research article a novel contribution-aware neighbor

Research ArticleA Novel Contribution-Aware Neighbor-Assist Video DeliverySolution over Mobile Content-Centric Networks

Lujie Zhong1 Shijie Jia23 and Mu Wang4

1 Information Engineering College Capital Normal University Beijing 100048 China2Central Plains Economic Zone Wisdom Tourism Cooperative Innovation Center Luoyang Normal UniversityLuoyang Henan 471934 China3Academy of Information Technology Luoyang Normal University Luoyang Henan 471934 China4State Key Laboratory of Networking and Switching Technology Beijing University of Posts and TelecommunicationsBeijing 100086 China

Correspondence should be addressed to Shijie Jia shjjiagmailcom

Received 13 September 2016 Accepted 10 November 2016

Academic Editor George Ghinea

Copyright copy 2016 Lujie Zhong et al 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

In this paper we propose a novel contribution-aware neighbor-assist video delivery solution over mobile content-centric network(CNVD) CNVD allows the nodes to build and maintain neighbor relationship with other nodes to share cached video resourcesand achieve unicast-based video lookup CNVD constructs an estimationmodel of contribution of neighbor nodes by investigationof lookup delay number of cached videos lookup success rate and geographical distance CNVD designs the estimation methodsof interest level and lookup capacity of nodes for video content which enables the nodes to decide whether to build neighborrelationship with other nodes A maintenance method of neighbor relationship between nodes is proposed which enables thenodes to update valid time of neighbor relationship in terms of contribution of neighbor nodes and decide whether to removeneighbor relationship in terms of the current valid time of neighbor relationship Further CNVD designs a contribution-basedvideo lookup algorithm which reduces lookup delay and improves lookup success rate Extensive tests show how CNVD achievesmuch better performance results in comparison with other state-of-the-art solutions

1 Introduction

The prominent advancement of bandwidth and networkingtechnologies in wireless mobile networks greatly promotesdevelopment of mobile Internet applications such as sociale-commerce and multimedia [1] The video streaming ser-vices rely on provision of rich visual content and convenienceof access using mobile devices to become the most popularapplications [2ndash5] (eg the mobile video users in Chinahave been four hundred and forty million in 2016) Thevideo services focus on user quality of experience (QoE)[6] Rich visual content and convenient access for videoservices in wireless mobile networks attract large amount ofusersThe large-scale video access consumesmassive networkbandwidth and results in strong competition between usersfor the network bandwidth which brings severely negativeinfluence for startup delay of usersThedynamic and complex

condition of video delivery in wireless heterogeneous net-works increases the probability of video data loss so that thedistorted frames reduce user QoE Obviously the increase inupload bandwidth supply and video delivery performance isvery important for improving user QoE By the investigationof node mobility to address the problems of dynamic pathof data transmission the P2PMP2P-based video systemsmake use of the resources of online video users to increasethe efficiency and capacity of bandwidth supply in wirelessmobile networks [7ndash10] However the limited resourcesof bandwidth computation storage and energy of mobiledevices result in the limited available resources in overlaynetworks which difficultly meets the demand for increasinghuge video traffic

Content-centric networking (CCN) a brand-new frame-work focuses content based on newly designed protocolstack instead of host and employs the all-to-cache method

Hindawi Publishing CorporationMobile Information SystemsVolume 2016 Article ID 5352165 12 pageshttpdxdoiorg10115520165352165

2 Mobile Information Systems

Internet

Media server

Video namePITNullCS

FIB Interface WiFi

Video name NullPITCS

FIB Interface WiFi


FIB Interface WiFi

Video namePITNullCS

FIB Interface WiFi

Video namePITNullCS

FIB Interface WiFi

Interest packet flow Datainterest packet flow discardedData packet flow

V1

V1

V1

V1

V1

Figure 1 Multimedia streaming services in mobile content-centric networks

to achieve nearby content fetching [11] which reduces delayof content lookup and delivery Figure 1 illustrates con-tent delivery process in mobile content-centric networks(MCCNs) Each mobile node in MCCNs maintains the datastructures content store (CS) pending interest table (PIT)and forwarding information base (FIB) [12] The mobilenodes send interest packets to their neighbor nodes in orderto obtain the desired video data If the neighbor nodeshave cached the requested video data in CS the data isdirectly returned to the request users to the request nodesOtherwise if the neighbor nodes do not cache the requestedvideo data locally the mobile nodes record the incominginterface of interest packets in PIT and broadcast to allneighbors if themobile nodes have recorded the informationof interest packets they discard the received interest packetsIf the neighbor nodes firstly receive the interest packets theyforward the interest packets to the next-hop nodes if themobile nodes receive the same interest packets they discardthe received packetsWhen the interest packets are forwardedto the content providers which cache the requested videodata the providers return the video data along the reversesearching path At themoment the intermediate nodes (relaynodes) in the data reverse path can cache the returned data inorder to have nearby access to the data in the future

The all-to-cache the traditional caching method inMCCNs consumes large number of storage resources ofnodes to support nearby content fetching in order to offloadtraffic in underlying network which leads to huge waste ofstorage resourcesThe low capacities of storage computationand energy of mobile devices difficultly support the massiveconsumption On the other hand because the nodes are the

content carriers the content movement in geographic areawith movement of carriers brings huge negative influencefor content caching effectiveness Moreover the traditionalbroadcasting-based content lookup approach inMCCNs alsowastes huge network bandwidth The network congestioncaused by large-scale request results in long startup delayIn order to address the problems that existed in traditionalmethods some unicast-based content delivery solutions inMCCNs make use of the collected and maintained contentinformation to achieve precise and economic content lookupHowever the nodes need tomaintain large number of contentcarriers in order to fast search content providers The largerthe scale of maintained information is the higher the prob-ability of fast and successful lookup is The maintenance oflarge-scale content information not only brings redundancylink between nodes but also leads to the overload of mobilenodes due to low capacities Otherwise the small-scalemaintenance reduces lookup hit rate which also results inlong startup delay In order to ensure high QoE of users andreducemaintenance cost of nodes a key issue is how to designthe economic and efficient content delivery solutions

In this paper we propose a novel contribution-awareneighbor-assist video delivery solution over mobile content-centric networks (CNVD) CNVD allows the mobile nodes tobuild and maintain neighbor relationship with other mobilenodes to share cached videos and support unicast-basedvideo lookup In order to achieve economic maintenancefor the video content in MCCNs to support efficient lookupand transmission of videos CNVD designs an estimationmethod of contribution level of neighbor nodes by inves-tigation for lookup success rate number of cached videos

Mobile Information Systems 3

forwarding delay of interest packets and video data andgeographical distance of selected neighbor nodes in lookupand transmission paths A maintenance strategy of neighbornodes composed of construction and removal of neighborrelationship in terms of contribution levels of neighbor nodesis proposed which reduces maintenance cost of neighbornodes Further CNVD designs a contribution-based videolookup algorithm which decreases the delay of lookup andtransmission of videos by selection of appropriate next-hopnodes in terms of contribution levels of neighbor nodesSimulation results show how CNVD achieves much betterperformance results in comparison with other state-of-the-art solutions

2 Related Work

The content lookup in MCCNs [12 13] mainly employs thebroadcasting-based method When the content requesterswant to search content they broadcast interest packets inthe whole network The mobile nodes which receive theinterest packets of requesters become the relay nodes in thelookup path of content If the relay nodes cannot store therequested content they continue to broadcast the receivedinterest packets to their neighbor nodes The broadcasting-based method consumes large amount of network band-width which increases risk of network congestion and wastesbattery energy of mobile nodes In order to reduce thecost of content lookup Rehman et al proposed a timer-based interest forwarding (REMIF) method [14] After therelay nodes in the lookup path of content receive the inter-est packets they monitor the channel state for a specifictime If the duplicated interest packets are received withinthe time window the relay nodes discard the duplicatedinterest packets in terms of the recorded information inPIT Otherwise the relay nodes forward interest packets toother neighbor nodes However the time window resultsin the extra delivery latency which is unsuitable for videostreaming services Yu et al proposed a neighborhood-awareinterest forwarding method (NAIF) [15] which forwardsinterest packets according to statistics information of interestforwarding The relay nodes in the lookup path of contentdecide to broadcast or drop the received interest packetsaccording to the forwarding rate By the adaptive adjustmentof forwarding rate NAIF promotes efficiency of interestforwarding and reduces consumption of network bandwidthHowever NAIF does not consider selection of appropriatenext-hop neighbor nodes If the interest packets always areforwarded to the nodes which cache the requested contentwith high probability in the process of content routingthe interest packets quickly are satisfied which reduces thecontent lookup delay Otherwise the interest packets stillneed to experience long-term forwarding which increasesthe lookup delay and reduces user QoE The performanceof interest forwarding easily is influenced by the selection ofneighbor nodes

In order to further promote efficiency of interest for-warding several content lookup methods employ unicast-based interest forwarding Bian et al proposed a geo-basedforwarding strategy for NDN in VANETs environment [16]

The location position information of data source is addedinto data name in the process of naming data By peri-odical exchange of location information among one-hopneighbor nodes if the nodes receive interest packets theyselect the forwarders with close geographical distance whichreduces delay of interest forwarding Qian et al proposed aprobability-based adaptive forwarding scheme (PAF) basedon the ant colony optimization (ACO) [17] PAF makesuse of the ACO to calculate the probability of next-hoprelay nodes according to the performance measurementsuch as delay The better the results of performance mea-surement are the higher the selection probability of next-hop nodes is The probability-based selection of next-hopnodes improves the delivery quality and balances the load ofinterest forwarding among nodes The centrality-based datadissemination method is proposed in [18] By investigationfor social contact patterns and interests of mobile users tomeasure the social centrality of nodes the nodes whichreceive interest packets makes use of unicast or multicast toforward the interest packets to next-hop nodes in terms ofcentrality values The selected next-hop nodes have the highcentrality value and have high probability of encounteringnodes which also achieves high performance of lookup anddissemination of data In fact the centrality-based methodrelies on a priori knowledge of centrality values of mobilenodes The mobile nodes need to continuously calculate andexchange centrality values of other encountered nodes whichconsumes large amount of resources ofmobile nodes Ahmedet al proposed a unicast-based forwarder selection (RUFS)in order to address the problems caused by interest broadcaststorm [19] RUFS requires each vehicle node to share statisticinformation of satisfied interests with neighbor nodes Allneighbor nodes maintain local neighbors satisfied list (NSL)to store cached content information which efficiently pro-motes content lookup performance by selection of optimalinterest forwarder However the mobile nodes have the lowcapacities in the resources of energy computation storageand bandwidth they do not bear high-cost maintenancefor large-scale items in NSL Otherwise the small-scalemaintenance of cached content information increases therisk of content lookup failure Obviously the unicast-basedmethods promote content lookup performance and reducethe consumption of bandwidth in the process of interestforwarding the key issue is how to balance content-awarecost and lookup performance for the content delivery inMCCNs

3 Model of Contribution Capacity

31 Estimation of Contribution Capacity of Neighbor NodesFor convenience some notations which are used in currentand following sections are defined in Notations The main-tenance of neighbor nodes supports forwarding messages todestination nodes so the selection of relay nodes is veryimportant for the forwarding performance The efficientselection of neighbor nodes in MCCNs not only speedsup convergence process of video lookup but also promoteslookup hit rate CNVD requires each node to maintain servallogical neighbor nodes in order to support video lookup


(these neighbor nodes are not always one-hop neighbors ingeographic area) For instance 119899119894 makes use of the mappingrelationship between cached content and nodes to build andmaintain a neighbor node list NL119894 = (119899119886 119899119887 119899119896) When119899119894 needs to watch video content it sends an interest packet toan item 119899ℎ in NL119894 119899ℎ shoulders the responsibility of lookup ofrequested content 119899ℎ checks local cached videos and searchesvideo information carried by neighbor nodes of 119899ℎ If 119899ℎand 119899ℎrsquos neighbor nodes do not have the requested video119899ℎ selects a neighbor node as next-hop node to forward theinterest packetTheneighbor node selection is very importantfor the video lookup performance If the nodes cache largeamount of videos in local buffer the request for videos maybe responded by them with high probability If the nodesmaintain large amount of neighbor nodes they are awareof the information of videos cached in neighbor nodeswhich increases the probability of successful response forthe requested videos In order to obtain high success rateand performance of content lookup CNVD constructs anestimation model of contribution capacity of neighbor nodesto estimate the performance of content lookup of neighbornodes

The delay-sensitive video services have high requirementfor the content delivery performance (lookup and transmis-sion of content) The startup delay of video requesters is animportant evaluation parameter for the quality of service(QoS) of video systems The startup delay includes lookupand transmission delay of video data which is the timespan from sending interest packets to receiving video data ofsupporting video playback The requesters always hope thatthe startup delay meets the requirement of the own QoE Let119889119906119894 denote the upper bound of startup delay of 119899119894 in termsof the requirement of 119899119894rsquos QoE (119889119906119894 is the maximum value ofsustainable startup delay of 119899119894) 119899ℎ is a neighbor node of 119899119894When 119899ℎ receives video request of 119899119894 and helps 119899119894 search videoprovider 119889119903ℎ is defined as the generated startup delay in theprocess of lookup and transmission of video data 119889119903ℎ is thereal startup delay of 119899119894 If 119899ℎ enables 119889119903ℎ to be less than 119889119906119894 119899119894considers that 119899ℎ successfully helps 119899119894 search the requestedvideo 119899119894 continues to maintain the logical link between 119899119894and 119899ℎ Otherwise if nℎ receives the interest packet of 119899119894 andcannot make 119889119903ℎ in [0 119889119906119894] 119899119894 considers that the lookup taskassigned by 119899119894 for 119899ℎ is failure 119899119894 needs to consider whetherto remove the logical link between 119899119894 and 119899ℎ 119899119894 considersthe logical link between 119899119894 and 119899ℎ is valueless for the lookupfailure so that the maintenance of logical link between themwastes valuable resources of bandwidth computation andenergy of 119899119894Therefore wemake use of119889119906119894 and119889119903ℎ to calculatecontribution value of a lookup task assigned by 119899119894 for 119899ℎaccording to the following equation

119862ℎ (VC119895) =

119908ℎ119903ℎ (1 minus 119889119903ℎ119889119906119894 ) 119889119903ℎ lt 119889119906119894

0 119889119903ℎ ge 119889119906119894(1)

where VC119895 is the category of requested video V119886 and V119886 isinVC119895 which points out the limited range for video lookup119862ℎ(VC119895) is the contribution value of 119899ℎ for the assignedlookup task corresponding to VC119895 119889119903ℎ lt 119889119906119894 denotes that

current video lookup is successful for 119899ℎ 1 minus 119889119903ℎ119889119906119894 is thedistance ratio between 119889119903ℎ and 119889119906119894 where 1 minus 119889119903ℎ119889119906119894 isin (0 1)The less the value of 119889119903ℎ is the higher the contribution valueof 119899ℎ is Otherwise 119889119903ℎ ge 119889119906119894 denotes that current videolookup is failure for 119899ℎ The contribution value of 119899ℎ is 0 Forinstance 119899119894 sends an interest packet to the neighbor node 119899ℎfor a video content V119886 at the time 119905119904 When 119899119894 receives therequested data at the time 119905119903 it calculates the real startupdelay by 119889119903ℎ = 119905119903 minus 119905119904 and estimates contribution valueof 119899ℎ according to (1) In MCCNs there are the two maininfluencing factors for the startup delay of video requestersnumber of relay nodes in lookup path and delivery capacityof interest packets and video data of these relay nodes Thenumber of relay nodes determines the forwarding frequencyof interest packets and video data The more the number ofrelay nodes is the longer the forwarding delay of interestpackets and video data is which results in long startup delayThe selection of relay nodes (neighbor nodes) is a key factorfor reducing startup delay If the neighbor nodes cache largeamount of videos related to the requested videos the demandof requesters is satisfied with high probability Otherwiseif the neighbor nodes cache small amount of videos andthese cached videos are not related to the requested videosthe interest packets still continue to be forwarded whichincreases the lookup delay On the other hand because therelay nodes in the paths of lookup and transmission of videodata are the logical neighbor nodes with each other thegeographical distance and communication quality betweenrelay nodes are neglected The long geographical distancebetween relay nodes may increase the delay of lookup andtransmission of video data the low communication qualitybetween relay nodes (eg network congestion) not onlyresults in the increase in delay of lookup and transmissionof video data but also causes the loss of interest packets andvideo data In order to investigate the influence of supply anddelivery capacity of neighbor nodes for the performance oflookup and delivery of video data we add the two impactfactors 119908ℎ isin [0 1] and 119903ℎ isin [0 1] for 1 minus 119889ℎ119903119889ℎ119897 119908ℎ and119903ℎ denote the weight values generated by capacity of videosupply and delivery of neighbor nodes respectively

32 Estimation of Video Supply Capacity of Neighbor NodesThe video supply capacity of neighbor nodes is very impor-tant for reducing the length of paths of video lookup andtransmission inMCCNsWefirstly investigate lookup successrate (LSR) and number of cached videos (NCVs) to calculatethe values of 119908ℎ of neighbor nodes The LSR reflects theresource supply capacity of neighbor nodes for the videorequesters If a neighbor node 119899ℎ always meets the demand ofrequesters for the requested videos 119899ℎ has high video supplycapacity to help 119899119894 fast search the requested videos whichreduces the lookup delay There is a close relation betweenLSR and NCV The number of cached videos maintained by119899ℎ includes local videos of 119899ℎ and videos cached in neighbornodes of 119899ℎThemore the number ofmaintained videos is thehigher the probability of request satisfied in the maintainedvideos is The investigation of LSR reflects the video supplycapacity of selected neighbor nodes On the other hand theNCV of 119899ℎ and 119899ℎrsquos neighbor nodes is the dynamic variation


with the interest for the video content The low capacityof storage computation bandwidth and energy makes themobile nodes only cache small amount of videos In order towatch new videos the mobile nodes need to remove the oldvideos and cache the new videos in local buffer with limitedsize The investigation of NCV reflects the influence level ofvariation of video distributionmaintained by 119899ℎ for the videosupply capacity of selected neighbor nodes Therefore wemake use of LSR and NCV of all relay nodes in lookup pathsto estimate supply capacity of selected neighbor nodes TheLSR of any neighbor node 119899ℎ of 119899119894 for a video category VC119895can be obtained according to the following equation

119877119894ℎ (VC119895) = 119877119873119904119877119873119905 (2)

where 119877119894ℎ(VC119895) is 119899ℎrsquos LSR for 119899119894rsquos request of videos inVC119895 119877119873119904 and 119877119873119905 denote successful and total lookupnumber respectively CNVD employs a provider-feedback-based lookup performance estimation method Because theproviders are the terminal point of lookup paths they cancollect the information of relay nodes in paths such as LSRand number of cached videos After the neighbor node 119899ℎof requester 119899119894 receives the interest packet 119899ℎ adds numberof cached videos corresponding to VC119895 and LSR of next-hopnode 119899119895 selected by 119899ℎ into the interest packet 119899119895 also adds theabove information into the interest packet After the iterationthe provider 119899119901 adds the collected information of relay nodesand the number of cached videos into the returned dataAfter 119899119894 receives the data it is aware of the supply capacityof relay nodes and provider 119899119894 obtains two datasets 119877119878119894 =(119877119894ℎ 119877119894119895 119877119894119901) and 119873119878119894 = (119873119877119894ℎ 119873119877119894119895 119873119877119894119901) 119877119894119895 and119873119877119894119895 denote LSR and NCV of relay nodes and provider in thepaths of lookup and transmission of video data respectivelyBecause the grey relational coefficient (GRC) canmeasure therelation level of variation process of two curves [20] wemakeuse of the GRC to estimate the relational level between 119877119878119894and 119873119878119894 The items in 119877119878119894 and 119873119878119894 are normalized accordingto the following equation

119909lowast (att) = 119909 (att) minus lowerattupperatt minus loweratt

119909lowast (att) isin [0 1] (3)

where att denotes the attribution of estimation parametersuch as LSR and NCV 119909(att) is the value of estimationparameter upperatt and loweratt are the upper and lowerbound of values of estimation parameter (minimum andmaximum values) The relational level between 119877119878119894 and 119873119878119894can be calculated according to the following equation

GRC (119877 119873119877) = 1sum 120579att |119909lowast (att) minus 1| + 1

GRC isin [0 1] (4)

where 120579att is the weight value of 119909lowast(att) GRC(119877 119873119877) is therelational value between RS119894 and 119873119878119894 which denotes therelational level of two curves composed of items in RS119894 and119873119878119894 The higher the GRC(119877 119873119877) is the more similar theinterests of relay nodes for cached videos in the lookup path

are For instance the LSR and NCV of relay nodes keepthe risefall trend which means that the variation process ofLSR and NCV meets the condition of risefall of LSR withincreasedecrease of NCV The relay nodes also have similarinterests for the content related to V119886 If the LSR of relaynodes keeps fallrise trend with increasedecrease of NCVthe relay nodes do not have the common interests for thecontent in VC119895 whichmay increase the risk of lookup failureWe use GRC(119877 119873119877) to assign the value of 119908ℎ namely 119908ℎ =GRC(119877 119873119877)33 Estimation of Video Delivery Capacity of Neighbor NodesExcept for the supply capacity of neighbor nodes we alsoinvestigate the video delivery capacity of relay nodes inlookup and transmission paths to estimate relational levelbetween geographical distance and forwarding delay betweenthemThe forwarding delay reflects the delivery performanceof interest packets and video data in lookup and transmissionpathsThe geographical distance and communication qualitybetween relay nodes are the main influencing factors for theforwarding delay The neighbor nodes form a logical lookuppath so the long geographical distance between logical relaynodes enables the interest packets or video data to experiencemany geographical relay nodes which increases forwardingdelay Generally the larger the geographical distance betweenlogical relay nodes is the longer the forwarding delay ofinterest packets and video data is [21ndash23] Moreover the lowcommunication quality between relay nodes also increasesthe forwarding delay of interest packets and video dataor causes loss of interest packets and video data evenif the two relay nodes have close geographical distanceThe investigation of forwarding delay between logical relaynodes denotes the video delivery capacity of neighbor nodesOn the other hand the geographical distance reflects thestability of mobility of relay nodes The mobility makes thegeographical distance continuously change which influencesthe delay of lookup and transmission of video data Theinvestigation of geographical distance between logical relaynodes denotes the influence level of node mobility forthe video delivery capacity of neighbor nodes Similarlyall nodes (the requester 119899119894 logical relay nodes and theprovider 119899119901) in the lookup and transmission paths add theirtimestamp of forwarding interest packets and geographicallocation into the interest packets 119899119901 collects the informationof timestamp and geographical location of all nodes Let119866119878119894 = ((119909119894 119910119894) (119909ℎ 119910ℎ) (119909119901 119910119901)) and 119879119878119894 = (119905119894 119905ℎ 119905119901)denote the set of timestamp and geographical locationrespectively where 119909119894 and 119910119894 denote the abscissa and verticalcoordinates of 119899119894 and 119905119894 is the timestamp of forwardinginterest packet of 119899119894 119899119901 also adds 119866119878119894 and 119879119878119894 into thereturned data When 119899119894 receives the requested data from 119899119901it calculates the forwarding delay and geographical distancebetween nodes according to the following equation

119889119894ℎ = 119905ℎ minus 119905119894 gd119894ℎ = radic(119909119894 minus 119909ℎ)2 + (119910119894 minus 119910ℎ)2 (5)

where 119889119894ℎ denotes the forwarding delay of interest packetsbetween 119899119894 and 119899ℎ gd119894ℎ denotes the geographical distancebetween 119899119894 and 119899ℎ119866119878119894 and119879119878119894 are converted to GD119894 and119879119863119894


namely GD119894 = (gd119894ℎ gdℎ119896 gdℎ119896) and 119879119863119894 = (119889119894ℎ 119889ℎ119896 119889119897119901) 119899119894 makes use of (3) to normalize items in GD119894 and119879119863119894 and further makes use of (4) to calculate relationallevel between forwarding delay and geographical distancebetween GD119894 and 119879119863119894 We use GRC(119889 gd) to assign thevalue of 119903ℎ namely 119903ℎ = GRC(119889 gd) 119899119894 calculates andrecords the contribution of 119899ℎ for the lookup of V119886 Theforwarding delay and geographical distance between relaynodes keep the same risefall trend which means thatthe variation process of forwarding delay and geographicaldistance meets the condition of risefall of forwarding delaywith increasedecrease of geographical distance There is thegood communication quality between relay nodes If theforwarding delay between relay nodes keeps fallrise trendwith increasedecrease of geographical distance there is thebad communication quality between relay nodes such asnetwork congestion which brings high risk of packet loss andlong delayThe requesters should avoid the reusage of currentlookup path for the subsequent video lookup

4 CNVD Detailed Design

41 Construction of Neighbor Relationship The nodes main-tain the logical connections with their neighbor nodes inorder to fast fetch desired video content For instance if theneighbor nodes of a node 119899ℎ store large number of videoresources the videos requested by 119899ℎ may be cached by theneighbor nodes Obviously the nodes tend to preferentiallyconstruct the neighbor relationship with the nodes cachedlarge-scale resources However if a node 119899119894 caches largeamount of video resources and 119899ℎ is uninterested in thecached videos of 119899119894 119899119894 also does not meet the demand of 119899ℎOtherwise if 119899ℎ is interested in large amount of videos cachedby 119899119894 119899ℎ preferentially constructs the neighbor relationshipwith 119899119894 The large amount of cached videos and the similarinterests enable the request of 119899ℎ to be met by 119899119894 with highprobability 119899ℎ sends an invitation message to 119899119894 where themessage includes the information of videos cached in 119899ℎ and119899ℎrsquos neighbor nodes namely119881119868ℎ = (VL119886VL119887 VL119896) VL119886is a video list and includes the ID of videos correspondingto the video category VC119886 Because 119899ℎ is aware of theinformation of videos stored in neighbor nodes by messageexchange the videos stored in neighbor nodes also areconsidered as the available resources of 119899ℎ If 119899119894 has the highinterested degree (interest level) for videos cached in 119899ℎthe video request of 119899119894 is met by 119899ℎ with high probability119899119894 estimates the interest level for the videos cached in 119899ℎaccording to the following equation

119868119894ℎ = sum119904119888=119886 1003816100381610038161003816VL119888 minus UV1198881003816100381610038161003816

sum119904119888=119886 1003816100381610038161003816VL1198881003816100381610038161003816 119868119894ℎ isin [0 1] (6)

where 119904 is the number of video categories interested by 119899119894UV119888 is the set of videos uninterested by 119899119894 corresponding tothe video category VC119888 |VL119888 minus UV119888| returns the number ofitems in the difference set between VL119888 and UV119888 sum119904119888=119886 |VL119888 minusUV119888| returns the number of interested videos of 119899119894 119868119894ℎ ge TH119894denotes that 119899119894 is interested in the videos cached by 119899ℎ whereTH119894 is the threshold of interest level of 119899119894 119899119894 accepts the

invitation of 119899ℎ and constructs the neighbor relationship with119899ℎ Otherwise if 119868119894ℎ lt TH119894 denotes that 119899119894 is uninterested inthe videos cached by 119899ℎ 119899119894 rejects the invitation of 119899ℎ

The scale of video resources cached in 119899ℎ and 119899ℎrsquosneighbor nodes is limited If 119899119894rsquos interest for the video contentis out of range of resources cached in 119899ℎ and 119899ℎrsquos neighbornodes (the requested videos are not located in the set ofvideos cached in 119899ℎ and 119899ℎrsquos neighbor nodes) 119899119894 still needsto search the requested videos with the help of neighbornodes Except for the supply capacity of local resources theresource lookup capacity of nodes also is an important factorfor the construction of neighbor relationship For instance 119899ℎand 119899ℎrsquos neighbor nodes only cache small amount of videosbut 119899ℎrsquos neighbor nodes have strong lookup capacity (highcontribution values for lookup tasks assigned by 119899ℎ) 119899119894 alsomay consider the construction of neighbor relationship with119899ℎ Even if 119899ℎ and 119899ℎrsquos neighbor nodes do not provide one-hop and two-hop video access for 119899119894 they may successfullysearch the videos requested by 119899119894 and enable the startup delaymeet the demand of 119899119894rsquos QoE 119899119894 also accepts the invitationof 119899ℎ and constructs the neighbor relationship with 119899ℎ 119899ℎsends an invitation message to 119899119894 where the message includesthe contribution values of all neighbor nodes of 119899ℎ namely119878con = (CL119886CL119887 CL119896) CL119886 is the list of contributionvalues of neighbor nodes corresponding to the video categoryVC119886 119899119894 estimates the lookup capacity of 119899ℎ for the videocategory VC119886 according to the following equation

LP119886ℎ = sum119892119890=1 119862119890119892

119862119909 = sumV119887=1 119862119887V

(7)

where 119862119909 is the average contribution value of a neighbornode 119899119909 of 119899ℎ corresponding to VC119886 V is the number ofvideo lookup tasks assigned by 119899ℎ for the requested videosin VC119886 119892 is the number of nodes which accept the lookuptasks assigned by 119899ℎ for the videos in VC119886 If 119899119894 is interestedin multiple video categories it estimates the video lookupcapacity of 119899ℎ for multiple video categories according to thefollowing equation

LPℎ = sum119896119890=1 120596119890 times LP119890ℎ119896 120596119890 isin (0 1) (8)

where 120596119890 is the weight value of the video category VC119890namely there are different weigh values between videocategories 119896 is the number of video categories interested by119899119894 Let LP119894 be the video lookup capacity of 119899119894 by making useof (8) where LP119894 and LPℎ are corresponding to the samevideo categories If LPℎ gt LP119894 119899119894 constructs the neighborrelationship with 119899ℎ CNVD allows the nodes construct theneighbor relationship with other nodes according to theinterest level for the cached videos and video lookup capacitywith each other

42 Removal ofNeighborRelationship Thenodeswhich buildlogical links not only need to consumebandwidth tomaintain


the state with each other but also are responsible for thevideo lookup If a neighbor node 119899ℎ always requests 119899119894 tohelp search desired videos and does not provide satisfiedlookup performance for 119899119894 the logical link between 119899119894 and119899ℎ is insignificant and redundant for 119899119894 In order to save theresources of bandwidth computation and energy CNVDallows 119899119894 to remove the redundant logical link with 119899ℎ (notall links always are maintained) In order to construct orkeep the neighbor relationship with the nodes with stronglookup capacity and large amount of interested videos thenodes need to store more videos and maintain the links withmore nodes The link removal is the punishment of nodeselfishness but the video sharing performance in the wholenetwork is not reduced by link removal

The contribution of neighbor nodes is the importantmetric for themaintenance of logical linksThe cached videosand maintained links of 119899119894 and neighbor nodes provide videosupply servicewith each otherThe contribution values reflectthe video supply capacities of neighbor nodes We define atime 119879119894ℎ to denote valid period of link between 119899119894 and 119899ℎ 119899119894or 119899ℎ removes the link between them when the time span oflink is greater than 119879119894ℎ For instance if the contribution valueof 119899ℎ for 119899119894 is 0 (119899119894 does not require 119899ℎ search videos or allvideo lookup of 119899ℎ is fail) 119899119894 removes the link between 119899ℎand 119899119894 when the valid time of their link is greater than 119879119894ℎIf 119899ℎ successfully searches the video requested by 119899119894 and hasa corresponding contribution 119862ℎ(VC119895) the valid time of linkbetween them is defined as

VT119894ℎ = VTlowast119894ℎ + 119879119894ℎ times 119862ℎ (VC119895) 119862ℎ (119881119862119895) isin [0 1] (9)

whereVTlowast119894ℎ is the remaining time of link VT119894ℎ is current validtime of link (neighbor relationship)Obviously the higher thecontribution of neighbor nodes is the longer the valid timeof link is The maintenance method of neighbor relationshipbased on the valid time of link reduces the consumptionof resources of bandwidth computation and energy andpromotes the video sharing

43 Neighbor Nodes Discovery and Video Lookup Initially allnodes do not construct the neighbor relationship with othernodes The nodes send invitation messages to their one-hopnodes Because the nodes do not have neighbor nodes thevideo lookup capacity of inviters is 0 The one-hop nodesdecide whether to accept the invitation by (6) Once thetwo nodes construct neighbor relationship theymaintain thelogical link by periodical exchange of messages containinginformation of cached videos If the link is overtime at a nodeside the node removes current link

On the other hand if a node 119899119894 wants towatch a video V119886 isinVC119895 119899119894 checks the local cached videos because it is aware ofvideos cached in all neighbor nodes by message exchange Ifthe requested video is cached in a neighbor node 119899ℎ 119899119894makesuse of the valid link to send the interest packet to 119899ℎ After119899ℎ checks local cached videos 119899ℎ directly returns the videodata to 119899119894 If the requested video is not cached in all neighbornodes of 119899119894 119899119894 requests the help of neighbor nodes to searchthe requested video data 119899119894 selects a neighbor node 119899119886 as thenext-hop node where 119899119886 has the highest contribution value

corresponding to VC119895 among all neighbor nodes After 119899119886receives the interest packet it check its videos and the videoscached in neighbor nodes If 119899119886 and 119899119886rsquos neighbor nodes donot have the requested video 119899119886 also selects a neighbor nodeas the next-hop node according to the contribution value ofall neighbor nodes corresponding to VC119895 In order to avoidthe loop circuit in the lookup process the requesters andrelay nodes add the information of their neighbor nodes intothe interest packet After iteration of the above process ifthe content provider is found the provider returns the videodata to 119899119894 and 119899119894 updates the contribution of 119899119886 Otherwiseif the interest packet is overtime 119899119894 broadcasts the interestpacket The pseudocode of the process of video lookup isdetailed in Algorithm 1The number of relay nodes in lookuppath and the number of their neighbor nodes determinethe complexity of Algorithm 1 Therefore the complexity ofAlgorithm 1 is 119874(119899)

After the nodes successfully fetch the requested videoswith the help of neighbor nodes (the lookup delay meets therequirement of QoE of requesters) they record the infor-mation of providers If the nodes have sufficient resourcesof computation bandwidth and energy to maintain a newlink they send the invitation messages to the providers alongthe lookup path The providers decide whether to accept theinvitation according to the capacities of video supply andlookup of inviters If the providers reject the invitation theinviters remove the information of providers The rejectionof providers drives the inviters continuously to find moreappropriate neighbor nodes

5 Testing and Test Results Analysis

51 Testing Topology and Scenarios We compare the per-formance of CNVD with RUFS which is a state-of-the-artunicast-based CCN forwarding strategy [19] CNVD andRUFS were modeled and implemented in Network Simulator3 (NS-3) The more the number of mobile nodes is themore the scale of requested video data is which bringssevere network congestion In order to reduce the influenceof network congestion for the experiment effect 200 mobilenodes are considered as vehicular nodes and are distributedin a 2000 times 2000m2 square area which has five horizontaland five vertical streets with two lanes The bandwidthof mobile nodes is 10Mbs The mobility results in thevariation of geographical distance between mobile nodesto influence the delay of lookup and transmission of videodata The random movement behaviors cannot reflect thereal movement environment so the movement behaviors ofmobile nodes follow the Manhattan mobility model [24] Inorder to simulate the real urban environment the movementspeed varies from 15ms to 20ms 33 road side units (RSUs)are evenly deployed in the square area and provide initialvideo data for mobile nodes The mobile nodes and RSUequip IEEE 80211p WAVE network interface to support datatransmission The maximum transmission unit (MTU) ofnetwork is set to 1500 BThe size of content store (CS) in eachnode is 10000MTU which is almost equal to 5 of the totalsize of video content The signal range of mobile nodes is setto 250m The simulation time is 1000 s


(1) flag = 0 119895 = 0(2) lowast NL is neighbor set of node V119886 isin VC119895 is video content requested by requester

119899119894 RS is set of relay nodes in lookup pathlowast(3) for (119896 = 0 119896 lt |NL119894| 119896++)(4) if NL119894[119896] includes V119886(5) 119899119894 sends interest packet to NL119894[119896] flag = 1 break(6) flag = 1(7) break(8) end if(9) end for(10) if (flag = 0)(11) 119899119894 sends interest to neighbor RS[119895] with the most contribution for VC119895(12) while (flag = 1 or 119895 gt TTL)(13) if RS[119895]rsquos neighbor NL[ℎ] includes V119886(14) RS[119895] forwards interest to NL[ℎ](15) flag = 1(16) else RS[119895] forwards interest to RS[119895 + 1] with the most contribution for

VC119895 119895++(17) end if(18) end while(19) end if(20) if (flag = 0)(21) 119873119894 broadcasts interest(22) end if

Algorithm 1 Process of video lookup

We group 100 video files into 20 video categories wherethe length of each file is 100 s Before the simulation wecreated 200 playback logs to define playback behaviors of200 mobile nodesThe 200mobile nodes watch diverse videocontent according to the created 200 playback logs where thewatched time is random When the nodes have watched avideo they request new videos according to the playback logs200mobile nodes join the video system following the Poissondistribution In CNVD the valid time 119879 of link betweennodes is set to 20 sThe threshold of interest level of all nodesis set to 05 and the number of neighbor nodes maintained byeach node is in the range [1 10] 120579att is set to 05 The valuesof 120596 corresponding to all video categories are 05 The upperbound of startup delay of all request nodes is set to 5 s

52 Performance Evaluation The performance of CNVD iscompared with RUFS in terms of lookup latency cache hitratio playback freeze frequency and maintain overheadrespectively

521 Lookup Latency The lookup latency is defined as thetime span between the time when the requester sends theinterest packet and the time when the provider receives theinterest packet

We use mean values of all lookup latency during every20 s as the average lookup latency Figures 2 and 3 show theperformance of lookup latency of CNVD and RUFS in termsof the variation in simulation time and number of mobilenodes As Figure 2 shows the blue curve corresponding toCNVD first experiences fast increase before 119905 = 200 s anddecreases to the lowest point (2 s) at 119905 = 400 s The lookup

15

2

25

3

35

4

45

Look

up la

tenc

y (s

)

Simulation time (s)0 200 400 600 800 1000

CNVDRUFS

Figure 2 Lookup latency versus simulation time

latency of CNVD fluctuates around 3 s and keeps relativelystable after 119905 = 600 s The red curve corresponding to RUFShas a severe fluctuation before 119905 = 200 s slightly increasesfrom 119905 = 200 s to 119905 = 600 s and maintains slight fluctuationafter 119905 = 600 s Obviously the lookup latency of CNVD islower than that of RUFS

We use mean values of all lookup latency during every20 nodes as the average lookup latency Figure 3 shows thevariation of lookup latency of two solutionsCNVDandRUFSwhen the number of nodes increases from 20 to 200 Theblack bars corresponding to CNVD have slight rise from 20


005

115

225

335

445

Look

up la

tenc

y (s

)

Number of mobile nodes 20 40 60 80 100 120 140 160 180 200

CNVDRUFS

Figure 3 Lookup latency versus number of mobile nodes

to 140 after fast fall from 140 to 200 where the lookup latencyof CNVD is between 25 s and 3 sThe red bars correspondingto the RUFS also have increase trend after fast decrease Therange of lookup latency of RUFS is [28 35] The lookuplatency of RUFS is higher than those of CNVD

Initially the small amount of mobile nodes requests andcaches video content The RSUs provide the initial videoresources for the request nodes so that the lookup latencykeeps low levels With increasing number of request nodesthe increase in the scale of requested videos brings hugevideo traffic which leads to the network congestion andcauses the rise of lookup latency When the nodes havewatched all videos they quit the systemThedecreasing trafficrelieves the congestion level and enables the lookup latencyfall In CNVD the nodes build the neighbor relationshipbased on the capacity of video supply and lookup andmaintain the neighbor relationship in terms of forwardingdelay geographical distance lookup success rate and numberof cached videos The neighbor nodes with strong capacityof video supply and delivery reduce the amount of interestforwarding and decrease lookup latency The link removalmechanism drives the nodes to continuously find moreappropriate neighbor nodes with similar interests and strongcapacity of video supply and delivery which promotes thelookup performance of neighbor nodes Therefore CNVDcan enable the lookup latency to keep low level with slightjitter In RUFS the selection of relay nodes relies on theforwarding capacity of neighbor nodes However the for-warding capacity of neighbor nodes only investigates theopportunistic encounter with other nodes which cannotguarantee the validation of content location informationAdditionally RUFS neglects mobility of mobile nodes whichalso brings negative influence for video delivery performancenamely the dynamic network topology caused by nodemobility may result in frequent change of paths of packetforwarding which increases the lookup latency AlthoughCNVD does not consider the mobility of mobile nodes thenodes investigate the variation trend of geographical distanceand transmission latency in the process of maintenance of

0

005

01

015

02

025

03

035

04

CHR

Simulation time (s)

CNVDRUFS

0 200 400 600 800 1000

Figure 4 CHR versus simulation times

neighbor nodes Therefore the lookup latency of CNVD islower than that of RUFS

522 Cache Hit Ratio (CHR) The cache hit ratio is definedas

CHR = 119867119873119877119873 CHR isin [0 1] (10)

where 119877119873 is the total number of interest packets receivedby nodes 119867119873 is the number of requests satisfied by videoscached in nodes The high CHR denotes the video providerscan efficiently supply the cached videos to reduce the amountof unsuccessful lookup Figures 4 and 5 illustrate the resultsof CHR of CNVD and RUFS in terms of the variation insimulation times and number of mobile nodes respectively

We use mean values of all CHR during every 20 s as theaverage CHR As Figure 4 shows the two curves experiencesimilar rise trend from 119905 = 0 s to 119905 = 200 s The blue curvecorresponding to CNVD keeps rise from 119905 = 200 s to 119905 =400 s reaches the peak 036 at 119905 = 460 s and has a fall withslight fluctuation from 119905 = 400 s to 119905 = 1000 s The red curvecorresponding to RUFS has the decrease trend with slightfluctuation from 119905 = 100 s to 119905 = 1000 s The blue curve ofCNVD is higher than that of RUFS

We use mean values of all CHR during every 20 nodesas the average CHR As Figure 5 shows the black barscorresponding toCNVDhave a fast rise trendwith increasingnumber of mobile nodes where the increase rate is gradualfrom 100 to 200 The red bars corresponding to RUFS alsoshow the rise trend but the CHR results of RUFS always keepcontinuous jitter The CHR of CNVD is almost 20 higherthan that of RUFS

Initially the small number of mobile nodes joins thesystem and requests video content The sent interest packetsalso are less The RSUs provide the initial video data sothat the CHR values keep fast rise The video content isfast disseminated to the whole network with the help ofcontent caching With the increase in the number of requestmembers the number of requested videos also fast increases


0

005

01

015

02

025

03

CHR

Number of mobile nodes

CNVDRUFS

20 40 60 80 100 120 140 160 180 200

Figure 5 CHR versus number of mobile nodes

When the request nodes need to fetch videos from thenetwork instead of RSUs the CHR values decrease due tothe unbalanced distribution of content In CNVD the nodesbuild the neighbor relationship based on the capacity of videosupply and lookup The similar interests between neighbornodes ensure the neighbor nodes cache and request thesimilar videos The neighbor nodes with large amount ofcached videos can support the video request falling in thecontent cached in neighbor nodes with high probabilityThe investigation of lookup success rate and number ofcached videos in the process of maintenance of neighbornodes drives the nodes continuously to findmore appropriateneighbor nodes with higher lookup success rate and largernumber of cached videos which further promotes the sharingperformance between neighbor nodes and increases the CHRvalues Therefore CNVD can obtain high CHR RUFS onlycollects the content location information by opportunisticencounter with other nodes and does not ensure the mobilenodes always obtain the location information of requestedcontent Additionally the node mobility also leads to fastinvalidation of collected information Therefore the CHR ofRUFS fast decreases and keeps low levels after 119905 = 100 s

523 Playback Freeze Frequency The times of occurrence ofplayback freeze per second are used to denote the playbackfreeze frequency during the whole simulation time

We use mean values of all playback freeze frequencyduring every 20 s as the average playback freeze frequencyFigure 6 shows the variation of playback freeze frequency ofthe two solutions with increasing simulation timeThe curvesof CNVDandRUFS experience fast rise with slight jitter from119905 = 0 s to 119905 = 1000 s The increment of RUFSrsquos results is higherthan that of CNVD

In RUFS the nodes collect the location informationof video content by making use of opportunistic messageexchange The node mobility speeds up the invalidationof collected information Moreover the nodes only collectinformation of cached content of adjacent nodes so that thesmall-scale collection of content information increases the


CNVDRUFS

005

115

225

335

445

5

Play

back

free

ze fr

eque

ncy

Figure 6 Playback freeze frequency versus simulation time

005

115

225

335

445

Mai

nten

ance

ove

rhea

d (k

bs)


CNVDRUFS

Figure 7 Maintenance overhead versus number of mobile nodes

risk of lookup fail Therefore the playback freeze frequencyof RUFS keeps fast rise during the whole simulation namelythe QoE of nodes is relatively low In CNVD the nodes whichbuild the neighbor relationship have strong capacity of videosupply and lookup Moreover the nodes remove the logicallink with neighbor nodes in terms of the contribution bythe investigation for capacity of video supply and deliveryof neighbor nodes which promotes the sharing performancesuch as high lookup success rate and low lookup delayTherefore the playback freeze frequency of CNVD keeps lowincrement namely the nodes in CNVD can obtain highQoE

524 Maintenance Overhead The average bandwidth usedto maintain node state and exchange content informationevery second is defined as the maintenance overhead

Figure 7 shows the variation of maintenance overhead ofthe two solutions with the increase in the number of mobilenodes The black and red bars of CNVD and RUFS keep fastrise trend with the growth of system scale The black barscorresponding to CNVD keep lower increment than those of


RUFS even if the maintenance overhead of CNVD is higherthan that of RUFS from 100 to 120

In RUFS the nodes maintain the information of recentlysatisfied interests of one-hop neighbor nodes In order toensure validity of exchanged information there is the high-frequency exchange of messages between nodes ThereforeRUFS has high maintenance overhead In CNVD the upperbound of number of neighbor nodes is 10 The small numberof neighbor nodes does not bring high maintenance costMoreover in order to further reduce the maintenance costthe nodes remove the link with neighbor nodes in termsof the contribution Because the nodes which keep long-term neighbor relationship have high lookup performancethe limited number of neighbor nodes does not bring morenegative influence for the content lookup Therefore themaintenance overhead of CNVD is lower than that of RUFS

6 Conclusion

In this paper we propose a novel contribution-awareneighbor-assist video delivery solution over mobile content-centric network (CNVD) CNVD constructs the estimationmodel of contribution of neighbor nodes by investigation oflookup success rate number of cached videos forwardingdelay and geographical distance In order to ensure theneighbor nodes have high capacity of video supply andlookup before construction of neighbor relationship thenodes estimate interest levels for cached content measure thelookup performance and further decide whether to build theneighbor relationship In order to stimulate nodes to improveutilization rate of cached content and find more appropriateneighbor nodes with strong capacity of video supply andlookup CNVD designs a removal method of neighborrelationship The nodes which keep long-term neighborrelationship have strong capacity of video supply and lookupwhich promotes video sharing performance and ensures QoEof the request nodes CNVD designs a contribution-basedvideo lookup algorithm which achieves fast video lookupThe simulation results show how CNVD has lower lookuplatency high cache hit ratio lower playback freeze frequencyand lower maintenance overhead than RUFS

Notations

119899119894 Mobile node 119894VC119895 Video category 119895119862ℎ(VC119895) Contribution of 119899ℎ for video lookup of 119899119894 in VC119895119889ℎ119903 The real startup delay of 119899119894119889ℎ119897 The lower bound of startup delay of 119899119894rsquos QoE

requirement119908ℎ Weight value of video supply capacity of 119899ℎ119903ℎ Weight value of video delivery capacity of 119899ℎ119877119894119895 119899119895rsquos lookup success rate for video request of 119899119894119873119877119894119895 Number of videos cached by neighbor node 119899119895

of 119899119894119889119894ℎ Forwarding delay of interest packets or videodata between 119899119894 and 119899ℎ

gd119894ℎ Geographical distance between 119899119894 and 119899ℎ

VL119886 Video list including ID of videos in VC119886UV119888 A set of uninterested videos of 119899119894 for videos

in VC119888TH119894 Threshold of interested degree of 119899119894 for video

contentCL119886 Set of contribution values of neighbor nodes

of 119899119894 for videos in VC119886120596119890 Weight value of VC119890LP119886ℎ Estimation value of lookup capacity of 119899ℎ for

request of videos in VC119886VT119894ℎ Valid time of neighbor relationship between

119899119894 and 119899ℎCompeting Interests

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

Acknowledgments

This work was supported in part by the National NaturalScience Foundation of China (NSFC) under Grant nos61402303 and 61501216 and the Project of Beijing MunicipalCommission of Education underGrantKM201510028016 andthe Science and Technology Plan Projects (Openness andCooperation) of Henan Province (152106000048)

References

[1] S Chen and J Zhao ldquoThe requirements challenges andtechnologies for 5G of terrestrial mobile telecommunicationrdquoIEEE Communications Magazine vol 52 no 5 pp 36ndash43 2014

[2] L Zhou Z Yang Y Wen and J J P C Rodrigues ldquoDistributedwireless video scheduling with delayed control informationrdquoIEEE Transactions on Circuits and Systems for Video Technologyvol 24 no 5 pp 889ndash901 2014

[3] D Bethanabhotla G Caire and M J Neely ldquoAdaptive videostreaming for wireless networks with multiple users andhelpersrdquo IEEE Transactions on Communications vol 63 no 1pp 268ndash285 2015

[4] S Jia C Xu J Guan H Zhang and G-M Muntean ldquoA novelcooperative content fetching-based strategy to increase thequality of video delivery to mobile users in wireless networksrdquoIEEE Transactions on Broadcasting vol 60 no 2 pp 370ndash3842014

[5] C Xu S Jia L Zhong and G-M Muntean ldquoSocially awaremobile peer-to-peer communications for community multime-dia streaming servicesrdquo IEEE Communications Magazine vol53 no 10 pp 150ndash156 2015

[6] C Xu F Zhao J Guan H Zhang and G-M Muntean ldquoQoE-driven user-centric vod services in urban multihomed P2P-based vehicular networksrdquo IEEE Transactions on VehicularTechnology vol 62 no 5 pp 2273ndash2289 2013

[7] N Kumar and J-H Lee ldquoPeer-to-peer cooperative caching fordata dissemination in urban vehicular communicationsrdquo IEEESystems Journal vol 8 no 4 pp 1136ndash1144 2014

[8] C Xu S Jia M Wang L Zhong H Zhang and G-MMuntean ldquoPerformance-aware mobile community-based VoDstreaming over vehicular Ad Hoc networksrdquo IEEE Transactionson Vehicular Technology vol 64 no 3 pp 1201ndash1217 2015


[9] Y Zhao Y Liu C Chen and J Zhang ldquoEnabling P2P one-viewmultiparty video conferencingrdquo IEEE Transactions on Paralleland Distributed Systems vol 25 no 1 pp 73ndash82 2014

[10] C Xu S Jia L Zhong H Zhang and G-M MunteanldquoAnt-inspired mini-community-based solution for video-on-demand services in wireless mobile networksrdquo IEEE Transac-tions on Broadcasting vol 60 no 2 pp 322ndash335 2014

[11] V Jacobson D K Smetters J D Thornton M F Plass NH Briggs and R L Braynard ldquoNetworking named contentrdquoin Proceedings of the 5th International Conference on EmergingNetworking Experiments and Technologies (CoNEXT rsquo09) pp 1ndash12 ACM Rome Italy December 2009

[12] G Grassi D Pesavento G Pau R Vuyyuru R Wakikawa andL Zhang ldquoVANET via named data networkingrdquo in Proceedingsof the IEEE Conference on Computer Communications Work-shops (INFOCOMWKSHPS rsquo14) pp 410ndash415 Toronto CanadaApril 2014

[13] L Zhang A Afanasyev J Burke et al ldquoNamed data network-ingrdquo ACM SIGCOMM Computer Communication Review vol44 no 3 pp 66ndash73 2014

[14] R A Rehman T D Hieu H-M Bae S-H Mah and B SKim ldquoRobust and efficient multipath Interest forwarding forNDN-based MANETsrdquo in Proceedings of the 9th IFIP Wirelessand Mobile Networking Conference (WMNC rsquo16) pp 187ndash192Colmar France July 2016

[15] Y-T Yu R BDilmaghani S CaloMY Sanadidi andMGerlaldquoInterest propagation in named data manetsrdquo in Proceedingsof the International Conference on Computing Networking andCommunications (ICNC rsquo13) pp 1118ndash1122 San Diego CalifUSA January 2013

[16] C Bian T Zhao X Li and W Yan ldquoBoosting named datanetworking for efficient packet forwarding in urban VANETscenariosrdquo in Proceedings of the 21st IEEE International Work-shop on Local and Metropolitan Area Networks (LANMAN rsquo15)pp 1ndash6 Beijing China April 2015

[17] HQian R Ravindran GQWang andDMedhi ldquoProbability-based adaptive forwarding strategy in named data networkingrdquoin Proceedings of the IFIPIEEE International Symposium onIntegrated Network Management (IM rsquo13) pp 1094ndash1101 GhentBelgium May 2013

[18] W Gao and G Cao ldquoUser-centric data dissemination indisruption tolerant networksrdquo in Proceedings of the 30th IEEEInternational Conference on Computer Communications (IEEEINFOCOM rsquo11) pp 3119ndash3127 Shanghai China April 2011

[19] SHAhmed SH Bouk andDKim ldquoRUFS RobUst forwarderselection in vehicular content-centric networksrdquo IEEE Commu-nications Letters vol 19 no 9 pp 1616ndash1619 2015

[20] A Razzaq and A Mehaoua ldquoLayered video transmission usingwireless path diversity based on grey relational analysisrdquo inProceedings of the IEEE International Conference on Communi-cations (ICC rsquo11) pp 1ndash6 Kyoto Japan June 2011

[21] C Xu T Liu J Guan H Zhang and G-M Muntean ldquoCMT-QA quality-aware adaptive concurrent multipath data transferin heterogeneous wireless networksrdquo IEEE Transactions onMobile Computing vol 12 no 11 pp 2193ndash2205 2013

[22] C Xu Z Li L Zhong H Zhang and G-M Muntean ldquoCMT-NC improving the concurrent multipath transfer performanceusing network coding in wireless networksrdquo IEEE Transactionson Vehicular Technology vol 65 no 3 pp 1735ndash1751 2016

[23] C Xu Z Li J Li H Zhang and G-M Muntean ldquoCross-layerfairness-driven concurrent multipath video delivery over het-erogeneous wireless networksrdquo IEEE Transactions on Circuits

and Systems for Video Technology vol 25 no 7 pp 1175ndash11892015

[24] F Bai N Sadagopan and A Helmy ldquoThe important frameworkfor analyzing the impact of mobility on performance of routingprotocols for Ad Hoc networksrdquo Ad Hoc Networks vol 1 no 4pp 383ndash403 2003

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



Internet

Media server

Video namePITNullCS

FIB Interface WiFi


FIB Interface WiFi


FIB Interface WiFi

Video namePITNullCS

FIB Interface WiFi

Video namePITNullCS

FIB Interface WiFi

Interest packet flow Datainterest packet flow discardedData packet flow

V1

V1

V1

V1

V1

Figure 1 Multimedia streaming services in mobile content-centric networks

to achieve nearby content fetching [11] which reduces delayof content lookup and delivery Figure 1 illustrates con-tent delivery process in mobile content-centric networks(MCCNs) Each mobile node in MCCNs maintains the datastructures content store (CS) pending interest table (PIT)and forwarding information base (FIB) [12] The mobilenodes send interest packets to their neighbor nodes in orderto obtain the desired video data If the neighbor nodeshave cached the requested video data in CS the data isdirectly returned to the request users to the request nodesOtherwise if the neighbor nodes do not cache the requestedvideo data locally the mobile nodes record the incominginterface of interest packets in PIT and broadcast to allneighbors if themobile nodes have recorded the informationof interest packets they discard the received interest packetsIf the neighbor nodes firstly receive the interest packets theyforward the interest packets to the next-hop nodes if themobile nodes receive the same interest packets they discardthe received packetsWhen the interest packets are forwardedto the content providers which cache the requested videodata the providers return the video data along the reversesearching path At themoment the intermediate nodes (relaynodes) in the data reverse path can cache the returned data inorder to have nearby access to the data in the future

The all-to-cache the traditional caching method inMCCNs consumes large number of storage resources ofnodes to support nearby content fetching in order to offloadtraffic in underlying network which leads to huge waste ofstorage resourcesThe low capacities of storage computationand energy of mobile devices difficultly support the massiveconsumption On the other hand because the nodes are the

content carriers the content movement in geographic areawith movement of carriers brings huge negative influencefor content caching effectiveness Moreover the traditionalbroadcasting-based content lookup approach inMCCNs alsowastes huge network bandwidth The network congestioncaused by large-scale request results in long startup delayIn order to address the problems that existed in traditionalmethods some unicast-based content delivery solutions inMCCNs make use of the collected and maintained contentinformation to achieve precise and economic content lookupHowever the nodes need tomaintain large number of contentcarriers in order to fast search content providers The largerthe scale of maintained information is the higher the prob-ability of fast and successful lookup is The maintenance oflarge-scale content information not only brings redundancylink between nodes but also leads to the overload of mobilenodes due to low capacities Otherwise the small-scalemaintenance reduces lookup hit rate which also results inlong startup delay In order to ensure high QoE of users andreducemaintenance cost of nodes a key issue is how to designthe economic and efficient content delivery solutions

In this paper we propose a novel contribution-awareneighbor-assist video delivery solution over mobile content-centric networks (CNVD) CNVD allows the mobile nodes tobuild and maintain neighbor relationship with other mobilenodes to share cached videos and support unicast-basedvideo lookup In order to achieve economic maintenancefor the video content in MCCNs to support efficient lookupand transmission of videos CNVD designs an estimationmethod of contribution level of neighbor nodes by inves-tigation for lookup success rate number of cached videos



2 Related Work









119862ℎ (VC119895) =

119908ℎ119903ℎ (1 minus 119889119903ℎ119889119906119894 ) 119889119903ℎ lt 119889119906119894

0 119889119903ℎ ge 119889119906119894(1)






119877119894ℎ (VC119895) = 119877119873119904119877119873119905 (2)



119909lowast (att) isin [0 1] (3)



GRC isin [0 1] (4)









119868119894ℎ = sum119904119888=119886 1003816100381610038161003816VL119888 minus UV1198881003816100381610038161003816

sum119904119888=119886 1003816100381610038161003816VL1198881003816100381610038161003816 119868119894ℎ isin [0 1] (6)




LP119886ℎ = sum119892119890=1 119862119890119892

119862119909 = sumV119887=1 119862119887V

(7)

























15

2

25

3

35

4

45

Look

up la

tenc

y (s

)


CNVDRUFS





005

115

225

335

445

Look

up la

tenc

y (s

)


CNVDRUFS




0

005

01

015

02

025

03

035

04

CHR

Simulation time (s)

CNVDRUFS

0 200 400 600 800 1000




CHR = 119867119873119877119873 CHR isin [0 1] (10)






0

005

01

015

02

025

03

CHR


CNVDRUFS

20 40 60 80 100 120 140 160 180 200







CNVDRUFS

005

115

225

335

445

5

Play

back

free

ze fr

eque

ncy


005

115

225

335

445

Mai

nten

ance

ove

rhea

d (k

bs)


CNVDRUFS








6 Conclusion


Notations












Acknowledgments


References


































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





2 Related Work









119862ℎ (VC119895) =

119908ℎ119903ℎ (1 minus 119889119903ℎ119889119906119894 ) 119889119903ℎ lt 119889119906119894

0 119889119903ℎ ge 119889119906119894(1)






119877119894ℎ (VC119895) = 119877119873119904119877119873119905 (2)



119909lowast (att) isin [0 1] (3)



GRC isin [0 1] (4)









119868119894ℎ = sum119904119888=119886 1003816100381610038161003816VL119888 minus UV1198881003816100381610038161003816

sum119904119888=119886 1003816100381610038161003816VL1198881003816100381610038161003816 119868119894ℎ isin [0 1] (6)




LP119886ℎ = sum119892119890=1 119862119890119892

119862119909 = sumV119887=1 119862119887V

(7)

























15

2

25

3

35

4

45

Look

up la

tenc

y (s

)


CNVDRUFS





005

115

225

335

445

Look

up la

tenc

y (s

)


CNVDRUFS




0

005

01

015

02

025

03

035

04

CHR

Simulation time (s)

CNVDRUFS

0 200 400 600 800 1000




CHR = 119867119873119877119873 CHR isin [0 1] (10)






0

005

01

015

02

025

03

CHR


CNVDRUFS

20 40 60 80 100 120 140 160 180 200







CNVDRUFS

005

115

225

335

445

5

Play

back

free

ze fr

eque

ncy


005

115

225

335

445

Mai

nten

ance

ove

rhea

d (k

bs)


CNVDRUFS








6 Conclusion


Notations












Acknowledgments


References


































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in






119862ℎ (VC119895) =

119908ℎ119903ℎ (1 minus 119889119903ℎ119889119906119894 ) 119889119903ℎ lt 119889119906119894

0 119889119903ℎ ge 119889119906119894(1)






119877119894ℎ (VC119895) = 119877119873119904119877119873119905 (2)



119909lowast (att) isin [0 1] (3)



GRC isin [0 1] (4)









119868119894ℎ = sum119904119888=119886 1003816100381610038161003816VL119888 minus UV1198881003816100381610038161003816

sum119904119888=119886 1003816100381610038161003816VL1198881003816100381610038161003816 119868119894ℎ isin [0 1] (6)




LP119886ℎ = sum119892119890=1 119862119890119892

119862119909 = sumV119887=1 119862119887V

(7)

























15

2

25

3

35

4

45

Look

up la

tenc

y (s

)


CNVDRUFS





005

115

225

335

445

Look

up la

tenc

y (s

)


CNVDRUFS




0

005

01

015

02

025

03

035

04

CHR

Simulation time (s)

CNVDRUFS

0 200 400 600 800 1000




CHR = 119867119873119877119873 CHR isin [0 1] (10)






0

005

01

015

02

025

03

CHR


CNVDRUFS

20 40 60 80 100 120 140 160 180 200







CNVDRUFS

005

115

225

335

445

5

Play

back

free

ze fr

eque

ncy


005

115

225

335

445

Mai

nten

ance

ove

rhea

d (k

bs)


CNVDRUFS








6 Conclusion


Notations












Acknowledgments


References


































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





119877119894ℎ (VC119895) = 119877119873119904119877119873119905 (2)



119909lowast (att) isin [0 1] (3)



GRC isin [0 1] (4)









119868119894ℎ = sum119904119888=119886 1003816100381610038161003816VL119888 minus UV1198881003816100381610038161003816

sum119904119888=119886 1003816100381610038161003816VL1198881003816100381610038161003816 119868119894ℎ isin [0 1] (6)




LP119886ℎ = sum119892119890=1 119862119890119892

119862119909 = sumV119887=1 119862119887V

(7)

























15

2

25

3

35

4

45

Look

up la

tenc

y (s

)


CNVDRUFS





005

115

225

335

445

Look

up la

tenc

y (s

)


CNVDRUFS




0

005

01

015

02

025

03

035

04

CHR

Simulation time (s)

CNVDRUFS

0 200 400 600 800 1000




CHR = 119867119873119877119873 CHR isin [0 1] (10)






0

005

01

015

02

025

03

CHR


CNVDRUFS

20 40 60 80 100 120 140 160 180 200







CNVDRUFS

005

115

225

335

445

5

Play

back

free

ze fr

eque

ncy


005

115

225

335

445

Mai

nten

ance

ove

rhea

d (k

bs)


CNVDRUFS








6 Conclusion


Notations












Acknowledgments


References


































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in







119868119894ℎ = sum119904119888=119886 1003816100381610038161003816VL119888 minus UV1198881003816100381610038161003816

sum119904119888=119886 1003816100381610038161003816VL1198881003816100381610038161003816 119868119894ℎ isin [0 1] (6)




LP119886ℎ = sum119892119890=1 119862119890119892

119862119909 = sumV119887=1 119862119887V

(7)

























15

2

25

3

35

4

45

Look

up la

tenc

y (s

)


CNVDRUFS





005

115

225

335

445

Look

up la

tenc

y (s

)


CNVDRUFS




0

005

01

015

02

025

03

035

04

CHR

Simulation time (s)

CNVDRUFS

0 200 400 600 800 1000




CHR = 119867119873119877119873 CHR isin [0 1] (10)






0

005

01

015

02

025

03

CHR


CNVDRUFS

20 40 60 80 100 120 140 160 180 200







CNVDRUFS

005

115

225

335

445

5

Play

back

free

ze fr

eque

ncy


005

115

225

335

445

Mai

nten

ance

ove

rhea

d (k

bs)


CNVDRUFS








6 Conclusion


Notations












Acknowledgments


References


































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in























15

2

25

3

35

4

45

Look

up la

tenc

y (s

)


CNVDRUFS





005

115

225

335

445

Look

up la

tenc

y (s

)


CNVDRUFS




0

005

01

015

02

025

03

035

04

CHR

Simulation time (s)

CNVDRUFS

0 200 400 600 800 1000




CHR = 119867119873119877119873 CHR isin [0 1] (10)






0

005

01

015

02

025

03

CHR


CNVDRUFS

20 40 60 80 100 120 140 160 180 200







CNVDRUFS

005

115

225

335

445

5

Play

back

free

ze fr

eque

ncy


005

115

225

335

445

Mai

nten

ance

ove

rhea

d (k

bs)


CNVDRUFS








6 Conclusion


Notations












Acknowledgments


References


































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




005

115

225

335

445

Look

up la

tenc

y (s

)


CNVDRUFS




0

005

01

015

02

025

03

035

04

CHR

Simulation time (s)

CNVDRUFS

0 200 400 600 800 1000




CHR = 119867119873119877119873 CHR isin [0 1] (10)






0

005

01

015

02

025

03

CHR


CNVDRUFS

20 40 60 80 100 120 140 160 180 200







CNVDRUFS

005

115

225

335

445

5

Play

back

free

ze fr

eque

ncy


005

115

225

335

445

Mai

nten

ance

ove

rhea

d (k

bs)


CNVDRUFS








6 Conclusion


Notations












Acknowledgments


References


































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




0

005

01

015

02

025

03

CHR


CNVDRUFS

20 40 60 80 100 120 140 160 180 200







CNVDRUFS

005

115

225

335

445

5

Play

back

free

ze fr

eque

ncy


005

115

225

335

445

Mai

nten

ance

ove

rhea

d (k

bs)


CNVDRUFS








6 Conclusion


Notations












Acknowledgments


References


































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in






6 Conclusion


Notations












Acknowledgments


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



research article a novel contribution-aware neighbor

Documents