social-aware cooperative video distribution via svc...

Research ArticleSocial-Aware Cooperative Video Distributionvia SVC Streaming Multicast

Lindong Zhao 12 Lei Wang 12 Xuguang Zhang 12 and Bin Kang3

1National Engineering Research Center for Communications and Network TechnologyNanjing University of Posts and Telecommunications China2National Mobile Communications Research Laboratory Southeast University China3School of Internet of Things Nanjing University of Posts and Telecommunications China

Correspondence should be addressed to Lei Wang wangleinjupteducn

Received 4 May 2018 Revised 4 October 2018 Accepted 10 October 2018 Published 25 October 2018

Academic Editor Jaime Lloret

Copyright copy 2018 Lindong Zhao 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

Scalable Video Coding (SVC) streaming multicast is considered as a promising solution to cope with video traffic overload andmulticast channel differences To solve the challenge of delivering high-definition SVC streaming over burst-loss prone channelswe propose a social-aware cooperative SVC streaming multicast schemeThe proposed scheme is the first attempt to enable D2Dcooperation for SVC streaming multicast to conquer the burst-loss and one salient feature of it is that it takes fully into account thehierarchical encoding structure of SVC in scheduling cooperation By using our scheme users form groups to share video packetsamong each other to restore incomplete enhancement layers Specifically a cooperative group formation method is designed tostimulate effective cooperation based on coalitional game theory and an optimal D2D links scheduling scheme is devised tomaximize the total decoded enhancement layers based on potential game theory Extensive simulations using real video tracescorroborate that the proposed scheme leads to a significant gain on the received video quality

1 Introduction

With the rapid development of mobile intelligent terminalswe have witnessed the explosive growth of usersrsquo demandsfor wireless live video services [1] Such growth in thebandwidth demanding will bring unprecedented challengesto current wireless network in the near future [2 3] Ascommunication technologies designed to cope with videotraffic overload video multicast can exploit the broadcastingnature of wireless communication to transmit video to mul-tiple users simultaneously [4] Due to its effectiveness videomulticast has been applied successfully in many scenarioseg breaking news video live sport videos mobile TVprograms and so on

In mobile video multicast how to solve varied fadingchannels from the base station (BS) to multicast users is akey problem Here we give an illustrative paragraph Thevideo transmission rate of a wireless link is determinedby the channel quality feedback For a specific link hightransmission rate with a bad channel quality will render

high bit error rate and degrade the goodput In a multicastscenario BS can only select one transmission rate while thechannel conditions of multicast users vary from each otherWhen BS multicast video to users according to the user withgood channel quality users with worse channel quality mayfail to play back the video smoothly Typically BS will choosetransmission rate according to the worst channel and thuseach multicast userrsquos received video quality is limited by theworst one

Since Scalable Video Coding streaming can be hierarchi-cally divided into one base layer and multiple enhancementlayers the SVC based video multicast can flexibly adjustthe quality of live video services according to channel feed-back [5] Thus SVC is a suitable video coding strategy toadapt to the diversity of multicast channels [6] One majorchallenge of employing SVC in video multicast is how tocope with the unavoidable packet loss over the burst-lossprone channels [7] which makes the traditional channelprotection method such as FEC insufficient [8] And due tothe stringent playback deadline the retransmission is also

HindawiWireless Communications and Mobile ComputingVolume 2018 Article ID 9315357 9 pageshttpsdoiorg10115520189315357

2 Wireless Communications and Mobile Computing

not preferred especially in multicast scenarios [9] Althoughsome related works have studied how to reduce the inter-ference and errors in SVC streaming [10ndash12] most of themneed heavy modification for video codec or introduce othernetwork components to the cellular system which is actuallyimpractical

Since the cooperative local recovery via D2D communi-cation can make up the packet loss without introducing anyredundancy or asking the base station to retransmit the lostpackets it is then a good option for SVC streaming multicastcountering the burst-loss There are some works that studiedhow to utilize D2D communication or other network inter-faces (like Wi-Fi) for video recovery [13] However previouscooperative recovery schemes did not take the hierarchicalencoding structure of SVC into account which may not leadto a significant gain to SVC-based video multicast In otherwords from the perspective of userrsquos Quality-of-Experience(QoE) [14] simply boosting the transmission rate does notnecessarily result in improvements in the user perception ofthe video quality [15]

SVC streaming has a hierarchical encoding structurewhich means that each enhancement layer can be decodedonly after the correct decoding of lower layers In suchcase either bit error or packet loss would lead to theloss of synchronization information and error propagationThe hierarchical encoding structure indicates that packetsbelonging to different enhancement layers should have differ-ent weight for improving video quality And the SVC basedvideo distribution scheme should also be structured anddealingwithmissing packets depends on the actual receptionExisting video distribution schemes often treat all themissingpackets equally hence those distribution schemes can notgive a good video quality for SVC streaming multicast In aword how to coordinate usersrsquo cooperation according to thehierarchical encoding structure of SVC is a critical issue forcooperative SVC streaming multicast

In this paper we propose a D2D-based cooperativeSVC streaming multicast scheme to improve the receivedvideo quality which is based on the consideration of thehierarchical encoding structure of SVC More specificallyusers in the same cooperative group share enhancementlayers usingD2Dcommunication under the scheduling of theBS To stimulate effective cooperation amongmulticast usersthe cooperative groups are constructed by using a coalitionalgame And to maximize the total decoded enhancementlayers an optimal D2D links scheduling scheme is devisedbased on potential game theory To the best of our knowledgethis paper is the first attempt to enable D2D cooperation forSVC streaming multicast to conquer the burst-lossThemaincontributions of this paper are summarized as follows

(i) We propose a social-aware cooperative SVC stream-ing multicast scheme to improve received video qual-ity by stimulating users to form cooperative groupsand share enhancement layers In particular thisscheme takes full account of the hierarchical encodingstructure of SVC to coordinate cooperation

(ii) To stimulate effective cooperation among users wepropose a cooperative group formationmethod based

on coalitional game theory in which the logicalrelationship among users is developed according tosocial attributes interest similarity and geographicallocation

(iii) To maximize the total received enhancement layersof the cooperative group we devise a D2D linksscheduling scheme for the base station The optimiza-tion problem for D2D links scheduling is cast as apotential game and solved by local search algorithmExtensive simulations with real video traces showthat the proposed scheme can achieve up to 659gain of video PSNR (Peak Signal-to-Noise Ratio)and valuable video frame ratio over the traditionalscheme

The rest of this paper is organized as follows Section 2gives a review of the related works in the field of wirelessvideo multicast and cooperative local recovery The systemmodel and workflow of the proposed scheme are discussed inSection 3 In Section 4 we study the problems of cooperativegroup formation and the optimal links scheduling in detailThe extensive numerical results are shown in Section 5 andconclusions are drawn in Section 6

2 Related Works

21 Wireless Video Multicast Wireless video multicast hasrecently become a hot research topic which is promising tohelp improve spectrum efficiency R Chandra et al proposeda multicast system DirCast based on Wi-Fi targeting tominimize the total transmission delay [16] In order toimprove the performance of video multicast a gateway wasintroduced between the base station and the cloud serversto determine the order of frames [17] and S Aditya et alintegrated a joint source-channel coding into the MPEG-4 codec [18] Deb et al proposed a greedy algorithm forresource allocation in SVC streaming multicast scenariovia WiMAX [11] S Jakubczak et al proposed a cross-layer design for multicasting scalable video with the goalof improving robustness to wireless interference and errors[12]

Some of the above works with nonscalable encodingsuffer from ldquothe limit of the worst onerdquo caused by multicastchannel differences as mentioned earlier To counter wirelessinterference and errors the others with scalable encodingintroduce heavy redundancy to video codec or add newnetwork interface to the system but present insufficient forburst-loss Different from these works this paper effectivelyconquers the burst-loss for SVC streaming multicast viaD2D cooperation and is compatible with existing cellularnetworks

22 Cooperative Local Recovery There have been someworkson the cooperative local recovery ie how to enable usersto share missing packets or act as relay nodes Reference[19] proposed a cooperative peer-to-peer recovery schemeoptimally scheduling the priority of packets Reference [9]recovered the video frames by constructing D2D assistedmultipath transmission Reference [20] realized peer-to-peer

Wireless Communications and Mobile Computing 3

Enhancement Layer Increase Quality

Enhancement Layer Increase Frame Rate

Enhancement Layer Increase Resolution

BaseLayer Basic InformationH

264

SVC

Stre

am

Figure 1 Scalable Video Coding structure

repair by combining batch processing and network codingReference [21] deployedD2D communication inside themul-ticast group and users who successfully decoded the messageretransmit it to those in need Reference [13] stimulatedusersrsquo cooperation for video multicast by utilizing two typesof social ties social reciprocity and social trust The abovevideo distribution schemes often treat all the missing packetsequally hence they cannot give a good video quality forSVC streamingmulticast Different from themwe coordinateusersrsquo cooperation based on the unique hierarchical encodingstructure of SVC

3 System Model

31 Preliminaries To facilitate our exposition it is necessaryto first introduce the scalable video encoding structure andlive video playback mechanism

The core idea of SVC is to divide the video streaminginto multiple layers according to different requirements Asillustrated in Figure 1 there are one base layer and multipleenhancement layers each of which is transmitted separatelyDevices can reconstruct the basic quality video image as longas it receives the base layer but the video reconstructed fromthe base layer is of poor quality The enhancement layerscontain details of the video streaming which means userscan get a better video quality by receiving more enhancementlayers within the range of the total video bit rate In additionan enhancement layer can be decoded after the correctdecoding of the base layer and lower enhancement layerswhile the base layer can be decoded separately

In general SVC streaming is composed of Group-of-Picture (GoP) sequences and a GoP consists of a base layerand one ormore enhancement layers with RLC encodingWedenote these layers by L = 1 2 sdot sdot sdot L each encoded layer119897 isin L is processed into 119875119897 packets and 119875119897 is also the decodingthreshold for layer 119897 The condition for successfully receivingthe layer 119897 can be expressed as

119877119897119894 ⩾ 119875119897 119897 isin L (1)

However receiving the full information of the layer 119897 doesnot mean that layer 119897 can be reconstructed the condition forsuccessfully decoding the layer 119897 is given by

1198771198971015840119894 ⩾ 1198751198971015840 forall1198971015840 ⩽ 119897 119897 isin L forall1198971015840 isin L (2)

where 1198771198971015840119894 denotes the packets which user 119894 received for layer1198971015840

Equation (2) means that to restore video quality to layer119897 not only should the received packets of layer 119897 be more thanits threshold but also the lower layers are decoded correctly

The live video playback mechanism is illustrated inFigure 2 Unlike data transmissions such as browsing a Webpage or downloading a file live video multicast is a delaysensitive application with real-time requirements Generallythe smooth video playback is tolerable with amaximum delayaccording to Persistence of Vision That is each GoP hasa fixed transmission deadline Once a packet in the GoPmisses this deadline it will be dropped For SVC streamingpackets on each layer are transmitted independently Usersreconstruct each video layer with the received packets andthe layer which does not receive enough packets before thedeadline cannot be reconstructed

32 System Model and Workflow From the previous dis-cussion we can see that affected by hierarchical encodingstructure of SVC and live video delay constraint users maynot be able to reconstruct a certain layer due to their failure toreceive the lower layer correctly Referring to SVC streamingonce a user receives enough packets for the correspondinglayer it can be reconstructed regardless of whether thepackets are coming from the base station or other users Dueto the diversity of multicast channels when some users losea certain layer other users may receive the layer successfullyFirst we propose to enable users to form ldquocooperative groupsrdquotaking into account social attributes preference similarityand geographical location according to the strategy proposedin Section 4Then we develop collaboration among users viaD2D communication inside each group and missing packetsare shared among users under the base stationrsquos scheduling torebuild lost layers

As shown in Figure 3 one base station and 119873 users arein a cell and partial users are divided into different multicastgroups receiving different videos via the cellular downlinkcalled multicast users Others who have no intention ofparticipating in multicast are referred to as ordinary usersSince the ordinary user is not covered by our discussion wemay assume that the 119873 users are all multicast users And allusers are equipped with both D2D and cellular interfacesso that they can receive packets by D2D communication ifnecessary

The workflow of our proposed intragroup recovery isshown in Figure 4 we divide one GoP delivery processinto two parts multicast round and cooperation round Atthe beginning of one GoP transmission each user receivespackets from the base station during multicast phase Inprocess of following cooperation round users would reportto the base station which layers are missing The D2D linkwill be established by the unified scheduling of the basestation to realize the complementation of the data receivedamong users If the time slots allocated to cooperationround are sufficient we can encapsulate the above reportingscheduling and sharing processes into a subround allowingusers to switch multiple times to obtain their missing layers


Application Layer GoP i-1 Playback GoP i Playback GoP i+1 Playback

Transport Layer GoP i Transmission GoP i+1 Transmission GoP i+2 Transmission

TimeDeadline

Figure 2 Live video transmission and playback

Group formation andScheduling

GeographicalLocation

Multicast

D2DMulticast users

Ordinary users

Figure 3 An illustration of cooperative SVC streaming multicast system

4 Cooperative Group Formulation andOptimal Links Scheduling

There are two technical challenges for our proposed scheme(1) how to effectively screen users into multicast group tostimulate cooperation and (2) how to schedule D2D links tomaximize the overall reconstructed SVC layers Now we willlook at these two subproblems

41 Cooperative Multicast Group Formation Now we for-mulate the cooperative multicast group formation for SVCstreaming recovery as a coalitional game

411 Utility Function When a user is ready to invite anotheruser to join the multicast group what are the main consider-ations

First consistency of user preferences within group isa prerequisite for multicast group formation in order toimprove the efficiency of spectrum resources in cellularsystem We use collaborative filtering method to obtain thesimilarity of usersrsquo interest Assume that the number ofmulticast video types available is 119883119870(119894) = 1198861198941 1198861198942 sdot sdot sdot 119886119894119883denoting user 119894rsquos interest set obtained from historical dataanalysis where 119886119894119909 is a binary variable which is equal to 1 ifuser 119894 is interested in video type 119909 otherwise 119886119894119909 = 0 Theinterest similarity between users 119894 and 119895 can be obtained bythe Jaccard coefficient [22] expressed by

119878 (119894 119895) =1003816100381610038161003816119870 (119894) cap 119870 (119895)10038161003816100381610038161003816100381610038161003816119870 (119894) cup 119870 (119895)1003816100381610038161003816 (3)

where 0 le 119878(119894 119895) le 1 the larger the value is the more likelythe users are to choose the same video content


Sub-round 1 Sub-round N

Multicastround

Cooperationround

GoP i-1 Transmission GoP i Transmission GoP i+1 Transmission

Time

Figure 4 Intragroup recovery workflow

Since social attributes represent long-term contactbetween users links maintained with them tend to be stable[23] With the help of social-related parameters to filterusers we can maximize the use of temporary links to achieveQoE promotion We use 119876(119894 119895) to denote the social affinityof users 119894 and 119895 which can be obtained by analyzing theirinteractions with social software Note that 0 le 119876(119894 119895) le 1the larger the value the closer the relationship between usersBesides since the next collaboration will take place with D2Dcommunication users have to consider location informationwhen inviting other members lest they become isolatednodes We characterize this property in terms of the ratioof distance between users to diameter of the cell expressedby 119866(119894 119895) = 119889119894119895119863 where 119863 denotes the diameter and 119889119894119895denotes the distance between users 119894 and 119895 0 lt 119866(119894 119895) le 1

After the above discussion we can see that there are threemain factors that affect the preference of user invitation andtheir relationship to utility functions can be expressed as

120597119880120597119878 ge 0120597119880120597119876 ge 0120597119880120597119866 le 0

(4)

And we will use the method of microeconomics to designutility function to quantify this preference In microeco-nomics utility is the quantitative expression of preferenceand utility is divided into ordinal utility and cardinality utility[24] Cardinal utility is measured by absolute value whileordinal utility is concerned with the relative relationship ofutility Here the users are concerned only with the relativerelationship of the utility brought about by inviting differentusers hence we use ordinal utility to quantify preferencesBecause there is no alternative or complementary relation-ship between interest similarity social attributes and geo-graphical location we use Cobb-Douglas preference modelto design our utility function expressed as

119880(119894 119895) = 119878 (119894 119895)120572 119876(119894 119895)120573119866(119894 119895)120574 (5)

where 120572 120573 and 120574 are the corresponding coefficients to adjustthe priorities and all positive Note that the value of ordinal

utility itself has no physical meaning we just use it to quantifypreference the forms of utility function are not uniqueBesides without considering the preference similarity theproposed scheme can also be used for users in the sameMBMS service area [25]

412 Coalitional Game Formulation Then we cast themulticast group formation as a coalitional game Ω =N ΦNW (≻119894)119894isinN as follows

(i) The set of players N is the set of multicast users(ii) The set of cooperation strategies is ΦN = (119891119894)119894isinN

119891119894 isin N119905 forall119894 isin N which denotes the set of currentinvestible users for all users

(iii) The characteristic function is119882(Δ) = (119891119894)119894isinN isin ΦN (119891119894)119894isinΔ = (119895)119895isinΔ 119886119899119889 119891119896 = 119896 119896 isin N Δ for eachmulticast group Δ isin N There are only two cases forthe players and the condition (119891119894)119894isinΔ = (119895)119895isinΔ denotesthat user 119894 chooses to invite user 119895 to join a multicastgroup Δ (a coalition) The condition 119891119896 = 119896 119896 isin N Δrepresents that user 119896 out of the coalition Δ is not inany group

(iv) The preference order (≻119894)119894isinN is defined as (119891119896)119896isinN≻119894(1198911015840119896)119896isinN if and only if 119880(119894 119891119894) ⩾ 119880(119894 1198911015840119894 ) In detailuser 119894 prefers the group formation strategy (119891119896)119896isinN to(1198911015840119896)119896isinN if and only if user 119894 invites assigned user 119891119894in strategy (119891119896)119896isinN to have more utility than invitingassigned user 1198911015840119894 in strategy (1198911015840119896)119896isinN according toabove-mentioned utility function

Definition 1 The core is the set of (119891lowast119894 )119894isinN isin 119882(N) for whicha coalition Δ does not exist and (119891119894)119894isinN isin 119882(Δ) such that(119891119894)119894isinN≻119894(119891lowast119894 )119894isinN

That is no set of participants will actively take a jointaction that is better for all of them to deviate from currentcoalition which is in the same spirit as Nash equilibriumin a noncooperative game By obtaining a core solution ofthe coalition game the formation of multicast groups iscompleted

413 Coalitional Game Core Solution Now we study theexistence of the core solution

Definition 2 A game = Z ΦZW (≻119894)119894isinZ is a subgameof the game Ω if and only if Z sube N 119886119899119889 Z = 0


Definition 3 A nonempty subset Δ isin Z is a top coali-tion for a subgame = Z ΦZW (≻119894)119894isinZ if and onlyif for strategy set (119891lowast119911 )119911isinZ isin 119882(Δ) and any strategy(119906119911)119911isinΔ isin 119882(C)CsubeZ we have 119880(119911 119891lowast119911 ) ⩾ 119880(119911 119906119911) for all119911 isin Z

In fact a subgame is a subset of the players of the originalgame Note that the strategy set to build a top coalition is thecore of the subgamemade up ofmembers in the top coalitionsince the group is collectively best for all its members

Definition 4 The game Ω = N ΦNW (≻119894)119894isinN owns onlyone core solution if and only if all its subgames have a topcoalition and any player in any subgame cannot belong tomore than one top coalition

That is the top coalitions set up by adopting the corestrategy are the multicast groups we need

Definition 5 For a coalitional subgame = Z ΦZW(≻119894)119894isinZ we define a user sequence (1198991 sdot sdot sdot 119899119862) as a multicastfriend cycle if and only if 119880(119911119888 119911119888+1) ⩾ 119880(119911119888 119911119896) forall119911119896 isin Z forall 119911 = 1 sdot sdot sdot 119862 minus 1 and 119880(119911119862 1199111) ⩾ 119880(119911119862 119911119896) forall119911119896 isin Z

Lemma 6 For any subgame = Z ΦZW (≻119894)119894isinZ we canobtain at least one multicast friends cycle Any multicast friendcycle is a top coalition of corresponding subgame Proof At the initial stage of any subgame we randomlychoose user 1199111 from the players subset ΦZ then we can findthe second user 1199112 who can maximize user 1199111rsquos utility Repeatthe steps above and we can find a multicast friend cycleUsing proofs by contradiction if such cycle does not exist theabove-mentioned user sequence will extend infinitely Sinceany two users are different and the total number of users inthe cell is limited this assumption is clearly not true Andaccording to Definition 5 each user takes the decisions thatare best for themselves no participant can take a joint actionthat is better for all of them to deviate from current coalitionAny multicast friend cycle generated through the above stepsis a top coalition

Theorem 7 For game Ω = N ΦNW (≻119894)119894isinN the strategyset for generating multicast friend groups is a core solution andthe set of coalitions resulting from a decision initiated by anyone of themulticast users is unique if no strategies for one playercan bring about the same utility

Proof According to Lemma 6 at the beginning of game Ωwe randomly choose a user to invite other users and obtain atop coalition 1198621 By Definition 3 players in this top coalitionhave achieved the maximum utility currently available theseplayers will not continue to participate in current game Afterremoving the players mentioned above the original gamebecomes its corresponding subgame = Z ΦZW (≻119894)119894isinZwhereZ = N1198621With the generation of onemulticast friendcycle after another using Lemma 6 we have Z = N 119862119905for corresponding subgame to each iteration 119905 = 1 sdot sdot sdot 119879According to Definition 4 this completes the proof For thesecond part of the theorem using proofs by contradictionif one player can join two different top coalitions it must

have more than one optimal strategy which contradictsDefinition 5

Based on Lemma 6 and Theorem 7 we can find the coresolution for cooperative group formation

42 User Cooperation Scheduling As we mentioned earlierwith a view to the hierarchical encoding structure of SVCstreaming the same layer of a GoP has different valuesaccording to the actual reception For example layer 4 enablesbetter video quality for user who owns layers 1 2 3 butit does not work for user who only has layers 1 2 Theproposed scheme targets to improve video quality of themulticast group as much as possible in a subphase which isdirectly related to layers that can be reconstructed Consid-ering the upper limit to the perceived video quality of usersand marginal utility on the improvement of QoE [26 27] theoptimization problem can be formulated as

max119873

sum119894=1

11 + exp (minus120596119897119894) (6)

where120596 is the marginal utility coefficient and 119897119894 is the numberof layers that user 119894 has successfully reconstructed from thiscollaboration Now we analyze this problem with an exactpotential game

Definition 8 A game Γ = C (119884119894)119894isinC (119880119894)119894isinC is an exactpotential game if there exists a function H 119884 997888rarr 119877 suchthat forall119894 isin C forall119910minus119894 isin 119884119894 forall119909 119911 isin 119884119894 we have 119880119894(119909 119910minus119894) minus119880119894(119911 119910minus119894) = H(119909 119910minus119894) minus 119867(119911 119910minus119894) C119867 119884119894 119880119894 are respec-tively player set potential function player 119894rsquos strategy set andutility function

A potential game must have equilibrium with purestrategy and a discrete potential game has finite improvementproperties [28] If the objective function is a potentialfunction of a game with discrete strategy space it must havean optimal solution and can be solved by local search Nowwe formulate a potential game Γ as follows the set of playersC is the set of users in a multicast group player 119894rsquos availablestrategy 119884119894 is to choose which player in the group to requestvideo layer utility function is 119865119894 = 1(1 + exp(minus120596119897119894)) andpotential function is H = sum119873119894=1 1(1 + exp(minus120596119897119894)) = sum119873119894=1 119865119894

According to [28] we can prove the proposed objectivefunction is a potential function of game Γ Since the numberof members in multicast group is finite game Γrsquos strategyspace is discrete The problem (6) can be solved by localsearch algorithm such as Tabu Search and Simulated Anneal-ing

5 Performance Analysis

In this section we evaluate the performance of the proposedcooperative SVC streaming multicast scheme through simu-lations by using MATLAB

Without loss of generality we suppose that the wirelessmulticast channel conditions are random but quasi-static andremain unchanged from one GoP that is sent out until thedeadline to be dropped Since channel capacity described by


the Shannon equation does not reflect the wireless channelconditions preciously where the wireless channels are burst-loss prone at upper layer we use the Gilbert-Elliot channelmodel to capture burst-loss features in the proposed systemwithout handling the details of the datalink layer and physicallayer Gilbert-Elliot model can be denoted by a two-statestationary continuous time Markov chain there are only twocases per link Good or Bad If the state is Bad one frame willbe lost otherwise it can be successfully decoded

For video streaming we considerH264SVCvideo codecwith YUV420 ie there are 8 layers per GoP and eachlayer consists of 16 encoded video frames We implement theexperiments with the real video traces Foreman where theaverage PSNR (Peak Signal-to-Noise) of the Foreman is 4332dB the resolution of the Foreman is 352times288 and defaultvideo encoding bitrate is 450Kbps

Besides in order to measure the performance of the pro-posed scheme we consider the following three measurementmetrics

Peak Signal-to-Noise PSNR (Peak Signal-to-Noise) is anobjective evaluation of video quality and is a function of themean square between the original and the received videosequences which can be formulated as

119875119878119873119877 = 20 sdot log10 (119872119860119883119868radic119872119878119864) (7)

where119872119860119883119868 is the maximum value of frame pixel and119872119878119864is themean square error between the original and the receivedvideo sequences

Valuable Video Frame Ratio the valuable video frameratio is defined as the ratio between the number of valuablevideo frames received by a user and the number of alltransmitted frames before playback

Mean Opinion Score MOS (Mean Opinion Score) is asubjective measurement of userrsquos view on the video qualityWhen the PSNR range is about gt37 31sim37 25sim31 20sim25lt20userrsquos subjective experience quality is ldquoExcellentrdquo ldquoGoodrdquoldquoFairrdquo ldquoPoorrdquo and ldquoBadrdquo respectively [13]

The performance of the proposed scheme is comparedwith direct multicast in our simulations According to mul-ticast channel conditions users with direct SVC multicastdiscard the enhancement layers that could not be decodeddue to burst-loss119873 users are scattered into a round cell andthe interest similarity and social affinity between each otherare also randomly obtained and coefficients 120572 120573 120574 120596 areequal to 1 The base station determines the average bit ratethat can transmit all layers of oneGoP inmulticast round andD2D cooperation round is set to transmit an enhancementlayer

First we evaluate the performance of our proposedcooperative group formation scheme the experiences areimplemented with different number of multicast users in thecell119873 = 100 200 300 400 500 600 700 800 To ensure thegenerality of the result for each given119873 we average over 1000runs Figure 5 shows the average size of cooperative multicastgroups We can see the size of groups has not skyrocketedwith the number of users but slows down the growth rategradually This is because users select clusters based on the

5

6

7

8

9

10

11

12

13

14

Aver

age S

ize o

f Mul

ticas

t Gro

ups

200 300 400 500 600 700 800100Number of Users

Figure 5 Average size of cooperative groups under various numberof users

coalitional game A player can only select one player as theoptimal strategy and each player can only be selected oncethus limiting the size of cooperative groups

Next we employ the above evaluation metrics to comparethe received video quality of the proposed scheme andtraditional direct multicast The default multicast group sizeis set to 10 Figure 6 shows the relationship between thereceived video quality in terms of PSNR and the probabilityof burst-loss during one GoP transmission Obviously thegrowing probability of burst-loss signifies worse channelqualities With the deterioration of channel quality thegain of the proposed scheme over direct multicast shows agradual increase and then remains stable This is because therestoration of video quality in the proposed scheme is basedon the complementarity among users With the degradationof channel conditions the probability of complementaritybetween users increases gradually But when the channelquality deteriorates to a certain extent most users have onlyfew of enhancement layers and the possibility of cooperationis reduced

In Figure 7 curves of valuable video frame ratios withthe probability of burst-loss during one GoP transmissionare depicted We can observe that the proposed scheme faroutperforms direct multicast especially when the channelquality is quite poor The valuable video frame ratio of theproposed scheme is 659 higher than direct multicast Thisis because users in the direct SVC streaming multicast systemhave to discard the corresponding enhancement layers andthe system resources are severely wasted

To better understand the performance of the proposedscheme Figure 8 shows the experience qualities in termsof MOS under various burst-loss probability and the SVCstreaming recovery with the real video traces Foreman isshown in Figure 9 As we can see the distortion of the videosequences with the proposed scheme is slight According toFigures 6 and 9 the average PSNR of the proposed schemeis higher than 35dB and usersrsquo MOS is ldquoExcellentrdquo or ldquoGoodrdquoBut the PSNR of direct multicast is between 24dB and 34dBwhich signifies the MOS is ldquoFairrdquo So the scheme proposed inthis paper greatly improves the userrsquos experience


Direct MulticastThe Proposed Scheme

20

25

30

35

40PS

NR(

dB)

015 02 025 03 035 04 045 0501Burst-loss probability during a GoP transmission

Figure 6 Relationship between the video quality and burst-lossprobability


20

30

40

50

60

70

80

90

100

Valu

able

Vid

eo F

ram

e Rat

io(

)


Figure 7 Valuable video frame ratio under various burst-lossprobability

6 Conclusions

This paper studied a cooperative SVC streaming multi-cast scheme based on D2D communication to improvethe received video quality over burst-loss prone channelsThe proposed scheme motivates multicast users to formcooperative groups based on coalitional game theory toshare enhancement layers The proposed optimal D2D linksscheduling based on potential game helps to improve thereceived video quality of multicast users while taking intoaccount the hierarchical encoding structure of SVC More-over the proposed scheme can make up the burst-losswithout introducing heavy redundancy to video codec or addnew network interface Extensive numerical analysis studieswith real video traces have corroborated the significant gainusing the proposed scheme


1

2

3

4

5

MO

S


Figure 8 Mean opinion score under various burst-loss probability

BurstndashLoss PR = 015



(a) (b) (c)

Figure 9 Video recovery performance under various burst-lossprobability (a) Foreman frame (b) direct multicast and (c) theproposed scheme

In future it is of great interest to design a distributedoptimal link scheduling algorithm to reduce the burden ofcentralized scheduling on the BS

Data Availability

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

Conflicts of Interest

The authors declare that they have no conflicts of interest

Acknowledgments

This work was partly supported by the National Natu-ral Science Foundation of China (61571240 61671253 and61801242) the Priority Academic Program Development ofJiangsu Higher Education Institutions the Natural ScienceFoundation of Jiangsu Province (BK20170915) the Qing


Lan Project The Major Projects of the Natural ScienceFoundation of the Jiangsu Higher Education Institutions(16KJA510004) National Science Foundation of Jiang SuHigher Education Institutions (17KJD510005) The OpenResearch Fund of National Engineering Research Center forCommunication and Network Technology Nanjing Univer-sity of Posts and Telecommunications (TXKY17005) TheOpen Research Fund of National Mobile CommunicationsResearch Laboratory Southeast University (2016D01) TheResearch Fund of Nanjing University of Posts and Telecom-munications (NY218012)

References

[1] L Rodrigo Munoz D Mora and R Barriga ldquoTeachers Ciscocertification and their impact in Cisco Networking Academyprogramrdquo Revista Cientıfica de la UCSA vol 4 no 1 pp 50ndash56 2017

[2] L Zhou D Wu J Chen and Z Dong ldquoGreening the SmartCities Energy-Efficient Massive Content Delivery via D2DCommunicationsrdquo IEEE Transactions on Industrial Informaticsvol 14 no 4 pp 1626ndash1634 2017

[3] L Zhou DWu Z Dong and X Li ldquoWhenCollaboration HugsIntelligence Content Delivery over Ultra-Dense NetworksrdquoIEEE Communications Magazine vol 55 no 12 pp 91ndash95 2017

[4] L Zhou D Wu J Chen and Z Dong ldquoWhen ComputationHugs Intelligence Content-Aware Data Processing for Indus-trial IoTrdquo IEEE Internet of Things Journal 2017

[5] T Schierl T Stockhammer and T Wiegand ldquoMobile videotransmission using scalable video codingrdquo IEEE Transactionson Circuits and Systems for Video Technology vol 17 no 9 pp1204ndash1217 2007

[6] G Feng Y Zhang J Lin H Wang and L Cai ldquoJoint optimiza-tion of downlink and D2D transmissions for SVC streamingin cooperative cellular networksrdquoNeurocomputing vol 270 pp178ndash187 2017

[7] S Tang and P R Alface ldquoImpact of random and burst packetlosses on H264 scalable video codingrdquo IEEE Transactions onMultimedia vol 16 no 8 pp 2256ndash2269 2014

[8] I-H Hou and P R Kumar ldquoScheduling heterogeneous real-time traffic over fading wireless channelsrdquo in Proceedings of theIEEE INFOCOM pp 1ndash9 San Diego CA USA 2010

[9] Z Liu M Dong H Zhou X Wang Y Ji and Y TanakaldquoDevice-to-device assisted video frame recovery for picocelledge users in heterogeneous networksrdquo in Proceedings of the ICC2016 - 2016 IEEE International Conference on Communicationspp 1ndash6 Kuala Lumpur Malaysia May 2016

[10] Y-J Yu P-C Hsiu and A-C Pang ldquoEnergy-efficient videomulticast in 4G wireless systemsrdquo IEEE Transactions on MobileComputing vol 11 no 10 pp 1508ndash1522 2012

[11] S Deb S Jaiswal and K Nagaraj ldquoReal-time video multicastin WiMAX networksrdquo in Proceedings of the 27th IEEE Com-munications Society Conference on Computer Communications(INFOCOM rsquo08) pp 2252ndash2260 April 2008

[12] S Jakubczak and D Katabi ldquoA cross-layer design for scalablemobile videordquo in Proceedings of the ACM 17th Annual Inter-national Conference pp 289ndash300 Las Vegas Nevada USASeptember 2011

[13] Y Cao T Jiang X Chen and J Zhang ldquoSocial-aware videomulticast based on device-to-device communicationsrdquo IEEE

Transactions on Mobile Computing vol 15 no 6 pp 1528ndash15392016

[14] A Canovas M Taha J Lloret and J Tomas ldquoSmart resourceallocation for improving QoE in IP Multimedia SubsystemsrdquoJournal of Network and Computer Applications vol 104 pp 107ndash116 2018

[15] M Taha J Lloret A Canovas and L Garcia ldquoSurvey ofTransportation of Adaptive Multimedia Streaming service inInternetrdquo Network Protocols and Algorithms vol 9 no 1-2 p85 2018

[16] R Chandra S Karanth T Moscibroda et al ldquoDirCast Apractical and efficient Wi-Fi multicast systemrdquo in Proceedingsof the 2009 17th IEEE International Conference on NetworkProtocols (ICNP) pp 161ndash170 Plainsboro NJ USA October2009

[17] J Yoon H Zhang S Banerjee and S Rangarajan ldquoVideomulticast with joint resource allocation and adaptive modula-tion and coding in 4G networksrdquo IEEEACM Transactions onNetworking vol 22 no 5 pp 1531ndash1544 2014

[18] S Aditya and S Katti ldquoFlexCast Graceful wireless videostreamingrdquo in Proceedings of the ACM 17th Annual InternationalConference pp 277ndash288 Las Vegas Nevada USA September2011

[19] S Raza D Li C-N Chuah andG Cheung ldquoCooperative peer-to-peer repair for wireless multimedia broadcastrdquo in Proceed-ings of the IEEE International Conference onMultimedia andExpo ICME 2007 pp 1075ndash1078 China July 2007

[20] X Liu S Raza C N Chuah and G Cheung ldquoNetworkCoding Based Cooperative Peer-to-Peer Repair inWireless Ad-Hoc Networksrdquo in Proceedings of the 2008 IEEE InternationalConference on Communications pp 2153ndash2158 Beijing 2008

[21] B Zhou H Hu S-Q Huang and H-H Chen ldquoIntraclusterdevice-to-device relay algorithm with optimal resource utiliza-tionrdquo IEEE Transactions on Vehicular Technology vol 62 no 5pp 2315ndash2326 2013

[22] X-F Zhang and D-Q Dai ldquoA framework for incorporatingfunctional interrelationships into protein function predictionalgorithmsrdquo IEEE Transactions on Computational Biology andBioinformatics vol 9 no 3 pp 740ndash753 2012

[23] Y Li T Wu P Hui D Jin and S Chen ldquoSocial-aware D2Dcommunications qualitative insights and quantitative analysisrdquoIEEE Communications Magazine vol 52 no 6 pp 150ndash1582014

[24] Z Weiqing and L Zhenyu ldquoInformation Measurement ModelBasedonOrdinalUtilityrdquo inProceedings of the 2011 InternationalConference on Future Computer Sciences and Application pp154ndash157 Hong Kong China 2011

[25] M Taha J M Jimenez A Canovas and J Lloret ldquoIntelligentAlgorithm for Enhancing MPEG-DASH QoE in eMBMSrdquoNetwork Protocols and Algorithms vol 9 no 3-4 p 94 2018

[26] J P Hansen S Hissam and L Wrage ldquoQoS optimization inad hoc wireless networks through adaptive control of marginalutilityrdquo inProceedings of the 2013 IEEEWireless Communicationsand Networking Conference WCNC 2013 pp 1192ndash1197 Shang-hai China 2013

[27] L Badia M Lindstrom J Zander and M Zorzi ldquoDemand andPricing Effects on the Radio Resource Allocation ofMultimediaCommunication Systemsrdquo in Proceedings of the Global Telecom-munications Conference 2003 GLOBECOM rsquo03 vol 7 pp 4116ndash4121 IEEE 2003

[28] D Monderer and L S Shapley ldquoPotential gamesrdquo Games andEconomic Behavior vol 14 no 1 pp 124ndash143 1996

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018


Active and Passive Electronic Components

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of



Journal ofEngineeringVolume 2018

SensorsJournal of



RotatingMachinery


Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation


Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom


not preferred especially in multicast scenarios [9] Althoughsome related works have studied how to reduce the inter-ference and errors in SVC streaming [10ndash12] most of themneed heavy modification for video codec or introduce othernetwork components to the cellular system which is actuallyimpractical

Since the cooperative local recovery via D2D communi-cation can make up the packet loss without introducing anyredundancy or asking the base station to retransmit the lostpackets it is then a good option for SVC streaming multicastcountering the burst-loss There are some works that studiedhow to utilize D2D communication or other network inter-faces (like Wi-Fi) for video recovery [13] However previouscooperative recovery schemes did not take the hierarchicalencoding structure of SVC into account which may not leadto a significant gain to SVC-based video multicast In otherwords from the perspective of userrsquos Quality-of-Experience(QoE) [14] simply boosting the transmission rate does notnecessarily result in improvements in the user perception ofthe video quality [15]

SVC streaming has a hierarchical encoding structurewhich means that each enhancement layer can be decodedonly after the correct decoding of lower layers In suchcase either bit error or packet loss would lead to theloss of synchronization information and error propagationThe hierarchical encoding structure indicates that packetsbelonging to different enhancement layers should have differ-ent weight for improving video quality And the SVC basedvideo distribution scheme should also be structured anddealingwithmissing packets depends on the actual receptionExisting video distribution schemes often treat all themissingpackets equally hence those distribution schemes can notgive a good video quality for SVC streaming multicast In aword how to coordinate usersrsquo cooperation according to thehierarchical encoding structure of SVC is a critical issue forcooperative SVC streaming multicast

In this paper we propose a D2D-based cooperativeSVC streaming multicast scheme to improve the receivedvideo quality which is based on the consideration of thehierarchical encoding structure of SVC More specificallyusers in the same cooperative group share enhancementlayers usingD2Dcommunication under the scheduling of theBS To stimulate effective cooperation amongmulticast usersthe cooperative groups are constructed by using a coalitionalgame And to maximize the total decoded enhancementlayers an optimal D2D links scheduling scheme is devisedbased on potential game theory To the best of our knowledgethis paper is the first attempt to enable D2D cooperation forSVC streaming multicast to conquer the burst-lossThemaincontributions of this paper are summarized as follows

(i) We propose a social-aware cooperative SVC stream-ing multicast scheme to improve received video qual-ity by stimulating users to form cooperative groupsand share enhancement layers In particular thisscheme takes full account of the hierarchical encodingstructure of SVC to coordinate cooperation

(ii) To stimulate effective cooperation among users wepropose a cooperative group formationmethod based

on coalitional game theory in which the logicalrelationship among users is developed according tosocial attributes interest similarity and geographicallocation

(iii) To maximize the total received enhancement layersof the cooperative group we devise a D2D linksscheduling scheme for the base station The optimiza-tion problem for D2D links scheduling is cast as apotential game and solved by local search algorithmExtensive simulations with real video traces showthat the proposed scheme can achieve up to 659gain of video PSNR (Peak Signal-to-Noise Ratio)and valuable video frame ratio over the traditionalscheme

The rest of this paper is organized as follows Section 2gives a review of the related works in the field of wirelessvideo multicast and cooperative local recovery The systemmodel and workflow of the proposed scheme are discussed inSection 3 In Section 4 we study the problems of cooperativegroup formation and the optimal links scheduling in detailThe extensive numerical results are shown in Section 5 andconclusions are drawn in Section 6

2 Related Works

21 Wireless Video Multicast Wireless video multicast hasrecently become a hot research topic which is promising tohelp improve spectrum efficiency R Chandra et al proposeda multicast system DirCast based on Wi-Fi targeting tominimize the total transmission delay [16] In order toimprove the performance of video multicast a gateway wasintroduced between the base station and the cloud serversto determine the order of frames [17] and S Aditya et alintegrated a joint source-channel coding into the MPEG-4 codec [18] Deb et al proposed a greedy algorithm forresource allocation in SVC streaming multicast scenariovia WiMAX [11] S Jakubczak et al proposed a cross-layer design for multicasting scalable video with the goalof improving robustness to wireless interference and errors[12]

Some of the above works with nonscalable encodingsuffer from ldquothe limit of the worst onerdquo caused by multicastchannel differences as mentioned earlier To counter wirelessinterference and errors the others with scalable encodingintroduce heavy redundancy to video codec or add newnetwork interface to the system but present insufficient forburst-loss Different from these works this paper effectivelyconquers the burst-loss for SVC streaming multicast viaD2D cooperation and is compatible with existing cellularnetworks

22 Cooperative Local Recovery There have been someworkson the cooperative local recovery ie how to enable usersto share missing packets or act as relay nodes Reference[19] proposed a cooperative peer-to-peer recovery schemeoptimally scheduling the priority of packets Reference [9]recovered the video frames by constructing D2D assistedmultipath transmission Reference [20] realized peer-to-peer






264

SVC

Stre

am



3 System Model




119877119897119894 ⩾ 119875119897 119897 isin L (1)












TimeDeadline




Multicast

D2DMulticast users

Ordinary users







119878 (119894 119895) =1003816100381610038161003816119870 (119894) cap 119870 (119895)10038161003816100381610038161003816100381610038161003816119870 (119894) cup 119870 (119895)1003816100381610038161003816 (3)




Multicastround

Cooperationround


Time




120597119880120597119878 ge 0120597119880120597119876 ge 0120597119880120597119866 le 0

(4)


119880(119894 119895) = 119878 (119894 119895)120572 119876(119894 119895)120573119866(119894 119895)120574 (5)
























max119873

sum119894=1

11 + exp (minus120596119897119894) (6)













119875119878119873119877 = 20 sdot log10 (119872119860119883119868radic119872119878119864) (7)






5

6

7

8

9

10

11

12

13

14

Aver

age S

ize o

f Mul

ticas

t Gro

ups

200 300 400 500 600 700 800100Number of Users








20

25

30

35

40PS

NR(

dB)




20

30

40

50

60

70

80

90

100

Valu

able

Vid

eo F

ram

e Rat

io(

)



6 Conclusions



1

2

3

4

5

MO

S






(a) (b) (c)



Data Availability




Acknowledgments




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia







264

SVC

Stre

am



3 System Model




119877119897119894 ⩾ 119875119897 119897 isin L (1)












TimeDeadline




Multicast

D2DMulticast users

Ordinary users







119878 (119894 119895) =1003816100381610038161003816119870 (119894) cap 119870 (119895)10038161003816100381610038161003816100381610038161003816119870 (119894) cup 119870 (119895)1003816100381610038161003816 (3)




Multicastround

Cooperationround


Time




120597119880120597119878 ge 0120597119880120597119876 ge 0120597119880120597119866 le 0

(4)


119880(119894 119895) = 119878 (119894 119895)120572 119876(119894 119895)120573119866(119894 119895)120574 (5)
























max119873

sum119894=1

11 + exp (minus120596119897119894) (6)













119875119878119873119877 = 20 sdot log10 (119872119860119883119868radic119872119878119864) (7)






5

6

7

8

9

10

11

12

13

14

Aver

age S

ize o

f Mul

ticas

t Gro

ups

200 300 400 500 600 700 800100Number of Users








20

25

30

35

40PS

NR(

dB)




20

30

40

50

60

70

80

90

100

Valu

able

Vid

eo F

ram

e Rat

io(

)



6 Conclusions



1

2

3

4

5

MO

S






(a) (b) (c)



Data Availability




Acknowledgments




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia





TimeDeadline




Multicast

D2DMulticast users

Ordinary users







119878 (119894 119895) =1003816100381610038161003816119870 (119894) cap 119870 (119895)10038161003816100381610038161003816100381610038161003816119870 (119894) cup 119870 (119895)1003816100381610038161003816 (3)




Multicastround

Cooperationround


Time




120597119880120597119878 ge 0120597119880120597119876 ge 0120597119880120597119866 le 0

(4)


119880(119894 119895) = 119878 (119894 119895)120572 119876(119894 119895)120573119866(119894 119895)120574 (5)
























max119873

sum119894=1

11 + exp (minus120596119897119894) (6)













119875119878119873119877 = 20 sdot log10 (119872119860119883119868radic119872119878119864) (7)






5

6

7

8

9

10

11

12

13

14

Aver

age S

ize o

f Mul

ticas

t Gro

ups

200 300 400 500 600 700 800100Number of Users








20

25

30

35

40PS

NR(

dB)




20

30

40

50

60

70

80

90

100

Valu

able

Vid

eo F

ram

e Rat

io(

)



6 Conclusions



1

2

3

4

5

MO

S






(a) (b) (c)



Data Availability




Acknowledgments




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia




Multicastround

Cooperationround


Time




120597119880120597119878 ge 0120597119880120597119876 ge 0120597119880120597119866 le 0

(4)


119880(119894 119895) = 119878 (119894 119895)120572 119876(119894 119895)120573119866(119894 119895)120574 (5)
























max119873

sum119894=1

11 + exp (minus120596119897119894) (6)













119875119878119873119877 = 20 sdot log10 (119872119860119883119868radic119872119878119864) (7)






5

6

7

8

9

10

11

12

13

14

Aver

age S

ize o

f Mul

ticas

t Gro

ups

200 300 400 500 600 700 800100Number of Users








20

25

30

35

40PS

NR(

dB)




20

30

40

50

60

70

80

90

100

Valu

able

Vid

eo F

ram

e Rat

io(

)



6 Conclusions



1

2

3

4

5

MO

S






(a) (b) (c)



Data Availability




Acknowledgments




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia














max119873

sum119894=1

11 + exp (minus120596119897119894) (6)













119875119878119873119877 = 20 sdot log10 (119872119860119883119868radic119872119878119864) (7)






5

6

7

8

9

10

11

12

13

14

Aver

age S

ize o

f Mul

ticas

t Gro

ups

200 300 400 500 600 700 800100Number of Users








20

25

30

35

40PS

NR(

dB)




20

30

40

50

60

70

80

90

100

Valu

able

Vid

eo F

ram

e Rat

io(

)



6 Conclusions



1

2

3

4

5

MO

S






(a) (b) (c)



Data Availability




Acknowledgments




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia







119875119878119873119877 = 20 sdot log10 (119872119860119883119868radic119872119878119864) (7)






5

6

7

8

9

10

11

12

13

14

Aver

age S

ize o

f Mul

ticas

t Gro

ups

200 300 400 500 600 700 800100Number of Users








20

25

30

35

40PS

NR(

dB)




20

30

40

50

60

70

80

90

100

Valu

able

Vid

eo F

ram

e Rat

io(

)



6 Conclusions



1

2

3

4

5

MO

S






(a) (b) (c)



Data Availability




Acknowledgments




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia




20

25

30

35

40PS

NR(

dB)




20

30

40

50

60

70

80

90

100

Valu

able

Vid

eo F

ram

e Rat

io(

)



6 Conclusions



1

2

3

4

5

MO

S






(a) (b) (c)



Data Availability




Acknowledgments




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia




RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia


social-aware cooperative video distribution via svc...

Documents