Research Article Adaptive MANET Multipath Routing ...downloads.hindawi.com/journals/tswj/2014/872526.pdfResearch Article Adaptive MANET Multipath Routing Algorithm Based on the Simulated

Research ArticleAdaptive MANET Multipath Routing Algorithm Based onthe Simulated Annealing Approach

Sungwook Kim

Department of Computer Science Sogang University Sinsu dong 1 Mapo-ku Seoul 121-742 Republic of Korea

Correspondence should be addressed to Sungwook Kim swkim01sogangackr

Received 11 April 2014 Revised 17 May 2014 Accepted 24 May 2014 Published 16 June 2014

Academic Editor T O Ting

Copyright copy 2014 Sungwook Kim This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Mobile ad hoc network represents a system of wireless mobile nodes that can freely and dynamically self-organize networktopologies without any preexisting communication infrastructure Due to characteristics like temporary topology and absenceof centralized authority routing is one of the major issues in ad hoc networks In this paper a new multipath routing scheme isproposed by employing simulated annealing approach The proposed metaheuristic approach can achieve greater and reciprocaladvantages in a hostile dynamic real world network situation Therefore the proposed routing scheme is a powerful method forfinding an effective solution into the conflict mobile ad hoc network routing problem Simulation results indicate that the proposedparadigm adapts best to the variation of dynamic network situations The average remaining energy network throughput packetloss probability and traffic load distribution are improved by about 10 10 5 and 10 respectively more than the existingschemes

1 Introduction

Due to the explosive growth of wireless communicationtechnology mobile ad hoc networks (MANETs) have beenused in many practical applications in the commercialmilitary and private sectors MANETs are self-creatingself-organizing and autonomous systems of mobile hostsconnected by wireless links with no static infrastructure suchas base station When making such networks operational akey question is how to effectively decide routing paths giventhe dynamic nature of the system and the limited knowledgeof the network topology In recent times a lot of attentionhas been attracted to designing efficient routing protocols forefficient MANET operations [1ndash4]

During the operation of MANETs unexpected growth oftraffic may develop in a specific routing path it may createlocal traffic congestion In order to alleviate this kind oftraffic overload condition load balancing strategy should beemployed In MANETs the meaning of load balancing is toease out the heavy traffic load in a specific path which canensure the balanced network resource assumption To ensurethe load balancing multipath routing algorithms have beendeveloped Multipath routing algorithm establishes multiplepaths between a source and a destination node and spreads

the traffic load along multiple routes It can alleviate trafficcongestion in a specific path Therefore multipath routingalgorithms can provide the route resiliencewhile ensuring thereliability of data transmission [5 6]

Nowadays metaheuristic approach is widely recognizedas a practical perspective to be implemented for real worldnetwork operations [7ndash9] Traditionally metaheuristic algo-rithms try to improve a candidate solution iteratively withregard to a given measure of quality Even though thisapproach does not guarantee an optimal solution it canbe widely applied to various network control problemsSimulated annealing is a well-known probabilistic meta-heuristic algorithm for finding an effective solution [10ndash12]To adaptively make a routing decision the basic concept ofsimulated annealing approach can be adopted

Motivated by the facts presented in the above discussiona new multipath routing scheme is proposed based on thesimulated annealing approach In this work we do not focuson trying to get an optimal solution itself but insteadan adaptive online feedback model is adopted Thereforethe proposed scheme repeatedly estimates the current net-work situations and dynamically makes a control decisionThis approach can significantly reduce the computational

Hindawi Publishing Corporatione Scientific World JournalVolume 2014 Article ID 872526 8 pageshttpdxdoiorg1011552014872526

2 The Scientific World Journal

complexity and overheads Due to this reason wireless nodesare assumed to be self-interested agents and make localdecisions in a distributed manner Therefore routing packetsare adaptively distributed through multiple paths in pursuitof the main goals such as load balancing and networkreliability Under diverse network environment changes theproposed scheme tries to approximate an optimal networkperformanceThe important features of our proposed schemeare (i) interactive process to get an efficient network per-formance (ii) distributed approach for large-scale networkoperations (iii) dynamic adaptability to current network and(iv) feasibility for practical implementation

11 Related Work Recently several routing schemes for adhoc networks have been presented in research literatureLeu et al developed the multicast power greedy clustering(MPGC) scheme [13] which is an adaptive power-awareand on-demand multicasting algorithm The MPGC schemeuses greedy heuristic clustering power-aware multicastingand clusteringmaintenance techniques thatmaximize energyefficiency and prolong network lifetime

To improve the network reliability and reduce the net-work traffic Kim et al propose the double-layered effectiverouting (DLER) scheme for peer-to-peer network systems[14] This scheme first chooses the shortest routing pathsamong possible routing paths and selects the path associatedwith the relay peer who has lower mobility to improve thereliability of the system Therefore in the DLER scheme thelower mobility of relay peers contributes to both the stabilityof clusters and the robustness of the system

Hieu and Hong proposed the entropy-based multiraterouting (EMRR) scheme [15] This scheme introduces a newapproach to modeling relative distance among nodes undera variety of communication rates due to nodersquos mobility inMANETs When mobile nodes move to another location therelative distance between communicating nodes will directlyaffect the data rate of transmission Therefore the stability ofa route is related to connection entropy Taking into accountthese issues the link weight and route stability based onconnection entropy are considered as a new routing metricIn the EMRR scheme the problem of determining the bestroute is formulated as the minimization of an object functionformed as a linear combination of the link weight and theconnection uncertainty of that link

The ant-colony based routing algorithm (ARA) schemewas proposed [16] in this scheme swarm intelligence andant-colony metaheuristic techniques are used This schemeconsists of three phases route discovery route maintenanceand route failure handling In the route discovery phasenew routes between nodes are discovered using forwardand backward ants Routes are maintained by subsequentdata packets that is as the data traverse the network nodepheromone values are modified to reinforce the routes

Wang et al developed the logical hypercube-based virtualdynamic backbone (HVDB) scheme for an 119899-dimensionalhypercube in a large-scale MANET [17] The HVDB schemeis a proactive QoS-aware and hybrid multicast routingprotocol Owing to the regularity symmetry properties

and small diameter of the hypercube every node playsalmost the same role In addition no single node is moreloaded than any other node and bottlenecks do not existunlike the case in tree-based architectures In particularthe HVDB scheme can satisfy the new QoS requirementsmdashhigh availability and good load balancingmdashby using locationinformation

Barolli et al proposed the genetic algorithmbased routing(GAR) scheme for mobile ad hoc networks [18] In the GARscheme only a small number of nodes are involved in routecomputation because a small population size is used As aresult routing information is transmitted only for the nodespresent in that population Different routes are ranked bysorting the first route is the best one and the remainingroutes are used as backup routes Because a tree-based geneticalgorithm method is used in the GAR scheme the delay andtransmission success rate are considered as QoS parametersin this scheme

The incentive-based repeated routing (IRR) scheme in[19] is an incentive-based routing model that captures thenotion of repetition To provide a desirable solution the IRRscheme examines certain fundamental properties to governthe behavior of autonomous agents The distributed routingmechanism (DRM) scheme in [20] is an adaptive and scalablerouting scheme for wireless ad hoc networks This schemeprovides a cost-efficient routing mechanism for strategicagents In addition theDRMscheme is designed tomaximizethe benefit of each agent

The proactive congestion reduction (PCR) scheme in [5]focuses on adaptive routing strategies to help congestionreduction Based on a nonlinear optimization method formultipath routings the PCR scheme calculates a traffic split-ting vector that determines a near-optimal traffic distributionover routing paths The shortest multipath source (SMS)scheme [6] is one of the most generally accepted on-demanddynamic routing schemes that build multiple shortest partialdisjoint pathsThe SMS scheme uses node-disjoint secondarypaths to exploit fault tolerance load balancing and band-width aggregation All the earlier work has attracted a lot ofattention and introduced unique challenges However theseexisting schemes have several shortcomings as described inSection 3 Compared to the PCR scheme and the SMS scheme[5 6] the proposed scheme attains better performance forwireless network managements

This paper is organized as follows Section 2 describesthe proposed algorithms in detail In Section 3 performanceevaluation results are presented along with comparisons withthe schemes proposed in [5 6] Finally in Section 4 conclud-ing remarks are given and some directions are identified forfuture work

2 Proposed MANET Routing Algorithms

Multipath routing algorithms are designed to split andtransmit the traffic load through two or more different pathsto a destination simultaneously In this paper we propose anew multipath routing scheme to balance the network loadwhile ensuring efficient network performance

The Scientific World Journal 3

21 Path Setup Algorithm Usually wireless link capacitycontinually varies because of the impacts from transmissionpower interference and so forthTherefore it is important toestimate the current link status by considering several controlfactors To configure the adaptive multihop routing path theproposed algorithm defines a link cost (119871 119875) for each link toestimate the degree of communication adaptability [21 22]In order to relatively handle dynamic network conditions the119871 119875 value from the node 119894 to the node 119895 is obtained as

119871 119875119894119895= [(1 minus 120572) timesC119894119895 + 120572 times (1 minus Θ119895 (119905))]

+ [120596 times (1 minus Ψ119894119895 (119905))]

st 120572 =119864119894

119864119872

C119894119895=

119889119894119895

119863119872

Ψ119894119895 (119905) =

120581119894119895

119905

(120581119894119895

119905+ 120599119894119895

119905)

(1)

where 119889119894119895is the distance from the node 119894 to the node 119895 and

119864119894is the remaining energy of the node 119894 119864

119872and 119863

119872are

the initial energy and the maximum coverage range of eachnodeTherefore the 119889

119894119895and 119864

119894are normalized by the119863

119872and

119864119872 the range is varied from 0 to 1 Θ

119895(119905) is the entropy for

the node 119895 at the time (119905) Usually entropy is the uncertaintyand a measure of the disorder in a system It represents thetopological change which is a natural quantification of theeffect of node mobility on MANETrsquos connectivity service[23] In this work the basic concept of entropy is adoptedfor supporting and evaluating stable routing routes For themobile node 119895 the entropyΘ

119895(119905) is calculated as follows [23]

Θ119895 (119905) =

minussum119896isin119865119895119875119896(119905 Δ119905) times log119875

119896(119905 Δ119905)

log119862 (119865119895)

st 119875119896(119905 Δ119905) =

119886119895119896

sum119894isin119865119895119886119895119894

(2)

where Δ119905is a time interval 119865

119895denotes the set of the

neighboring nodes of node 119895 and 119862(119865119895) is the cardinality

(degree) of set 119865119895 To estimate the stability of a part of a

specific route 119886119895119894represents ameasure of the relativemobility

among two nodes 119895 and 119894 as

119886119895119894=

1

119868 119879

times

119868 119879

sum

119897=1

1003816100381610038161003816V (119895 119894 119905

119897)1003816100381610038161003816

st V (119895 119894 119905) = V (119895 119905) minus V (119894 119905)

(3)

where V(119895 119905) and V(119894 119905) are the velocity vectors of node 119895 andnode 119894 at time 119905 respectively 119868 119879 is the number of discretetimes 119905

119897that mobility information can be calculated and

disseminated to other neighboring nodeswithin time intervalΔ119905 V(119895 119894 119905) is the relative velocity between nodes 119895 and 119894 at

time 119905 Any change can be described as a change of variablevalues 119886

119895119894in the course of time 119905 such as 119886

119895119894(119905) rarr 119886

119895119894(119905+Δ119905)

The entropy Θ119895(119905) is normalized as 0 le Θ

119895(119905) le 1 If Θ

119895(119905)

value is close to 1 the part of the route that represents the linksof the path associated with an intermediate node 119895 is stableIf Θ119895(119905) value is close to 0 the local route is unstable [23] In

(1) Ψ119894119895(119905) is the link 119894119895rsquos trust value at the time 119905 After the 119905th

iteration Ψ119894119895(119905) is using the number of packets successfully

serviced in the link 119894119895 (120581119894119895119905) divided by the total number of

packets that have been sent from the node 119894 to the relay node119895 (120581119894119895119905+ 120599119894119895

119905)

To relatively estimate the current link situation by using(1) the control parameters 120572 and 120596 should be adjusteddynamically The C

119894119895reflects the cost of the wireless com-

munication the closer a next node the more attractive forrouting due to the less communication cost The 119864

119894is the

current residual energy of node 119894 which reflects the remain-ing lifetime of a wireless node Due to the characteristicsof wireless propagation the energy consumption rate forwireless communications is strongly related to the internodedistance The parameter 120572 controls the relative weights givento distance and entropy of corresponding relay node Underdiverse network environments a fixed value of 120572 cannoteffectively adapt to the changing conditions [21 22] In thispaper we treat it as an online decision problem and adaptivelymodify 120572 value When the remaining energy of the node 119894is high we can put more emphasis on the stability status ofnext node 119895 that is on (1 minus Θ

119895(119905)) In this case a higher

value of 120572 is more suitable When the remaining energy ofthe node 119894 is not enough due to traffic overhead the pathselection should strongly depend on the energy dissipationfor data transmission In this case a lower value of 120572 is moresuitable for the energy consumption rate that is onC

119894119895 since

the distance between two neighbor nodes directly affectsthe energy consumption rate In the proposed algorithmthe value of 120572 of the node 119894 is dynamically adjusted basedon the current rate of its remaining energy per initiallyassigned energy (119864

119894119864119872) Therefore the system can be more

responsive to current network conditions by the real-timenetwork monitoring The parameter 120596 is an impact factor toevaluate the trust level of the link In this paper to avoid thedetrimental packet loss effect each linkrsquos trust level is fullyconsidered to estimate 119871 119875 value the 120596 value is fixed as 1

The 119871 119875 value can represent the normalized commu-nication cost of each link With the 119871 119875 value we definethe path cost parameter (PC) to calculate total routing pathcost PC is computed as the sum of all link costs from thesource node to the current node Based on the PC valuethe proposed routing algorithm constructs adaptivemultihoprouting paths to reach the destination node At the initialtime for routing operations the source node broadcasts itsinitial PC value (ie PC = 0) Within the power coveragearea message receiving relay nodes individually estimatethe link cost according to (1) and estimate its PC valueas PC + 119871 119875 Some nodes can receive multiple PC valuesfrom reachable different neighbor nodes For self-organizingand independent-effective controlling each node keeps thisinformation For example the node 119894 can have receivedmultiple PC values that is PC

1 PC119896 and PC

119873119894 where PC

119896

is the receiving PC value of the message-sending neighbornode 119896 (1 le 119896 le 119873

119894) and119873

119894is the number of total reachable

neighbor nodes In this case the node 119894 calculates its own PC119894

value as follows

PC119894= argmin

119896isin119873119894

(PC119896+ 119871 119875

119894119896) (4)


According to (4) the node 119894 adaptively selects one neighbornode as a relay node while minimizing PC

119894value which

potentially incorporates more global network informationThe estimated PC value is recursively forwarded to establishthe routing path This route formation process is repeateduntil all available multipaths from the source to the destina-tion node are configured

22 Simulated Annealing Routing Algorithm Generally mul-tipath routing algorithms face an essential challengemdashhowto distribute the volume of traffic to a specific path Inorder to produce good solutions within a reasonable amountof computer time the proposed scheme does not seek theoptimal allocation Based on feedbacks of the real-timetraffic measurements it is designed in a simple but efficientmetaheuristic algorithm

Simulated annealing (SA) is a well-known metaheuristicmethod that has been applied successfully to combinatorialoptimization problems [8] The term simulated annealingderives from the roughly analogous natural phenomena ofannealing of solids which is accomplished by heating up asolid and allowing it to cool down slowly so that thermalequilibrium is maintained Each step of the SA processreplaces the current solution by a random ldquonearbyrdquo solutionchosen with a probability that depends on the differencebetween the corresponding function values and on a globalparameter 119879 (called the temperature) The 119879 is graduallydecreased during the process to reach steady state or thermalequilibrium [8 10 12]

In the proposed algorithm the SA approach is used tosolve themultipath routing problemThe basic concept of theproposed algorithm is to proportionally load traffic on eachroute according to its adaptability To transmit packets eachnode selects a next relay node based on the PC informationFrom the point of view of the node 119894 selection probability ofthe neighbor node 119896 (SP

119896) is defined as follows

SP119896=

119879119862119896

sum119899

119895=1119879119862119895

where 119879119862119896= 1 minus

(PC119895+ 119871 119875

119894119895)

sum119899

119895=1(PC119895+ 119871 119875

119894119895)

119896 isin 119873119894

(5)

where 119899 is the total number of neighbor nodes Based on theroulette-wheel function [24] of SP values a next relay node istemporarily selected For example the probability of node 119896rsquosselection is SP

119896 Therefore we can make the more adaptable

nodesmore likely to be selected than the less adaptable nodesIn addition to avoid a local optimal solution the Boltzmannprobability (BP) is adopted The BP is defined as follows [8]

BP = exp(minus119873pc minus 119862pc

119879 (119905)

) (6)

where 119873pc is the SP value of new selected node and 119862pcis the SP value of previously connected node In (6) thedifference between 119873pc and 119862pc (ie 119873pc minus 119862pc) means thepath adaptability alteration 119879(119905) is a parameter to controlthe BP value Metaphorically it is the time 119905rsquos temperature of

the system As an annealing process the 119879(119905) is decreasedaccording to a cooling schedule At the beginning of theannealing algorithm run the initialization temperature ishigh enough so that possibility of accepting any decisionchanges whether it improves the solution or not While timeis ticking away the 119879(119905) value decreases until the stoppingcondition is met In this paper 119879(119905) value is set to the currentratio of the remaining packet amount to the total routingpacket amount

At the routing decision time there are two cases

(i) If the 119873pc value is higher than the 119862pc (ie 119873pc minus119862pc ge 0) the new selected neighbor node replaces thecurrent relay node

(ii) If the 119873pc value is less than the 119862pc value (ie 119873pc minus119862pc lt 0) the new selected neighbor node is noteligible to replace the current relay node Howeverthis node might still be accepted as a new relay nodeto potentially avoid local optima It is analogous to theuphill move acceptance to reach an optimal point Inthis case a random number 119883 is generated where 119883is in the range of 0 sdot sdot sdot 1

(a) If the 119883 is less than BP (ie 119883 lt BP) thenew selected neighbor node replaces the currentrelay node

(b) Otherwise the current routing route is notchanged

Based on the SA approach individual nodes in our proposedscheme locally make routing decisions to select next relaynodes In an entirely distributed fashion this hop-by-hoppath selection procedure is recursively repeated until thepacket reaches the destination nodeTherefore our proposedrouting algorithm can have the self-adaptability for networkdynamics

23 The Main Steps of MANET Routing Algorithm In thispaper we propose a new multipath routing algorithm forwireless mobile ad hoc networks In the proposed schemerouting is guided by employing a simulated annealing pro-cess Therefore self-interested ad hoc nodes make routingdecisions according to private preferences while adaptingthe current network situations To solve the dynamic anddistributed routing problem the main steps of the proposedmultipath routing algorithm are given next

Step 1 Each node dynamically estimates the 119889 C 119864 Θ(sdot)Ψ(sdot) and 120572 values based on the real-time measurement

Step 2 The 119871 119875 value is locally calculated according to (1)

Step 3 At the initial time for routing operations the sourcenode broadcasts the initial PC value to neighbor nodes Eachnode calculates its PC value by using (4) and recursivelyforwards this information

Step 4 Based on the PC value route configuration processcontinues repeatedly until all available multipaths from thesource to the destination node are configured


Table 1 Type of traffic and system parameters used in the simulation experiments

(a)

Traffic type Bandwidth requirement Connection duration (avesec)I 128 Kbps 60 sec (1min)II 256Kbps 120 sec (2min)III 512 Kbps 180 sec (3min)

(b)

Parameter Value Descriptionunit time 1 second Equal interval of time axis119890dis 1 pJbitm2 Energy dissipation coefficient for the packet transmission119864co 10 nJbit System parameter for the electronic digital coding energy dissipation119863119872

10m Maximum wireless coverage range of each node119864119872

10 joules Initial assigned energy amount of each node120596 1 The weighted factor for the trust levelI 119879 10 seconds The number of discrete times to estimate entropy119883 0sim1 Generated random number

(c)

Parameter Initial Description Values120572 1 The ratio of remaining to initial energy of node 0sim1 (119864

119894119864119872)

119879(t) 1 The ratio of remaining to initial packet amount at time t 0sim1

Step 5 To transmit packets each relay node temporarilyselects a next relay node with the selection probability whichis estimated according to (5)

Step 6 If the119873pc value is higher than the119862pc (ie119873pcminus119862pc gt0) the new selected neighbor node replaces the current relaynode proceed to Step 8 Otherwise go to Step 7

Step 7 When the 119873pc value is less than the 119862pc value (ie119873pc minus 119862pc lt 0) a random number 119883 is generated If agenerated119883 is less than the BP (ie119883 ltBP) the new selectedneighbor node replaces the current relay nodeOtherwise theestablished routing route is not changed

Step 8 In an entirely distributed fashion this hop-by-hoppath selection procedure is recursively repeated until thepacket reaches the destination node

3 Performance Evaluation

In this section the effectiveness of the proposed algorithms isvalidated through simulation we propose a simulationmodelfor the performance evaluation With a simulation study theperformance superiority of the proposed multipath routingscheme can be confirmed The assumptions implemented inour simulation model were as follows

(i) 100 nodes are distributed randomly over an area of500 times 500 meter square

(ii) Each data message is considered CBR traffic with thefixed packet size

(iii) Network performancemeasures obtained on the basisof 50 simulation runs are plotted as functions of thepacket generation per second (packetss)

(iv) Data packets are generated at the source according tothe rate 120582 (packetss) and the range of offered loadwas varied from 0 to 30

(v) The bandwidth of the wireless link was set to 5Mbsand the unit time is one second

(vi) The source and destination nodes are randomlyselected

(vii) For simplicity we assume the absence of noise orphysical obstacles in our experiments

(viii) The mobility of each mobile node is randomlyselected from the range of 0ndash10ms and mobilitymodel is random way point model

(ix) At the beginning of simulation all nodes started withan initial energy of 10 joules

(x) Three different traffic types were assumed they weregenerated with equal probability

Table 1 shows the traffic types and system parametersused in the simulation Each type of traffic has its ownrequirements in terms of bandwidth and service time Inorder to emulate a real wireless network and for a faircomparison we used the system parameters for a realisticsimulation model [21 22]

Recently the PCR scheme [5] and the SMS scheme [6]have been published and introduced unique challenges forthe issue of multipath routing in MANETs Even thoughthese existing schemes have presented novel multipath rout-ing algorithms there are several disadvantages First these


0 05 1 15 2 25 33

4

5

6

7

8

9

10

Offered load (packet generation rate)

Aver

age r

emai

ning

ener

gy

Proposed schemeThe PCR schemeThe SMS scheme

Figure 1 Average remaining energy

schemes cannot adaptively estimate the current networkconditions Therefore each node is unaware of effectiverouting paths to reach a destination Second some nodescarry a disproportionately large amount of the entire trafficdrastically decreasing the throughput of the flows theyforward Third the PCR and SMS schemes are based ona centralized approach The ideas for practical implemen-tations are left for future study As mentioned earlier wecompare the performance of the proposed scheme with theseexisting schemes to confirm the superiority of the proposedapproach In our simulation analysis of Figures 1ndash5 the 119909-axis (a horizontal line) marks the traffic intensities which isvaried from 0 to 30The 119910-axis (a vertical line) represents thenormalized value for each performance criterion

Figure 1 compares the performance of each scheme interms of the average remaining energy of wireless nodesTo maximize a network lifetime the remaining energy isan important performance metric All the schemes havesimilar trends However based on (1) the proposed schemeeffectively selects the next routing link by considering theremaining energy information Therefore we attain muchremaining energy under heavy traffic load intensities itguarantees a longer node lifetime

Figure 2 shows the performance comparison of networkthroughput Usually network throughput is the rate of suc-cessful message delivery over a communication channel Thethroughput is usually measured in bits per second (bits orbps) and sometimes in data packets per second or data pack-ets per time slot In this work network throughput is definedas the ratio of data amount received at the destination nodesto the total generated data amount For a fair comparison itis the best realistic way Due to the inclusion of the adaptiveonline approach the proposed scheme can have the bestthroughput gain

In Figure 3 the packet loss probabilities are presentedpacket loss means the failure of one or more transmitted

05 1 15 2 25 301

02

03

04

05

06

07

08

09

1


Net

wor

k th

roug

hput


Figure 2 Network throughput

0 05 1 15 2 25 30

01

02

03

04

05

06


Pack

et lo

ss p

roba

bilit

y


Figure 3 Packet loss probability

packets to arrive at their destinations As the offered trafficload increases wireless nodes will run out of the energy orcapacity for data transmissions and data packets are likely tobe dropped Therefore the packet loss probability increaseslinearly with the traffic load Based on the real-time onlinemanner our dynamic SA approach can improve the systemreliability so we achieve a lower packet loss rate than otherschemes under various traffic loads

The curves in Figures 4 and 5 indicate the average energy-exhaustion ratio and normalized traffic load distribution Inthis paper traffic load distribution means the average rateof traffic dispersion among wireless nodes In an entirely


0 050

01

02

03

04

05

06

07

08

09

1


Ener

gy-e

xhau

stion

nod

e rat

io


1 2 315 25

Figure 4 Energy-exhaustion ratio

05 1 15 2 25 30

01

02

03

04

05

06

07

08

09

1



Nor

mal

ized

traffi

c loa

d di

strib

utio

n

Figure 5 Normalized traffic load distribution

distributed fashion individual node in our scheme moni-tors the current network situation and updates all controlparameters periodically for the adaptive routing Thereforeunder various system constraints the proposed scheme isable to decrease the number of energy expiration nodes andadaptively distribute routing packets to avoid traffic conges-tions which is highly desirable property for the MANETmanagement

The simulation results shown in Figures 1ndash5 demon-strate that the proposed multipath routing scheme generallyexhibits better performance comparedwith the other existingschemes [5 6] Based on the adaptive simulated annealing

approach the proposed scheme constantly monitors thecurrent traffic conditions and gets an efficient solutionThrough the simulation experiments it could be seen thatthe proposed strategy is proved to be an effective paradigmto solve complex routing problems in a dynamic networkenvironment

4 Summary and Conclusions

Recent advances in wireless technology and availability ofmobile computing devices have generated a lot of interestin mobile ad hoc networks For these networks the biggestchallenge is to find routing paths to satisfy varying require-ments In this paper new multipath routing algorithmsare developed based on the effective simulated annealingapproach For real network implementation the proposedscheme is designed in self-organizing dynamic online andinteractive process Therefore each individual node has anability to provide more adaptive control mechanism andmakes a local routing decision to find an efficient pathUnder dynamic network environments this approach candynamically reconfigure the established path to adapt tonetwork changes From simulation results the proposedscheme outperforms existing schemes in terms of networkreliability energy efficiency and so forth

In the future we expect our methodology to be usefulin developing new adaptive ad hoc routing algorithmsIn particular the metaheuristic approach can be extendedto support delay sensitive data services In addition thebasic concept of adaptive online algorithms has become aninteresting research topic in highly mobile ad hoc networks

Conflict of Interests

The author Sungwook Kim declares that there is no conflictof interests regarding the publication of this paper

References

[1] P Deepalakshmi and S Radhakrishnan ldquoQoS routing algo-rithm for mobile ad hoc networks using ACOrdquo in Proceedingsof the International Conference on Control Automation Commu-nication and Energy Conservation (INCACEC rsquo09) pp 1ndash6 June2009

[2] J C-PWangM Abolhasan D R Franklin and F Safaei ldquoEnd-to-end path stability of reactive routing protocols in IEEE 80211ad hoc networksrdquo in Proceedings of the IEEE 34th Conference onLocal Computer Networks (LCN rsquo09) pp 20ndash23 October 2009

[3] F Qin and Y Liu ldquoMultipath based QoS routing in MANETrdquoJournal of Networks vol 4 no 8 pp 771ndash778 2009

[4] M Abolhasan T Wysocki and E Dutkiewicz ldquoA review ofrouting protocols for mobile ad hoc networksrdquo Journal of AdHoc Networks vol 2 no 1 pp 1ndash22 2004

[5] M Klinkowski D Careglio and J Sole-Pareta ldquoReactive andproactive routing in labelled optical burst switching networksrdquoIET Communications vol 3 no 3 pp 454ndash464 2009

[6] H Zafar D Harle I Andonovic and Y Khawaja ldquoPerformanceevaluation of shortest multipath source routing schemerdquo IETCommunications vol 3 no 5 pp 700ndash713 2009


[7] J Leino Applications of game theory in Ad Hoc networks [MSthesis] Helisnki University of Technology 2003

[8] M Randall G McMahon and S Sugden ldquoA simulated anneal-ing approach to communication network designrdquo Journal ofCombinatorial Optimization vol 6 no 1 pp 55ndash65 2002

[9] M Dirani and T Chahed ldquoFramework for resource allocationin heterogeneous wireless networks using game theoryrdquo inProceedings of the 3rd InternationalWorkshop of the EURO-NGINetwork of Excellence pp 144ndash154 2006

[10] T K Varadharajan and C Rajendran ldquoA multi-objectivesimulated-annealing algorithm for scheduling in flowshops tominimize the makespan and total flowtime of jobsrdquo EuropeanJournal of Operational Research vol 167 no 3 pp 772ndash7952005

[11] T Hussain and S J Habib ldquoOptimization of network clusteringand hierarchy through simulated annealingrdquo in Proceedings ofthe 7th IEEEACS International Conference onComputer Systemsand Applications (AICCSA rsquo09) pp 712ndash716 May 2009

[12] K Bouleimen andH Lecocq ldquoA new efficient simulated anneal-ing algorithm for the resource-constrained project schedulingproblem and its multiple mode versionrdquo European Journal ofOperational Research vol 149 no 2 pp 268ndash281 2003

[13] J J Y Leu M H Tsai C Tzu-Chiang et al ldquoAdaptive poweraware clustering and multicasting protocol for mobile ad-hocnetworksrdquo in Ubiquitous Intelligence and Computing pp 331ndash340 2006

[14] J Kim K Lee T Kim and S Yang ldquoEffective routing schemesfor double-layered peer-to-peer systems in MANETrdquo Journal ofComputing Science and Engineering vol 5 no 1 pp 19ndash31 2011

[15] C T Hieu and C Hong ldquoA connection entropy-based multi-rate routing protocol for mobile Ad Hoc networksrdquo Journal ofComputing Science and Engineering vol 4 no 3 pp 225ndash2392010

[16] M Gunes U Sorges and I Bouazizi ldquoARAmdashthe ant-colonybased routing algorithm for MANETsrdquo in Proceedings of theInternational Conference on Parallel Processing Workshops pp79ndash85 August2002

[17] G Wang J Cao L Zhang K C C Chan and J Wu ldquoA novelQoSmulticastmodel inmobile ad hoc networksrdquo inProceedingsof the 19th IEEE International Parallel andDistributed ProcessingSymposium (IPDPS rsquo05) pp 206ndash211 April 2005

[18] L Barolli A Koyama T Suganuma and N ShiratorildquoGAMAN a GA based QoS routing method for mobile ad hocnetworksrdquo Journal of Interconnection Networks vol 4 no 3 pp251ndash270 2003

[19] M Afergan ldquoUsing repeated games to design incentive-basedrouting systemsrdquo in Proceedings of the 25th IEEE InternationalConference on Computer Communications (INFOCOM rsquo06) pp1ndash13 April 2006

[20] M-Y Wu and W Shu ldquoRPF a distributed routing mechanismfor strategic wireless ad hoc networksrdquo in Proceedings of theIEEEGlobal Telecommunications Conference (GLOBECOM rsquo04)pp 2885ndash2889 December 2004

[21] S Kim ldquoCooperative game theoretic online routing scheme forwireless network managementsrdquo IET Communications vol 4no 17 pp 2074ndash2083 2010

[22] S Kim ldquoGame theoretic multi-objective routing scheme forwireless sensor networksrdquo Ad-Hoc amp Sensor Wireless Networksvol 10 no 4 pp 343ndash359 2010

[23] H Shen B Shi L Zou and H Gong ldquoA Distributed entropy-based long-life QoS routing algorithm in Ad Hoc networkrdquo inProceedings of the Canadian Conference on Electrical and Com-puter Engineering Toward a Caring and Humane Technology(CCECE rsquo03) pp 1535ndash1538 May 2003

[24] Y Zou Z Mi and M Xu ldquoDynamic load balancing basedon roulette wheel selectionrdquo in Proceedings of the InternationalConference on Communications Circuits and Systems (ICCCASrsquo06) pp 1732ndash1734 June 2006

Submit your manuscripts athttpwwwhindawicom

Computer Games Technology

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Distributed Sensor Networks


Advances in

FuzzySystems

Hindawi Publishing Corporationhttpwwwhindawicom

Volume 2014


ReconfigurableComputing

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014


Applied Computational Intelligence and Soft Computing

thinspAdvancesthinspinthinsp

Artificial Intelligence

HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014

Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Electrical and Computer Engineering

Journal of

Journal of

Computer Networks and Communications


Hindawi Publishing Corporation

httpwwwhindawicom Volume 2014

Advances in

Multimedia


Biomedical Imaging


ArtificialNeural Systems

Advances in


RoboticsJournal of



Computational Intelligence and Neuroscience

Industrial EngineeringJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Human-ComputerInteraction

Advances in

Computer EngineeringAdvances in



complexity and overheads Due to this reason wireless nodesare assumed to be self-interested agents and make localdecisions in a distributed manner Therefore routing packetsare adaptively distributed through multiple paths in pursuitof the main goals such as load balancing and networkreliability Under diverse network environment changes theproposed scheme tries to approximate an optimal networkperformanceThe important features of our proposed schemeare (i) interactive process to get an efficient network per-formance (ii) distributed approach for large-scale networkoperations (iii) dynamic adaptability to current network and(iv) feasibility for practical implementation

11 Related Work Recently several routing schemes for adhoc networks have been presented in research literatureLeu et al developed the multicast power greedy clustering(MPGC) scheme [13] which is an adaptive power-awareand on-demand multicasting algorithm The MPGC schemeuses greedy heuristic clustering power-aware multicastingand clusteringmaintenance techniques thatmaximize energyefficiency and prolong network lifetime

To improve the network reliability and reduce the net-work traffic Kim et al propose the double-layered effectiverouting (DLER) scheme for peer-to-peer network systems[14] This scheme first chooses the shortest routing pathsamong possible routing paths and selects the path associatedwith the relay peer who has lower mobility to improve thereliability of the system Therefore in the DLER scheme thelower mobility of relay peers contributes to both the stabilityof clusters and the robustness of the system

Hieu and Hong proposed the entropy-based multiraterouting (EMRR) scheme [15] This scheme introduces a newapproach to modeling relative distance among nodes undera variety of communication rates due to nodersquos mobility inMANETs When mobile nodes move to another location therelative distance between communicating nodes will directlyaffect the data rate of transmission Therefore the stability ofa route is related to connection entropy Taking into accountthese issues the link weight and route stability based onconnection entropy are considered as a new routing metricIn the EMRR scheme the problem of determining the bestroute is formulated as the minimization of an object functionformed as a linear combination of the link weight and theconnection uncertainty of that link

The ant-colony based routing algorithm (ARA) schemewas proposed [16] in this scheme swarm intelligence andant-colony metaheuristic techniques are used This schemeconsists of three phases route discovery route maintenanceand route failure handling In the route discovery phasenew routes between nodes are discovered using forwardand backward ants Routes are maintained by subsequentdata packets that is as the data traverse the network nodepheromone values are modified to reinforce the routes

Wang et al developed the logical hypercube-based virtualdynamic backbone (HVDB) scheme for an 119899-dimensionalhypercube in a large-scale MANET [17] The HVDB schemeis a proactive QoS-aware and hybrid multicast routingprotocol Owing to the regularity symmetry properties

and small diameter of the hypercube every node playsalmost the same role In addition no single node is moreloaded than any other node and bottlenecks do not existunlike the case in tree-based architectures In particularthe HVDB scheme can satisfy the new QoS requirementsmdashhigh availability and good load balancingmdashby using locationinformation

Barolli et al proposed the genetic algorithmbased routing(GAR) scheme for mobile ad hoc networks [18] In the GARscheme only a small number of nodes are involved in routecomputation because a small population size is used As aresult routing information is transmitted only for the nodespresent in that population Different routes are ranked bysorting the first route is the best one and the remainingroutes are used as backup routes Because a tree-based geneticalgorithm method is used in the GAR scheme the delay andtransmission success rate are considered as QoS parametersin this scheme

The incentive-based repeated routing (IRR) scheme in[19] is an incentive-based routing model that captures thenotion of repetition To provide a desirable solution the IRRscheme examines certain fundamental properties to governthe behavior of autonomous agents The distributed routingmechanism (DRM) scheme in [20] is an adaptive and scalablerouting scheme for wireless ad hoc networks This schemeprovides a cost-efficient routing mechanism for strategicagents In addition theDRMscheme is designed tomaximizethe benefit of each agent

The proactive congestion reduction (PCR) scheme in [5]focuses on adaptive routing strategies to help congestionreduction Based on a nonlinear optimization method formultipath routings the PCR scheme calculates a traffic split-ting vector that determines a near-optimal traffic distributionover routing paths The shortest multipath source (SMS)scheme [6] is one of the most generally accepted on-demanddynamic routing schemes that build multiple shortest partialdisjoint pathsThe SMS scheme uses node-disjoint secondarypaths to exploit fault tolerance load balancing and band-width aggregation All the earlier work has attracted a lot ofattention and introduced unique challenges However theseexisting schemes have several shortcomings as described inSection 3 Compared to the PCR scheme and the SMS scheme[5 6] the proposed scheme attains better performance forwireless network managements

This paper is organized as follows Section 2 describesthe proposed algorithms in detail In Section 3 performanceevaluation results are presented along with comparisons withthe schemes proposed in [5 6] Finally in Section 4 conclud-ing remarks are given and some directions are identified forfuture work

2 Proposed MANET Routing Algorithms

Multipath routing algorithms are designed to split andtransmit the traffic load through two or more different pathsto a destination simultaneously In this paper we propose anew multipath routing scheme to balance the network loadwhile ensuring efficient network performance




+ [120596 times (1 minus Ψ119894119895 (119905))]

st 120572 =119864119894

119864119872

C119894119895=

119889119894119895

119863119872

Ψ119894119895 (119905) =

120581119894119895

119905

(120581119894119895

119905+ 120599119894119895

119905)

(1)



119872and 119863

119872are


119894119895and 119864


119872and





Θ119895 (119905) =


119896(119905 Δ119905)

log119862 (119865119895)

st 119875119896(119905 Δ119905) =

119886119895119896

sum119894isin119865119895119886119895119894

(2)







119886119895119894=

1

119868 119879

times

119868 119879

sum

119897=1

1003816100381610038161003816V (119895 119894 119905

119897)1003816100381610038161003816

st V (119895 119894 119905) = V (119895 119905) minus V (119894 119905)

(3)






119895119894(119905) rarr 119886

119895119894(119905+Δ119905)


119895(119905) le 1 If Θ

119895(119905)






119905)




119894is the




119894119895 since





1 PC119896 and PC

119873119894 where PC

119896


119894) and119873



value as follows

PC119894= argmin

119896isin119873119894

(PC119896+ 119871 119875

119894119896) (4)



119894value which






SP119896=

119879119862119896

sum119899

119895=1119879119862119895

where 119879119862119896= 1 minus

(PC119895+ 119871 119875

119894119895)

sum119899

119895=1(PC119895+ 119871 119875

119894119895)

119896 isin 119873119894

(5)





119879 (119905)

) (6)
















(a)


(b)




(c)


119894119864119872)





















0 05 1 15 2 25 33

4

5

6

7

8

9

10


Aver

age r

emai

ning

ener

gy







05 1 15 2 25 301

02

03

04

05

06

07

08

09

1


Net

wor

k th

roug

hput



0 05 1 15 2 25 30

01

02

03

04

05

06


Pack

et lo

ss p

roba

bilit

y






0 050

01

02

03

04

05

06

07

08

09

1


Ener

gy-e

xhau

stion

nod

e rat

io


1 2 315 25


05 1 15 2 25 30

01

02

03

04

05

06

07

08

09

1



Nor

mal

ized

traffi

c loa

d di

strib

utio

n










References

































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in






+ [120596 times (1 minus Ψ119894119895 (119905))]

st 120572 =119864119894

119864119872

C119894119895=

119889119894119895

119863119872

Ψ119894119895 (119905) =

120581119894119895

119905

(120581119894119895

119905+ 120599119894119895

119905)

(1)



119872and 119863

119872are


119894119895and 119864


119872and





Θ119895 (119905) =


119896(119905 Δ119905)

log119862 (119865119895)

st 119875119896(119905 Δ119905) =

119886119895119896

sum119894isin119865119895119886119895119894

(2)







119886119895119894=

1

119868 119879

times

119868 119879

sum

119897=1

1003816100381610038161003816V (119895 119894 119905

119897)1003816100381610038161003816

st V (119895 119894 119905) = V (119895 119905) minus V (119894 119905)

(3)






119895119894(119905) rarr 119886

119895119894(119905+Δ119905)


119895(119905) le 1 If Θ

119895(119905)






119905)




119894is the




119894119895 since





1 PC119896 and PC

119873119894 where PC

119896


119894) and119873



value as follows

PC119894= argmin

119896isin119873119894

(PC119896+ 119871 119875

119894119896) (4)



119894value which






SP119896=

119879119862119896

sum119899

119895=1119879119862119895

where 119879119862119896= 1 minus

(PC119895+ 119871 119875

119894119895)

sum119899

119895=1(PC119895+ 119871 119875

119894119895)

119896 isin 119873119894

(5)





119879 (119905)

) (6)
















(a)


(b)




(c)


119894119864119872)





















0 05 1 15 2 25 33

4

5

6

7

8

9

10


Aver

age r

emai

ning

ener

gy







05 1 15 2 25 301

02

03

04

05

06

07

08

09

1


Net

wor

k th

roug

hput



0 05 1 15 2 25 30

01

02

03

04

05

06


Pack

et lo

ss p

roba

bilit

y






0 050

01

02

03

04

05

06

07

08

09

1


Ener

gy-e

xhau

stion

nod

e rat

io


1 2 315 25


05 1 15 2 25 30

01

02

03

04

05

06

07

08

09

1



Nor

mal

ized

traffi

c loa

d di

strib

utio

n










References

































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





119894value which






SP119896=

119879119862119896

sum119899

119895=1119879119862119895

where 119879119862119896= 1 minus

(PC119895+ 119871 119875

119894119895)

sum119899

119895=1(PC119895+ 119871 119875

119894119895)

119896 isin 119873119894

(5)





119879 (119905)

) (6)
















(a)


(b)




(c)


119894119864119872)





















0 05 1 15 2 25 33

4

5

6

7

8

9

10


Aver

age r

emai

ning

ener

gy







05 1 15 2 25 301

02

03

04

05

06

07

08

09

1


Net

wor

k th

roug

hput



0 05 1 15 2 25 30

01

02

03

04

05

06


Pack

et lo

ss p

roba

bilit

y






0 050

01

02

03

04

05

06

07

08

09

1


Ener

gy-e

xhau

stion

nod

e rat

io


1 2 315 25


05 1 15 2 25 30

01

02

03

04

05

06

07

08

09

1



Nor

mal

ized

traffi

c loa

d di

strib

utio

n










References

































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





(a)


(b)




(c)


119894119864119872)





















0 05 1 15 2 25 33

4

5

6

7

8

9

10


Aver

age r

emai

ning

ener

gy







05 1 15 2 25 301

02

03

04

05

06

07

08

09

1


Net

wor

k th

roug

hput



0 05 1 15 2 25 30

01

02

03

04

05

06


Pack

et lo

ss p

roba

bilit

y






0 050

01

02

03

04

05

06

07

08

09

1


Ener

gy-e

xhau

stion

nod

e rat

io


1 2 315 25


05 1 15 2 25 30

01

02

03

04

05

06

07

08

09

1



Nor

mal

ized

traffi

c loa

d di

strib

utio

n










References

































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




0 05 1 15 2 25 33

4

5

6

7

8

9

10


Aver

age r

emai

ning

ener

gy







05 1 15 2 25 301

02

03

04

05

06

07

08

09

1


Net

wor

k th

roug

hput



0 05 1 15 2 25 30

01

02

03

04

05

06


Pack

et lo

ss p

roba

bilit

y






0 050

01

02

03

04

05

06

07

08

09

1


Ener

gy-e

xhau

stion

nod

e rat

io


1 2 315 25


05 1 15 2 25 30

01

02

03

04

05

06

07

08

09

1



Nor

mal

ized

traffi

c loa

d di

strib

utio

n










References

































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




0 050

01

02

03

04

05

06

07

08

09

1


Ener

gy-e

xhau

stion

nod

e rat

io


1 2 315 25


05 1 15 2 25 30

01

02

03

04

05

06

07

08

09

1



Nor

mal

ized

traffi

c loa

d di

strib

utio

n










References

































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





























Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in










Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in



Documents

Research Article Adaptive MANET Multipath Routing ...downloads.hindawi.com/journals/tswj/2014/872526.pdfResearch Article Adaptive MANET Multipath Routing Algorithm Based on the Simulated