epsoc:social-basedepidemic-basedroutingprotocolin...

Research ArticleEpSoc Social-Based Epidemic-Based Routing Protocol inOpportunistic Mobile Social Network

Halikul Lenando 1 and Mohamad Alrfaay2

1Faculty of Computer Science and Information Technology Universiti Malaysia Sarawak Kota Samarahan Malaysia2Faculty of Computer and Information Sciences Jouf University Jouf Saudi Arabia

Correspondence should be addressed to Halikul Lenando coolunimasmy

Received 17 October 2017 Revised 20 February 2018 Accepted 11 March 2018 Published 4 April 2018

Academic Editor Fabio Gasparetti

Copyright copy 2018Halikul Lenando andMohamadAlrfaayis is an open access article distributed under the Creative CommonsAttribution License which permits unrestricted use distribution and reproduction in anymedium provided the original work isproperly cited

In opportunistic networks the nature of intermittent and disruptive connections degrades the efficiency of routing Epidemicrouting protocol is used as a benchmark for most of routing protocols in opportunistic mobile social networks (OMSNs) due to itshigh message delivery and latency However Epidemic incurs high cost in terms of overhead and hop count In this paper wepropose a hybrid routing protocol called EpSoc which utilizes the Epidemic routing forwarding strategy and exploits an importantsocial feature that is degree centrality Two techniques are used in EpSocMessagesrsquo TTL is adjusted based on the degree centralityof nodes and the message blocking mechanism is used to control replication Simulation results show that EpSoc increases thedelivery ratio and decreases the overhead ratio the average latency and the hop counts as compared to Epidemic and Bubble Rap

1 Introduction

Opportunistic mobile social network (OMSN) [1ndash4] is apromising networking model for data dissemination In theOMSN mobile nodes grab the opportunity of encounteringthe peer (they are in the communication range of each other)to forward the data e OMSN incurs intermittent and dis-ruptive connectivity due to node mobility To tackle with thiscomplex environment the store-carry-forward scheme is ap-plied in the OMSN If no connection is available at a particulartime a mobile node stores data in its buffer and carries themuntil it encounters other mobile nodes to forward the data[5ndash8] Various approaches have been proposed to address theinformation delivery problem in the OMSN such as in [9ndash12]emain concerns of OMSN routing approaches are yieldinghigh delivery ratio low delay and low overhead or cost onnetworks and nodes

Flooding is one of the dominant schemes to disseminatedata in the OMSN [13] Each message will be flooded to everynode in the network Multiple copies of each message aregenerated and spread in the network Epidemic [9] routingprotocol is the flooding-based routing protocol When two

nodes encounter they exchange all of their messages isresults inmessages spread over the whole network by pairwisecontacts between two nodes If no buffer constraints are appliedEpidemic represents the upper bound in message delivery andlatency Epidemic routing is used as a benchmark and a refer-ence for themost of other routing protocols in the opportunisticnetworkemain drawback of the Epidemic scheme is its highoverhead Many schemes are proposed to decrease the over-head in Epidemic-based approaches by limiting the number ofmessage replicas [14ndash16] An effective scheme to controlreplication spread is the vaccine [17] It applies the antipacketmechanism to control replica distribution in Epidemic-basedrouting In [18] a new scheme is proposed to control thereplication of epidemically distributed information Based onthe vaccine scheme signal distribution is early controlled bythe fully immunized vaccine In addition a partially immu-nized vaccine is initiated when there is a local-forwardingopportunity to vaccine more packets

In the OMSN mobile devices are portable by humans sothat social features of people can be exploited for network-ing purposes [19ndash21] Social-based protocols utilize socialproperties of mobile users such as similarity centrality and

HindawiMobile Information SystemsVolume 2018 Article ID 6462826 8 pageshttpsdoiorg10115520186462826

friendship to improve routing efficiency in the opportunisticmobile social networkis is because social features are morestable and less changeable than other features like mobilitypatterns HiBOp [22] and CiPRO [23] exploit the similaritysocial feature and the userrsquos context information to forwarddata LASS [24] takes into account the difference of membersrsquoactivity within the nodersquos community for data disseminationIn ML-SOR [25] node centrality (different types of central-ities) the similarity between communities and social ties areall exploited to effectively select the forwarding node MCAR[26] exploits the preferred communities of people during theirdaily lives for effective information delivery Direct (inside onecommunity) and indirect (via different communities) contactsare considered in MCAR In SPRINT-SELF [27] network andnode overhead is decreased by exploiting social information ofmobile users e authors consider the social community ofnodes and utilize it to predict future behavior based on contacthistory In addition they proposed a newmechanism to avoidselfish nodes for more improvements

Social features are utilized widely for buffer managementLiu et al [28] utilized social features and the congestion levelto develop the forwarding strategy that drops the messagewith the minimum social link rather than random droppingIn SRAMSW [29] the buffer management mechanism iscombined with social features to enhance the spray-basedrouting Expired messages are deleted successfully deliveredmessages are acknowledged and messages are prioritizedaccording to their spray times and residence time In additionthree social features centrality similarity and friendship areadopted for better forwarding decision and to avoid thedead-end problem

We hypothesize that combining social features with theEpidemic-based forwarding scheme improves the efficiencyof routing in the OMSN In this paper we present anEpidemic-based Social-based routing protocol (EpSoc) thatcombines the advantages of the forwarding strategy used inthe Epidemic routing protocol with the positive impact ofexploiting social features EpSoc exploits the degree centralitysocial feature to adapt the time to live (TTL) of the routedmessage If a message is forwarded to a node which has higherdegree centrality (socially active node) its TTL value will bedecreased If these messages are dropped in the active nodewhen TTL is zero the blocking mechanism is used to rejectreceiving replications of these messages

e rest of this paper is organized as follows the nextsection reviews the related work We describe in detail ourproposed algorithm EpSoc in Section 3 In Section 4 weintroduce the performance evaluation and the discussionresults Finally Section 5 concludes the paper

2 Related Works

Delivering data to the destination at the right time withminimum resources is an optimal condition for any givenrouting protocols Epidemic has optimal performance interms of delivery ratio and latency However it suffers fromhigh overhead cost One of the solutions is to exploit socialinformation to improve Epidemic-based routing protocolsin the OMSN Degree centrality is a social feature that is

exploited widely in the literature to improve routing in themobile social networks For example Bubble Rap [11] utilizesthe mobile nodersquos degree centrality to provide cost-effectiverouting as compared to Epidemic routing Bubble Rap ex-ploits two social and structural metrics namely centrality andcommunity It selects high centrality nodes and communitymembers of destination as relays In the Bubble Rap algo-rithm nodes belong to different sizes of communities andhave different levels of popularity (ie rank) Each node isassumed to have two rankings global denotes the popularity(ie connectivity) of the node in the entire society and localdenotes the popularity within its community Messages areforwarded to nodes that have higher global ranking until anode in the destinationrsquos community is found en themessages are forwarded to nodes having a higher local rankingwithin the destinationrsquos community

Similar to Bubble Rap CAOR [30] exploits degree cen-trality and similarity social features to improve routingHowever CAOR constructed autonomous communities basedon common interest locations between mobile nodes Mem-bers of a community with high centralities act as the home ofthis community CAOR also has a mechanism to turn therouting between lots of nodes to the routing between a fewcommunity homes It also applies the reverse Dijkstra algo-rithm to determine the optimal relays and compute theminimum expected delivery delay In [31] cultural algorithm(CA) ant colony optimization (ACO) and social connectivitybetween users are combined to address the routing problemSocial metrics of nodes including degree and betweennesscentralities are analyzed to support forwarding decision in theopportunistic network environment

Besides exploited in the routing problem social featureswere also applied to solve different type of problems Forexample works in [32] exploit social features (degree andbetweenness centralities) to solve the throwbox placementproblem based on the given graph A userrsquos degree is equal tothe total number of its neighbors and betweenness is thetotal number of shortest paths passing through the nodeework also introduced the concept of location degree tomeasurehow many mobile users have the location as one of their topvisited locations and the location betweenness to show howimportant the location is for the entire social graph Socialmetrics such as degree centrality social activeness and com-munity acquaintance are also applied to enhance data deliveryin VSNs in [33] e social interactions between nodes wherenodes of similar interests or nodes belonging to the samecommunity have greater probability to encounter each otherSocial centrality is also utilized for congestion control in thepostdisaster environment [34]

Unlike the aforementioned works instead of consideringcommunity or homing structuring our proposed protocol isbased on the flooding-based forwarding strategy of Epi-demic is is because we intended to design our forwardingprotocol to have high delivery ratio and low delay as theEpidemic-based forwarding strategy Moreover we alsoaimed to avoid extra time required to form and maintain thecommunity structure us to decrease the overhead weutilized degree centrality and adapted the messagersquos TTL(Time to live) to control the forwarding nodesrsquo activity e

2 Mobile Information Systems

messagersquos TTL is considered in literature when developingefficient routing protocols Miao et al [33] proposed theadaptive multistep routing protocol for mobile delay-tolerantnetworks (MDTNs)eir aim is to get the balance between thedelay and cost of message delivery e time to live of themessage is used in order to allocate the minimum number ofcopies necessary to achieve a given delivery probability In [34]the authors considered the contact information of nodes andthe time to live messagersquos property to make route decisionand improve performance ey built a replica distributioncriterion between two encountered nodes based on residualmessagesrsquo TTL In our proposed solution we do not limitthe number of replicas but allowmessages to be spread in thenetwork and then we utilize degree centrality to decreasethe TTL and consequently decrease the overhead Also theblocking mechanism is adopted in the active node to cancelreceiving replications of the same message which is seenpreviously

3 EpSoc Routing Protocol

EpSoc is the routing algorithm designed to decrease overheadin the Epidemic protocol by embedding social features inrouting the message in the opportunistic network To achievethis objective we propose two mechanisms First the mes-sagersquos TTL is adapted based on degree centrality of the nodesin the OMSN e second one is the message blockingmechanism that is used to prevent receiving the replicationsof runout TTL messages in active nodes

31 Degree Centrality Node centrality indicates the popu-larity of the node in the network whereby node centrality ina social network is the reflection of its social relative im-portance [35] A higher node degree centrality means thenode connects with many numbers of nodes in the networkA degree centrality for a given node i can be calculated asfollows

DCi 1113944N

k1

a(i k) (1)

whereN is the number of nodes in the network and a(i k) 1if a direct link exists between node i and node k and ine k

We adopt the CWindow [11] calculation algorithm tocalculate degree centrality CWindow divides the day intotime windows and calculates the average of nodesrsquo degree overthese windows to estimate the nodersquos centrality To decreasethe processing overhead results from processing any changesof degree centrality CWindow calculates the nodersquos degreecentrality at regular intervals instead of every time the cen-trality changes

We pick the CWindow algorithm to calculate degreecentrality because it takes into account the changes in nodecentrality over time and averages the nodersquos centralities for fewwindow intervals which is suitable for peoplersquos behavior in theOMSN It is also used by other social-based protocols thatconsider the nodersquos degree centrality such as Bubble Rap [11]Dlife [36] and SCORP [37]

32 EpSoc Forwarding Strategy Figure 1 depicts the for-warding process in EpSoc where two mechanisms are applied

In Figure 1(a) node N1 encountered three nodes N2N3 and N4 N2 has higher degree centrality (DC) than N1e TTL value of the forwarded messages to N2 is decreasedby dividing it by the DC value of N2 We call these messagessocially infected messages Node N2 registers the ID of thesemessages as seen messages in its blocking register Node N2will drop the socially infected message when it expires InFigure 1(b) the Epidemic forwarding strategy which weadopted in our algorithm causes replications of socially in-fected messages to be sent to node N2 from other nodes suchas N4 In this case node N2 will reject receiving the messagewith the reason that it is a seen message (its ID is stored in theblock register)

Based on (1) the TTL value is adapted using the fol-lowing equation

TTLnew TTLoldDCk

if DCk gtDCi1113864 1113865 (2)

e two combined mechanisms adopted in EpSoc en-hance routing performance If a node is socially active itmeets more nodes in the network and has a higher potential todeliver more messages Decreasing messagesrsquo TTL value inactive nodes results in releasing space in their buffers andincreases the ability to deliver more messages is will in-crease the delivery ratio e blocking mechanism controlsthe replications by preventing decreased TTL messages fromhitting again the previously traversed actives nodese resultis decreasing network overhead Regarding average latencythe messages that are delivered by active nodes have a shorterend-to-end delay compared to other messages delivered bylower activity nodes so average latency will be decreased tooWe conclude that decreasing message life socially by com-bining the message blocking scheme affects positively theperformance of the Epidemic routing scheme in the OMSN

33 Pseudocode of the EpSoc Message Forwarding We usedCWindow to calculate the centrality of the node Each nodeNi records the encountered nodes in the networkWhen nodeNi encounters Nj the degree centrality values are calculatedusing CWindow en centrality values are exchanged be-tween both nodes Ni compares its centrality value DCi withthe Nj centrality value DCj If DCj is greater than DCi whichmeans that Nj is socially more active than Ni then for eachmessage mi in the Ni buffer BufNi

the TTL value miTTLis

decreased by dividing it by the node Nj centrality value DCje ID of each socially infectedmessage is registered in theNj

blocking register BlockNj e time complexity of the EpSoc

forwarding algorithm is Ο(M) where M is the number ofmessages carried in the nodersquos buffer and needed to be sent orforwarded

Algorithm 1 shows the complete pseudocode of EpSoc

4 Performance Evaluations

41 Data Set We adopt the Cambridge experimental dataset of haggle [38] is data set includes traces of Bluetooth

Mobile Information Systems 3

sightings by groups of users carrying small devices (iMotes) fora number of days in campus environments e experimentsare conducted in the computer laboratory that includes theundergraduate rst-year and second-year students and alsosome PhD and postgraduate students which lasted for 11 days

42 Simulation Setup We use the opportunistic networkenvironment (ONE) [39] simulator to evaluate our algo-rithm Also comparison with Epidemic and Bubble Raprouting protocols is included to measure the performance ofour proposed algorithm EpSoc We want to justify that theEpidemic algorithm performs better when considering socialfeatures in forwarding messages e simulator settings aretabulated in Table 1

In each experiment we compare the performance of theprotocols EpSoc Epidemic and Bubble Rap based on thefollowing metrics

421 Successful Delivery Ratio It is the ratio between thenumber of deliveredmessages and the total number of createdmessages e ideal value of the successful delivery ratio is 10when all created messages are delivered to their destinations

422 Overhead Ratio It is the additional bytes that are sentfor successfully delivering a message to a destination

423 Average Latency It is the average of the time elapsedbetween message creation and delivery

424 Average Hop Count It is the average of the numberof hops that messages must take in order to reach thedestination

43 Experiments and Discussion To evaluate our work wewill consider two features buer size and message TTLvalue ese two features have a high impact on routingperformance in the OMSN In our work we aect thesetwo features We adjust the TTL value socially and use theblocking mechanism to manage buer storing Consequentlyour experiments are to measure the performance of EpSocwhen varying the TTL and buer sizee comparison will bewith Epidemic and Bubble Rap protocols

431 Varying Buer Size For varying the buer size wexed the value of TTL to 25 d Figures 2ndash5 show the per-formance comparisons between EpSoc Epidemic and BubbleRap in terms of delivery ratio overhead ratio average latencyand average hop count respectively

Figure 2 shows the delivery ratio with buer size Gen-erally for Epidemic Bubble Rap and EpSoc increasing the

N2_DC gt N1_DC (true) (i) Decrease message TTL(ii) Register message in N2 block register

N3_DC = N1_DC (true)No change in message TTL

N2

N1 N3

N4 N4_DC lt N1_DC (true)No change in message TTL

(a)

(i) Decrease message TTL(ii) Register message in N5 block

register

N6_DC lt= N2_DC (true)No change in message TTL

N1 N2

N4 N6

N5

N5_DC gt N2_DC (true)

(b)

Figure 1 EpSoc forwarding scheme

(1) For all Ni isin N(2) if Ni encounter Nj(3) Calculate DCi DCj(4) DCiDCj(5) For all mi isin BufNi

(6) if miIDnotin BlockNj

(7) if DCj gtDCi(8) miTTL

larrmiTTLDCj

(9) BufNjlarrmi forward message to Nj

(10) BlockNjlarrmiID

(11) End if(12) End if(13) End for(14) End if(15) End for

ALGORITHM 1 Pseudocode of the forwarding strategy in EpSoc

Table 1 Simulation settingsSimulation time 987529 secondsInterface Bluetooth interfaceNo of nodes 36Transmit speed 250k (2Mbps)Mobility Real trace data (Cambridge)Buer size 1 5 15 25 35 45 and 55MBRoutingprotocols Epidemic EpSoc and Bubble Rap

Message size 128kEvent interval 30 to 40 secondsInitial messageTTL

10m 30m 1 h 3 h 5 h 12 h 1 d 15 d 25 d3 d 4 d and 1w


buer size will increase the delivery ratiois is becausemoremessages can be carried by intermediate nodes which con-sequently deliver more messages to destinations Changingthe number of delivered messages aects the delivery ratiooverhead ratio average latency and average hop count Re-garding the delivery ratio it is clear that delivering moremessages results in a higher value while the decrease in de-livered messages results in a lower value In the lower buersize scenario (1ndash25MB) the delivery ratio of Epidemic is thelowest because of the redundancy Bubble Rap and EpSocoutperform Epidemic due to utilizing social features In higherbuer size scenarios (35ndash55MB) Epidemic achieves higherdelivery ratio than Bubble Rap Larger buer size alleviates thenegative impact of dropping messages because of replicationand enables Epidemic to deliver more messages Our protocolEpSoc outperforms both Epidemic and Bubble Rap Blockingrunout TTL messages from being received by active nodesresults in more space in their buer and therefore can carrymore dierent messages when encountering other nodes andlater deliver them to destinations In addition the decrease ofTTL of the messages that are forwarded to the more activenodes results in better utilization of the buerrsquos space De-creased TTL message copies are dropped earlier enablingcarrying more other messages Active nodes deliver themessage quickly erefore the number of delivered messagesis increased and consequently the delivery ratio is increased

e relation of overhead ratio with buer size is shownin Figure 3 When increasing the buer size the overhead isdecreased for all algorithms Regarding Bubble Rap over-head is decreased with buer increase because no replicationexists We observe from Figure 3 that both protocols BubbleRap and EpSoc outperform Epidemic signicantly andEpSoc outperforms Epidemic We mentioned formerly thatapplying the strategy of blocking messages and decreasingTTL increase the number of delivered messages In additionthe blocking message to be resent to active nodes decreases

the number of replications in the network and hence de-creases the forwardings As a result decreasing the for-wardings and increasing the delivered messages decrease theoverhead ratio in the network Bubble Rap achieves loweroverhead than EpSoc for lower buer size scenarios (1 2 5and 15MB) is is because of two reasons rst the rep-lication and Epidemic strategy adopted in EpSoc and sec-ond low buer size causes dropping relayed messages earlydue to buer overcentow which in turn decreases the epoundciencyof our proposed algorithm compared to Bubble Rap in termsof overhead However for larger buer size scenarios (25 3545 and 55MB) EpSoc is more epoundcient and its overheadratio is very close to that of Bubble Rap (for 35 45 and55MB it is slightly better than Bubble Rap)

Figure 4 shows that average latency for all the threeprotocols which increases when the buer size is increased Alow-sized buer only relays low-latency messages which willreach their destinations quickly On the other hand a higher-sized buer allows messages to be carried for a longer timewhich contribute to a higher average latency EpSoc decreasesthe average latency signicantly EpSoc optimizes the buerusage where messages forwarded to active nodes have lowTTL With the low TTL a relay node has more free space fornew messages in its buer as the buer quickly dropped theold messages

In Figure 5 the average hop count is recorded eaverage hop count increases when the buer size is in-creased Epidemic has more hop count which indicates morenodes experiencing the duplicatedmessages whereas BubbleRap has lower hop count because it prevents replication ofmessages For EpSoc the number of hop count is con-tributed by allowing replication as in Epidemic

e worst achievement is because of redundancy BubbleRap does not apply replication so that it outperforms EpSocEpSoc achieves better average hop count than Epidemicis is because of the message blocking strategy whichdecreases the number of replications in the network andhence decreases the forwardings

0 10 20 30 40 50 600

005

01

015

02

025

03

035

04

Buffer size (MB)

EpidemicEpSocBubble Rap

Del

iver

y ra

tio

Figure 2 Delivery ratio versus buer size

Buffer size (MB)


Ove

rhea

d ra

tio

0 10 20 30 40 50 600

100

200

300

400

500

600

700

Figure 3 Overhead ratio versus buer size


432 Varying Initial TTL TTL determines the life of themessage in the network A high value of TTL gives highchances of the message delivered to the target destinationand vice versa

In Figure 6 the relation of delivery ratio with TTL isdepicted For lower TTL (10mndash3 h) the delivery ratio ofEpidemic is slightly higher than that of Bubble Rap andEpSoc e reason is that for short TTL messages exploitingsocial features will not be very eective due to quicklymessages dropping In addition the higher replication oc-curred in Epidemic increases the number of deliveredmessages When TTL is increased Bubble Rap and EpSocoutperform Epidemic due to utilizing the social features Inhigh TTL scenarios (15 dndash1w) EpSoc outperforms BubbleRap is indicates an epoundciency of EpSoc in its social featureselection

Shortening the life of messages in the active nodes andblocking runout TTL messages from hitting again the activenodes cause an increment in the delivered messages anddecrement in forwardings erefore delivery ratio grows up

Figure 7 compares the performance of the protocols withdierent values of TTL in terms of overhead ratio WhenTTL is very low (10mndash1 h) Epidemic EpSoc and BubbleRap achieve low overhead e reason is that messages aredropped quickly For high TTL values Bubble Rap achievesthe best and Epidemic the worst

Our protocol EpSoc manages to decrease the overheadbetter than Epidemicis is because EpSoc is a combinationof the centooding-based forward strategy with social features

From Figure 8 in terms of average latency EpSoc ap-pears to be outperforming others especially with high TTL


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

Del

iver

y ra

tio

002

004

006

008

01

012

014

016

018

02

TTL (minutes)

Figure 6 Delivery ratio versus TTL

Ove

rhea

d ra

tio

0

50

100

150

200

250

300

350

400


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

TTL (minutes)

Figure 7 Overhead ratio versus TTL


Aver

age l

aten

cy

0 10 20 30 40 50 600

1

2

3

4

5

6

7(times104)

Buffer size (MB)

Figure 4 Average latency versus buer size


Aver

age h

op co

unt

0 10 20 30 40 50 601

2

3

4

5

6

7

8

Buffer size (MB)

Figure 5 Average hop count versus buer size


values (12 hndash1w) is is because our algorithm alwaysenables the active node to carry messages with lower TTL

From Figure 9 we observed that if TTL is very low(10mndash1 h) all routing protocols in the experiment have lowaverage hop count is is due to the low number of for-wardings between nodes as message life exhausted quicklyBubble Rap and EpSoc have almost stable performance whenTTL is increased Exploited social features in Bubble Rap andEpSoc and applying the blocking mechanism in EpSocdecrease the eect of changing the value of TTL on thenumber of traversed nodes to the destination On thecontrary when the TTL is increased the average hop countof Epidemic increases signicantly because of the centooding-

based forwarding strategy For EpSoc the average hop countis decreased signicantly compared to Epidemic because ofthe social eect of active nodes

5 Conclusion

In this paper we investigate the centooding-based forwardingstrategy that is Epidemic with social features to improve therouting performance in the opportunistic mobile social net-work (OMSN) Inspired by the advantages of Epidemic routingprotocols in terms of delivery ratio and delivery delay and byexploiting the social activity of nodes we formulated a centooding-based social-based routing protocol named as EpSoc Simu-lation experiments using real data sets are conducted to evaluateour protocol performance From the presented results ourapproach increases the delivery ratio and decreases the deliveryoverhead average latency and average hop count as com-pared to the Epidemic protocol As for benchmark social-basedrouting protocol performance (Bubble Rap) our protocoldecreases the average latency signicantly with a better deliveryratio in high buer size and low TTL scenarios Generally wemanage to exploit the advantage of Epidemic and Bubble Rapto improve the data dissemination in the OMSN

Additional Points

Signicance Social features of a node can be utilized to have aneective role in the routing protocol in the OMSN networkis paper presents the routing protocol that exploits degreecentrality to increase the delivery ratio and decrease theoverhead and latency Moreover exploiting other social fea-tures such as similarity and communitymay lead to beingmoreepoundcient routing protocol erefore combining social featureswith other forwarding techniques such as the Epidemic-basedstrategy is signicant to have an epoundcient forwarding strategy inthe OMSN to support green technologies

Conflicts of Interest

e authors declare that there are no concenticts of interestregarding the publication of this paper

References

[1] A Chaintreau A Mtibaa L Massoulie and C Diot ldquoediameter of opportunistic mobile networksrdquo in Proceedings ofthe 2007 ACM CoNEXT Conference p 12 ACM New YorkNY USA December 2007

[2] M Conti S Giordano M May and A Passarella ldquoFromopportunistic networks to opportunistic computingrdquo IEEECommunications Magazine vol 48 no 9 pp 126ndash139 2010

[3] K Zhu W Li X Fu and L Zhang ldquoData routing strategies inopportunistic mobile social networks taxonomy and openchallengesrdquo Computer Networks vol 93 pp 183ndash198 2015

[4] Y Liu Z Yang T Ning and H Wu ldquoEpoundcient quality-of-service (QoS) support in mobile opportunistic networksrdquoIEEE Transactions on Vehicular Technology vol 63 no 9pp 4574ndash4584 2014

[5] L Pelusi A Passarella and M Conti ldquoOpportunistic net-working data forwarding in disconnected mobile ad hoc

Aver

age l

aten

cy


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0TTL (minutes)

0

05

1

15

2

25

3

35(times104)

Figure 8 Average latency versus TTL

Aver

age h

op co

unt


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

TTL (minutes)

1

2

3

4

5

6

7

8

9

10

Figure 9 Average hop count versus TTL


networksrdquo IEEE Communications Magazine vol 44 no 11pp 134ndash141 2006

[6] T Le and M Gerla ldquoSocial-distance based anycast routing indelay tolerant networksrdquo in Proceedings of the 2016 Medi-terranean Ad Hoc Networking Workshop (Med-Hoc-Net)Vilanova i la Geltru Spain June 2016

[7] K Fall ldquoA delay-tolerant network architecture for challengedinternetsrdquo in Proceedings of the 2003 Conference on Appli-cations Technologies Architectures and Protocols for Com-puter Communications New York NY USA August 2003

[8] M Alajeely R Doss and A Ahmad ldquoRouting protocols inopportunistic networksndasha surveyrdquo IETE Technical Reviewpp 1ndash19 2017

[9] A Vahdat and D Becker ldquoEpidemic routing for partially-connected ad hoc networksrdquo Tech Rep CS-200006 DukeUniversity Durham NC USA 2000

[10] A Lindgren A Doria and O Schelen Probabilistic Routing inIntermittently Connected Networks Springer Berlin Ger-many 2004

[11] P Hui J Crowcroft and E Yoneki ldquoBUBBLE Rap social-basedforwarding in delay-tolerant networksrdquo IEEE Transactions onMobile Computing vol 10 no 11 pp 1576ndash1589 2011

[12] P Pholpabu and L L Yang ldquoRouting protocols for mobilesocial networks achieving trade-off among energy con-sumption delivery ratio and delayrdquo in Proceedings of the 2015IEEECIC International Conference on Communications inChina (ICCC) Shenzhen China November 2015

[13] M Nikoo F Ramezani M Hadzima-Nyarko E K Nyarkoand M Nikoo ldquoFlood-routing modeling with neural networkoptimized by social-based algorithmrdquo Natural Hazardsvol 82 no 1 pp 1ndash24 2016

[14] E Bulut Z Wang and B K Szymanski ldquoCost-effectivemultiperiod spraying for routing in delay-tolerant networksrdquoIEEEACM Transactions on Networking (TON) vol 18 no 5pp 1530ndash1543 2010

[15] T Matsuda and T Takine ldquo(pq)-epidemic routing for sparselypopulated mobile ad hoc networksrdquo IEEE Journal on SelectedAreas in Communications vol 26 no 5 pp 783ndash793 2008

[16] P Mundur M Seligman and J N Lee ldquoImmunity-basedepidemic routing in intermittent networksrdquo in Proceedings ofthe 2008 5th Annual IEEE Communications Society Conferenceon Sensor Mesh and Ad Hoc Communications and NetworksSan Francisco CA USA June 2008

[17] P Y Chen S M Cheng and K C Chen ldquoOptimal controlof epidemic information dissemination over networksrdquo IEEETransactions on Cybernetics vol 44 no12 pp 2316ndash2328 2014

[18] C Y Aung I W H Ho and P H J Chong ldquoStore-carry-cooperative forward routing with information epidemicscontrol for data delivery in opportunistic networksrdquo IEEEAccess vol 5 pp 6608ndash6625 2017

[19] Y Zhu B Xu X Shi and Y Wang ldquoA survey of social-basedrouting in delay tolerant networks positive and negativesocial effectsrdquo IEEE Communications Surveys amp Tutorialsvol 15 no 1 pp 387ndash401 2013

[20] C C Sobin V Raychoudhury G Marfia and A Singla ldquoAsurvey of routing and data dissemination in delay tolerantnetworksrdquo Journal of Network and Computer Applicationsvol 67 pp 128ndash146 2016

[21] P Pholpabu and L-L Yang ldquoSocial contact probabilityassisted routing protocol for mobile social networksrdquo inProceedings of the IEEE 79th Vehicular Technology Conference(VTC Spring) Seoul South Korea May 2014

[22] C Boldrini M Conti and A Passarella ldquoExploiting usersrsquosocial relations to forward data in opportunistic networks the

HiBOp solutionrdquo Pervasive and Mobile Computing vol 4no 5 pp 633ndash657 2008

[23] H Nguyen and S Giordano ldquoContext information predictionfor social-based routing in opportunistic networksrdquo Ad HocNetworks vol 10 no 8 pp 1557ndash1569 2012

[24] Z Li C Wang S Yang C Jiang and X Li ldquoLASS local-activity and social-similarity based data forwarding in mobilesocial networksrdquo IEEE Transactions on Parallel and Distrib-uted Systems vol 26 pp 174ndash184 2015

[25] A Socievole E Yoneki F De Rango and J Crowcroft ldquoML-SOR message routing using multi-layer social networks inopportunistic communicationsrdquo Computer Networks vol 81pp 201ndash219 2015

[26] P Pholpabu and L-L Yang ldquoA mutual-community-awarerouting protocol for mobile social networksrdquo in Proceedings ofthe Global Communications Conference (GLOBECOM)Austin TX USA December 2014

[27] R Ciobanu C Dobre V Cristea F Pop and F XhafaldquoSPRINT-SELF social-based routing and selfish node de-tection in opportunistic networksrdquo Mobile Information Sys-tems vol 2015 Article ID 596204 12 pages 2015

[28] Y Liu K Wang H Guo Q Lu and Y Sun ldquoSocial-awarecomputing based congestion control in delay tolerant networksrdquoMobile Networks and Applications vol 22 no 2 pp 1ndash12 2016

[29] J Guan Q Chu and I You ldquoe social relationship basedadaptive multi-spray-and-wait routing algorithm for dis-ruption tolerant networkrdquo Mobile Information Systemsvol 2017 Article ID 1819495 13 pages 2017

[30] M Xiao J Wu and L Huang ldquoCommunity-aware oppor-tunistic routing in mobile social networksrdquo IEEE Transactionson Computers vol 63 pp 1682ndash1695 2014

[31] A C K Vendramin A Munaretto M R Delgado M Fonsecaand A C Viana ldquoA social-aware routing protocol for oppor-tunistic networksrdquo Expert Systems with Applications vol 54pp 351ndash363 2016

[32] Y Zhu C Zhang X Mao and Y Wang ldquoSocial basedthrowbox placement schemes for large-scale mobile socialdelay tolerant networksrdquo Computer Communications vol 65pp 10ndash26 2015

[33] J Miao O Hasan S B Mokhtar L Brunie and G GianinildquoA delay and cost balancing protocol for message routing inmobile delay tolerant networksrdquo Ad Hoc Networks vol 25pp 430ndash443 2015

[34] H Chen and W Lou ldquoContact expectation based routing fordelay tolerant networksrdquo Ad Hoc Networks vol 36pp 244ndash257 2016

[35] F Xia L Liu J Li J Ma and A V Vasilakos ldquoSocially awarenetworking a surveyrdquo IEEE Systems Journal vol 9pp 904ndash921 2015

[36] W Moreira P Mendes and S Sargento ldquoOpportunisticrouting based on daily routinesrdquo in Proceedings of the Worldof wireless mobile and multimedia networks (WoWMoM) SanFrancisco CA USA June 2012

[37] W Moreira P Mendes and S Sargento ldquoSocial-aware op-portunistic routing protocol based on userrsquos interactions andinterestsrdquo in Proceedings of the International Conference onAd Hoc Networks vol 129 pp 100ndash115 Springer In-ternational Publishing Barcelona Spain August 2014

[38] Website of CRAWDAD project httpcrawdadorgcambridgehaggle20090529

[39] A Keranen J Ott and T Karkkainen ldquoe ONE simulatorfor DTN protocol evaluationrdquo in Proceedings of the 2nd In-ternational Conference on Simulation Tools and TechniquesBrussels Belgium March 2009


Computer Games Technology

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

Advances in

FuzzySystems


Volume 2018


ReconfigurableComputing



Applied Computational Intelligence and Soft Computing

thinspAdvancesthinspinthinsp

thinspArtificial Intelligence

Hindawiwwwhindawicom Volumethinsp2018


Civil EngineeringAdvances in


Electrical and Computer Engineering

Journal of

Journal of

Computer Networks and Communications


Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia


Biomedical Imaging



Engineering Mathematics


RoboticsJournal of



Computational Intelligence and Neuroscience


Mathematical Problems in Engineering

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018


Human-ComputerInteraction

Advances in


Scientic Programming

Submit your manuscripts atwwwhindawicom

friendship to improve routing efficiency in the opportunisticmobile social networkis is because social features are morestable and less changeable than other features like mobilitypatterns HiBOp [22] and CiPRO [23] exploit the similaritysocial feature and the userrsquos context information to forwarddata LASS [24] takes into account the difference of membersrsquoactivity within the nodersquos community for data disseminationIn ML-SOR [25] node centrality (different types of central-ities) the similarity between communities and social ties areall exploited to effectively select the forwarding node MCAR[26] exploits the preferred communities of people during theirdaily lives for effective information delivery Direct (inside onecommunity) and indirect (via different communities) contactsare considered in MCAR In SPRINT-SELF [27] network andnode overhead is decreased by exploiting social information ofmobile users e authors consider the social community ofnodes and utilize it to predict future behavior based on contacthistory In addition they proposed a newmechanism to avoidselfish nodes for more improvements

Social features are utilized widely for buffer managementLiu et al [28] utilized social features and the congestion levelto develop the forwarding strategy that drops the messagewith the minimum social link rather than random droppingIn SRAMSW [29] the buffer management mechanism iscombined with social features to enhance the spray-basedrouting Expired messages are deleted successfully deliveredmessages are acknowledged and messages are prioritizedaccording to their spray times and residence time In additionthree social features centrality similarity and friendship areadopted for better forwarding decision and to avoid thedead-end problem

We hypothesize that combining social features with theEpidemic-based forwarding scheme improves the efficiencyof routing in the OMSN In this paper we present anEpidemic-based Social-based routing protocol (EpSoc) thatcombines the advantages of the forwarding strategy used inthe Epidemic routing protocol with the positive impact ofexploiting social features EpSoc exploits the degree centralitysocial feature to adapt the time to live (TTL) of the routedmessage If a message is forwarded to a node which has higherdegree centrality (socially active node) its TTL value will bedecreased If these messages are dropped in the active nodewhen TTL is zero the blocking mechanism is used to rejectreceiving replications of these messages

e rest of this paper is organized as follows the nextsection reviews the related work We describe in detail ourproposed algorithm EpSoc in Section 3 In Section 4 weintroduce the performance evaluation and the discussionresults Finally Section 5 concludes the paper

2 Related Works

Delivering data to the destination at the right time withminimum resources is an optimal condition for any givenrouting protocols Epidemic has optimal performance interms of delivery ratio and latency However it suffers fromhigh overhead cost One of the solutions is to exploit socialinformation to improve Epidemic-based routing protocolsin the OMSN Degree centrality is a social feature that is

exploited widely in the literature to improve routing in themobile social networks For example Bubble Rap [11] utilizesthe mobile nodersquos degree centrality to provide cost-effectiverouting as compared to Epidemic routing Bubble Rap ex-ploits two social and structural metrics namely centrality andcommunity It selects high centrality nodes and communitymembers of destination as relays In the Bubble Rap algo-rithm nodes belong to different sizes of communities andhave different levels of popularity (ie rank) Each node isassumed to have two rankings global denotes the popularity(ie connectivity) of the node in the entire society and localdenotes the popularity within its community Messages areforwarded to nodes that have higher global ranking until anode in the destinationrsquos community is found en themessages are forwarded to nodes having a higher local rankingwithin the destinationrsquos community

Similar to Bubble Rap CAOR [30] exploits degree cen-trality and similarity social features to improve routingHowever CAOR constructed autonomous communities basedon common interest locations between mobile nodes Mem-bers of a community with high centralities act as the home ofthis community CAOR also has a mechanism to turn therouting between lots of nodes to the routing between a fewcommunity homes It also applies the reverse Dijkstra algo-rithm to determine the optimal relays and compute theminimum expected delivery delay In [31] cultural algorithm(CA) ant colony optimization (ACO) and social connectivitybetween users are combined to address the routing problemSocial metrics of nodes including degree and betweennesscentralities are analyzed to support forwarding decision in theopportunistic network environment

Besides exploited in the routing problem social featureswere also applied to solve different type of problems Forexample works in [32] exploit social features (degree andbetweenness centralities) to solve the throwbox placementproblem based on the given graph A userrsquos degree is equal tothe total number of its neighbors and betweenness is thetotal number of shortest paths passing through the nodeework also introduced the concept of location degree tomeasurehow many mobile users have the location as one of their topvisited locations and the location betweenness to show howimportant the location is for the entire social graph Socialmetrics such as degree centrality social activeness and com-munity acquaintance are also applied to enhance data deliveryin VSNs in [33] e social interactions between nodes wherenodes of similar interests or nodes belonging to the samecommunity have greater probability to encounter each otherSocial centrality is also utilized for congestion control in thepostdisaster environment [34]

Unlike the aforementioned works instead of consideringcommunity or homing structuring our proposed protocol isbased on the flooding-based forwarding strategy of Epi-demic is is because we intended to design our forwardingprotocol to have high delivery ratio and low delay as theEpidemic-based forwarding strategy Moreover we alsoaimed to avoid extra time required to form and maintain thecommunity structure us to decrease the overhead weutilized degree centrality and adapted the messagersquos TTL(Time to live) to control the forwarding nodesrsquo activity e






DCi 1113944N

k1

a(i k) (1)







TTLnew TTLoldDCk

if DCk gtDCi1113864 1113865 (2)























N2

N1 N3


(a)


register


N1 N2

N4 N6

N5


(b)





larrmiTTLDCj















0 10 20 30 40 50 600

005

01

015

02

025

03

035

04

Buffer size (MB)


Del

iver

y ra

tio


Buffer size (MB)


Ove

rhea

d ra

tio

0 10 20 30 40 50 600

100

200

300

400

500

600

700










0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

Del

iver

y ra

tio

002

004

006

008

01

012

014

016

018

02

TTL (minutes)


Ove

rhea

d ra

tio

0

50

100

150

200

250

300

350

400


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

TTL (minutes)



Aver

age l

aten

cy

0 10 20 30 40 50 600

1

2

3

4

5

6

7(times104)

Buffer size (MB)



Aver

age h

op co

unt

0 10 20 30 40 50 601

2

3

4

5

6

7

8

Buffer size (MB)






5 Conclusion


Additional Points




References






Aver

age l

aten

cy


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0TTL (minutes)

0

05

1

15

2

25

3

35(times104)


Aver

age h

op co

unt


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

TTL (minutes)

1

2

3

4

5

6

7

8

9

10













































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in








DCi 1113944N

k1

a(i k) (1)







TTLnew TTLoldDCk

if DCk gtDCi1113864 1113865 (2)























N2

N1 N3


(a)


register


N1 N2

N4 N6

N5


(b)





larrmiTTLDCj















0 10 20 30 40 50 600

005

01

015

02

025

03

035

04

Buffer size (MB)


Del

iver

y ra

tio


Buffer size (MB)


Ove

rhea

d ra

tio

0 10 20 30 40 50 600

100

200

300

400

500

600

700










0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

Del

iver

y ra

tio

002

004

006

008

01

012

014

016

018

02

TTL (minutes)


Ove

rhea

d ra

tio

0

50

100

150

200

250

300

350

400


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

TTL (minutes)



Aver

age l

aten

cy

0 10 20 30 40 50 600

1

2

3

4

5

6

7(times104)

Buffer size (MB)



Aver

age h

op co

unt

0 10 20 30 40 50 601

2

3

4

5

6

7

8

Buffer size (MB)






5 Conclusion


Additional Points




References






Aver

age l

aten

cy


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0TTL (minutes)

0

05

1

15

2

25

3

35(times104)


Aver

age h

op co

unt


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

TTL (minutes)

1

2

3

4

5

6

7

8

9

10













































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in
















N2

N1 N3


(a)


register


N1 N2

N4 N6

N5


(b)





larrmiTTLDCj















0 10 20 30 40 50 600

005

01

015

02

025

03

035

04

Buffer size (MB)


Del

iver

y ra

tio


Buffer size (MB)


Ove

rhea

d ra

tio

0 10 20 30 40 50 600

100

200

300

400

500

600

700










0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

Del

iver

y ra

tio

002

004

006

008

01

012

014

016

018

02

TTL (minutes)


Ove

rhea

d ra

tio

0

50

100

150

200

250

300

350

400


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

TTL (minutes)



Aver

age l

aten

cy

0 10 20 30 40 50 600

1

2

3

4

5

6

7(times104)

Buffer size (MB)



Aver

age h

op co

unt

0 10 20 30 40 50 601

2

3

4

5

6

7

8

Buffer size (MB)






5 Conclusion


Additional Points




References






Aver

age l

aten

cy


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0TTL (minutes)

0

05

1

15

2

25

3

35(times104)


Aver

age h

op co

unt


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

TTL (minutes)

1

2

3

4

5

6

7

8

9

10













































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in










0 10 20 30 40 50 600

005

01

015

02

025

03

035

04

Buffer size (MB)


Del

iver

y ra

tio


Buffer size (MB)


Ove

rhea

d ra

tio

0 10 20 30 40 50 600

100

200

300

400

500

600

700










0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

Del

iver

y ra

tio

002

004

006

008

01

012

014

016

018

02

TTL (minutes)


Ove

rhea

d ra

tio

0

50

100

150

200

250

300

350

400


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

TTL (minutes)



Aver

age l

aten

cy

0 10 20 30 40 50 600

1

2

3

4

5

6

7(times104)

Buffer size (MB)



Aver

age h

op co

unt

0 10 20 30 40 50 601

2

3

4

5

6

7

8

Buffer size (MB)






5 Conclusion


Additional Points




References






Aver

age l

aten

cy


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0TTL (minutes)

0

05

1

15

2

25

3

35(times104)


Aver

age h

op co

unt


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

TTL (minutes)

1

2

3

4

5

6

7

8

9

10













































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in











0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

Del

iver

y ra

tio

002

004

006

008

01

012

014

016

018

02

TTL (minutes)


Ove

rhea

d ra

tio

0

50

100

150

200

250

300

350

400


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

TTL (minutes)



Aver

age l

aten

cy

0 10 20 30 40 50 600

1

2

3

4

5

6

7(times104)

Buffer size (MB)



Aver

age h

op co

unt

0 10 20 30 40 50 601

2

3

4

5

6

7

8

Buffer size (MB)






5 Conclusion


Additional Points




References






Aver

age l

aten

cy


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0TTL (minutes)

0

05

1

15

2

25

3

35(times104)


Aver

age h

op co

unt


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

TTL (minutes)

1

2

3

4

5

6

7

8

9

10













































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in







5 Conclusion


Additional Points




References






Aver

age l

aten

cy


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0TTL (minutes)

0

05

1

15

2

25

3

35(times104)


Aver

age h

op co

unt


0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1000

0

TTL (minutes)

1

2

3

4

5

6

7

8

9

10













































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in














































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in




epsoc:social-basedepidemic-basedroutingprotocolin...

Documents