


References













































Volume 2014




Journal of











Journal of


Function Spaces






Algebra




































Volume 2014




Journal of











Journal of


Function Spaces






Algebra
















Volume 2014




Journal of











Journal of


Function Spaces






Algebra









Documents

Research Article Location Prediction-Based Data ...downloads.hindawi.com/journals/mpe/2014/453564.pdf · Research Article Location Prediction-Based Data Dissemination Using Swarm