research article adaptive streaming of scalable videos...

Research ArticleAdaptive Streaming of Scalable Videos over P2PTV

Youssef Lahbabi1 El Hassan Ibn Elhaj2 and Ahmed Hammouch3

1ENSET Mohammed V University 10100 Rabat Morocco2National Institute of Posts and Telecommunications (INPT) 10112 Rabat Morocco3ENSET Mohammed V University LRGE 10100 Rabat Morocco

Correspondence should be addressed to Youssef Lahbabi lahbabi youssefyahoofr

Received 7 April 2015 Revised 29 June 2015 Accepted 22 July 2015

Academic Editor Antonio Liotta

Copyright copy 2015 Youssef Lahbabi et alThis is an open access article distributed under theCreative CommonsAttribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

In this paper we propose a new Scalable Video Coding (SVC) quality-adaptive peer-to-peer television (P2PTV) system executedat the peers and at the network The quality adaptation mechanisms are developed as follows on one hand the Layer LevelInitialization (LLI) is used for adapting the video qualitywith the static resources at the peers in order to avoid long startup timesOnthe other hand the Layer Level Adjustment (LLA) is invoked periodically to adjust the SVC layer to the fluctuation of the networkconditions with the aim of predicting the possible stalls before their occurrence Our results demonstrate that our mechanismsallow quickly adapting the video quality to various system changes while providing best Quality of Experience (QoE) that matchescurrent resources of the peer devices and instantaneous throughput available at the network state

1 Introduction

Video streaming has become the most traffic intensive appli-cation in the world of multimedia According to a studyconducted by Cisco [1] the sum of all forms of video (TVvideo on demand [VoD] Internet and P2P) will be in therange of 80 to 90 percent of global consumer traffic by 2018In recent years Peer-to-peer systems have been successfullygainingmuch attention both in academic communities and inresearch industries P2P has played an important role in widebroadcasting growth ofmultimedia applications and providesseveral favorable characteristics such as self-organizationand configuration high scalability and robustness

Although P2P streaming takes advantage of the P2Parchitecture to lessen server load it gives reliable servicesand it still faces several challenges However current P2Pstreaming systems still suffer from a major limitation suchthese systems try to provide the video quality to all userswithout taking into account the varying bandwidth andthe heterogeneity of devices where each end terminal hasdifferent requirement for the video In this regard newmedia compression standard such as SVC (Scalable VideoCoding) has revolutionized the digital media industry byproviding high quality and scalable delivery of video contentto heterogeneous end-user devices and network conditions

End users have been enjoying benefits of peer-to-peertelevision (P2PTV) systems for over ten years P2PTV is acategory of P2P software applications specially designed toredistribute video streaming media based on a P2P networkCompared to P2P software designed for file sharing whichhas few concerns of timeliness property P2PTV is born forreal-time contents Typically the broadcasted video streamsare channels from all over the world but may also comefrom other sources for example recorded video files Thedraw to these applications is significant because they have thepotential tomake anyTVchannel globally available Based onthe type of distribution topology graph P2PTV systems canbe categorized as BitTorrent-like mesh-based [2] and Tree-based A BitTorrent-like P2PTV system can be viewed asa real-time version of BitTorrent [3] Similar to BitTorrentrent file sharing systems mechanisms that encourage fairnessare also implementable for BitTorrent-like P2PTV systemsAs the most widely used BitTorrent-like P2PTV systemsSopCast [4] PPLive [5] PPStream [6] and UUsee [7] areconsidered to be the most typical examples

Many researchers are currently investigating how to im-plement IPTV in P2P networks in order to provide an ef-ficient scalable real-time distribution and robust P2P stream-ing services PALs [8] present a new framework for quality

Hindawi Publishing CorporationInternational Journal of Digital Multimedia BroadcastingVolume 2015 Article ID 283097 10 pageshttpdxdoiorg1011552015283097

2 International Journal of Digital Multimedia Broadcasting

adaptive playback in P2P video streaming system HoweverPALs only consider single-dimensional scalability whichcannot adapt to heterogeneous static resources of peers andnetwork condition Chameleon [9] is a new adaptive P2Pstreaming protocol that combines the advantages of networkcoding (NC) and Scalable Video Coding (SVC) The core ofchameleon is studied including neighbor selection qualityadaptation receiver-driven peer coordination and senderselection with different design options But it cannot adaptto dynamic bandwidth and catastrophic failures Authors in[10] propose simple modifications to existing protocols thathave the potential to lead to significant benefits in terms oflatency jitter throughput packet loss and PSNR Abboudet al [11] developed quality adaptive in P2P VoD systemsbased on SVC Quality adaptation mechanisms are mainlycomposed of two stages The IQA is used for adapting thehighest possible layerwith static resources (Screen resolutionbandwidth and processing power) Moreover the PQA [11]is responsible for adjusting the layer according to the changesof dynamic resources (network condition and throughput)However it cannot adapt the Quantization Parameter (QP)with instantaneous throughput available and peer churn

In contrast to thementioned pieces of work in this paperwe present our P2PTV system with full support for qualityadaptation using SVC Therefore we use three-dimensionalscalability as defined by theH264SVC standard [12] to adaptto different peer resources and available real-time resources

Our results show that our proposed algorithms can pro-vide the best quality level adaptation with available resourcesand contribute to giving more robustness against heteroge-neous peers and high churn rates LLI [13] is demonstrated viathe effectiveness of quality adaptationwith the heterogeneousnetwork terminal devices that have distinct characteristics interms of subjective quality for the actual playback we usea modified version of the MPlayer [14] that supports SVC[12] scaling method MOS and objective QoE metric likeSSIM By the same token LLA [13] is evaluated and comparedwith PQA [11] on PSNR metric of objective QoE using PSIMsimulator [15] Deployment and evaluation show that theLLA can achieve a remarkable video streaming PSNR valueincrease

The rest of this paper is organized as follows Backgroundon SVC P2PTV streaming adaptive streaming and Qualityof Experience are given in Section 2 The proposed of ourquality adaptation algorithms that use SVC is described indetail in Section 3 Section 4 illustrates the simulation resultsas well as our analysis andwe conclude the paper in Section 5

2 Background

21 Scalable Video Coding Scalable Video Coding (SVC)[16 17] is an extension of the H264MPEG-4 AdvancedVideo Coding standard [18] which introduces a speciallayered encoding this is similar to progressive JPEG usedfor image compression and transmission over the Web Avideo encoded in SVC format is created composing multiplesubstreams derived from the original video signal Eachsubstream can be transmitted independently from the others

However in order to reconstruct the video the client hasto start from the base layer and then sequentially use theavailable improvement layers This way starting from asingle encoded video stream is possible to achieve a multibitrate streaming service In contrast other currently usedtechnologies require separate encoding of each individualstream at different bit rate This operation needs to be doneonce if the streams are stored as independent files or eachtime the content is required if done on the fly In the firstcase for a typical streaming service between 3 and morethan 10 copies of each video are created and stored onthe transmitting source introducing a great redundancyWhile the storage costs are becoming less of a factor thisapproach greatly increases the initial efforts and the ongoingmaintenance complexity of the system In the second case weremove the storage requirements but the encoding processhas significantly higher computing cost SVC tries to reducethose costs while keeping encoding efficiency and an efficientgranularity

22 P2PTV Streaming Traditional Internet TV servicesbased on a simple unicast approach are restricted tomoderatenumbers of clientsThe overwhelming resource requirementsmake these solutions impossible when the number of usersgrows to millions By multiplying servers and creating acontent distribution network (CDN) the solution will scaleonly to a larger audience with regard to the number ofdeployed servers which may be limited by infrastructurecosts Finally the lack of widespread deployment of IP-multicast limits the availability and scope of this solution fora TV service on the Internet scale Therefore the use of P2Poverlay networks [19] to deliver live television in the Internet(P2PTV) is achieving popularity and has been considered asa promising alternative to IP unicast and multicast models[20] The raising popularity of this solution is confirmedby the amount of new P2PTV [21] applications that havebecome available including PPLive SOPCast Tvants TVU-Player Joost [22] Babelgum and Zattoo and by constantlyincreasing amount of their users As the P2PTV is notwithoutdrawbacks currently the popularity is gaining solutionscombiningmulticast CDN and P2P approaches However incertain performance evaluation scenarios the components ofsuch hybrid systems can be considered separately The nodesin a P2PTV network called peers self-organize themselvesto act both as clients and servers to exchange TV contentbetween themselves

23 Adaptive Streaming Adaptive streaming [23 24] has agood potential to replace widely used progressive downloadAdaptive streaming can dynamically adjust the video bitrateto the varying available bandwidth and prevent prefetchingtoo much future video data when the extra bandwidthis available but the data are eventually left unused Foradaptive streaming video servers need to maintain multiplecopies of the same video with different bitrates for differentclients and clients with different kinds of connectivity whichrequires additional server storage and reduces cache hit ratioRecently Scalable Video Coding (H264SVC) [17 18] is

International Journal of Digital Multimedia Broadcasting 3

Table 1 Mapping of oQoE to sQoE

MOS PSNR SSIM5 ge45 gt0994 ge33 amp lt45 ge095 amp lt0993 ge274 amp lt33 ge088 amp lt0952 ge187 amp lt274 ge05 amp lt0881 lt187 lt05

considered to be able to save server storage and increase hitratio using the existing web cache infrastructure However arate adaptation algorithm still needs to be carefully designedfor streaming scalable video [25]

24 Quality of Experience Quality of Experience [26ndash28]is defined as the subjectively perceived acceptability of aservice [29] The perceived quality can be investigated insubjective tests where presented stimuli such as impairedvideo sequences are rated by subjects under controlled condi-tions The obtained rating expresses the subjective Quality ofExperience (sQoE) typically described by theMean OpinionScore (MOS) However subjective tests are time-consumingand expensive and have to be undertaken manually whichdoes not allow for automatic quality ratings by software Thisaspect motivates objective metrics which are designed tocorrelate with human perception and thus avoid cost andtime intensive empirical evaluations Estimates for the qualityobtained bymetrics are called objectiveQuality of Experience(oQoE) In this paper we focus on two full reference metricsPSNR and SSIM PSNR [30] is commonly used formeasuringpicture quality degradation for the received image qualitycompared to the original image whereas SSIM [31] indexmeasures the structural similarity between the original andthe received image Based on results obtained for still imagesin [31] we introduce a mapping of PSNR and SSIM (oQoE)to a nominal 5-point MOS scale (sQoE) according to Table 1for expressing an approximation of sQoE

3 Proposed Quality Adaptive Streaming

In this section we present the main idea of our proposednovel quality adaptation algorithms using SVC and runningat every peer in the P2PTV Figure 1 depicts the basic archi-tecture of the quality adaptation loop

Quality adaptation is achieved by adjusting qualityaccording to the different peer resources and networkdynamics It is performed by two modules the Layer LevelInitialization (LLI) and the Layer Level Adjustment (LLA)Both modules form the algorithms that match the layerswith resources available at the peer On one hand the LLIis used for determining the highest possible layer that apeer can retrieve and play and is performed at session startOn the other hand the LLA is performed periodically toadjust the layer according to the dynamic changes of thenetwork environment Next we give more details on thequality adaptation modules and their role in the P2PTVsystem

Chunkblock selection

Tracker

Streaming

Layer Level Initialization (LLI)

Layer LevelAdjustment (LLA)

MPlayerQoE

Peer selection

Figure 1 The quality adaptive P2PTV system

31 Layer Level Initialization (LLI) Layer Level Initialization(LLI) is invoked at the beginning of video playback in order toavoid long startup times and match resources at session startThe architecture of the LLI is depicted in Figure 2

The basic idea behind the LLI is to compare the require-ments of each SVC video quality with the local resources ofeach user device The subtle property of the LLI is that ithas to make a decision on the layer level without having anyinformation about effective network conditions and systemdynamics An initial layer set 119871

0= (1198890 1199050 1199020) with base

quality level is populated at first According to the threedimensions of scalability in the SVC model we have identi-fied the following relevant local resources of a devices power(CPU RAM and battery life) and user preference (displayresolution frame rate and PSNR level) And then bitrateadaptation complexity adaptation [32] and distributionsvideo length relative battery life [33 34] adaptation modulesselect out all compatible quality level based on static resourcesof a peer considering the user preference limitations All thecompatible combinations are appended as candidates thefinal decision is made by selecting the filtered layer level 119871

119897=

(119889119897 119905119897 119902119897) set 119871119889119897 119905119897119902119897LLI where values of all three dimensions of

scalability are at their maximum

32 Layer Level Adjustment (LLA) The Layer Level Adjust-ment (LLA) module is the dynamic part of the qualityadaptation loop The LLA is activated periodically duringthe SVC video playback every time interval Its main task isto adjust the selected SVC layer according to the real-timeinformation in order to predict possible stalls before theyhappen and avoid stalls by temporary switching layer Thearchitecture of the LLA is depicted in Figure 3 LLA uses timevarying information regarding block availability churn rateand throughput respectively to adapt to status adaptationmeasured through the SVC stream rate adaptation framerate adaptation Quantization Parameter (QP) bitrate adap-tation PSNR level adaptation and complexity adaptation Itstarts with the prefiltered of LLI output layer level parameters119871119889119897 119905119897119902119897

LLI that contains all quality level combinations supportedin terms of the static peer resources as explained aboveSimilar to the LLI the LLA final decision algorithm receives


Evaluate static peer resources

Screen resolution

Bandwidth

Device power

User preference

Display

Frame rate

SNR level

Spatial adaptation

Temporal adaptation

Quality adaptationCPU RAM Battery life

Complexity adaptation Video length

Bitrate adaptation

Final decision

Initial layer set L0 = (d0 t0 q0)

dl

tl

ql

Ld119897t119897 q119897

LLI

Figure 2 Algorithm Layer Level Initialization (LLI)

Evaluate real-timeresources

Block availability

Churn rate

Throughput

Peer resources

Device power

Status adaptation

Stream rateadaptation

Complexity adaptation

Bitrate adaptation

Final decision

Video SVC

Frame rate

Quantization Parameter(QP)

PSNR level

Initial layer set Ld119897t119897 q119897LLI

dl

t998400

l

q998400

l

Ld119897t

998400

119897q998400

119897

LLA

Figure 3 Algorithm Layer Level Adjustment (LLA)

the 1198711198891198971199051015840

1198971199021015840

119897

LLA from the different stages of the LLA and thenmakes a final decision on the three types of scalability to befetched

321 Status Adaptation This part of the LLA keeps track ofthe block availability of all connected peers Its objective isto check whether the current layer can be supported by theavailable blocks from current neighbors Using the net-status

adaptation the SVC layer of a peer can be adapted accordingto the real-time resources of its connected peers so thatthe playback does not need to stop and wait for unavailableblocks Consequently the number of stalls can be reducedduring the playback

322 Streaming Rate We define the streaming rate 119877 [35]as the total amount of received video data per second The


Li Li+1

QPi+1

QPi

Figure 4 State diagram of the Gilbert Elliot model used for SVClevel

average streaming rate is calculated across all active peers andrepresents a basic performance metric that is given by

119877 = (119906119901+119906119904

119873) (1)

where 119873 is the number of all active peers in the system Weuse 119880

119901to denote the average upload capacity participating

peers and 119880119904to denote the upload capacity of dedicated

streaming server

323 Churn Rate We refer to the ratio of the total number ofpeers 120582 that join the streaming system during the simulationtime to the total number 120583 of peers that leave the system asthe churn rate 120588 [36]

120588 =120582

120583 (2)

324 Bitrate Adaptation This step of the LLA changes theSVC layer by using the active download throughput Thegoal of the bitrate adaptation is to predict possible bufferunderruns due to slow block supply Therefore it adapts theSVC layer so that the bit rate fits the dynamic throughputthereby avoiding potential stalls

325 Quantization Parameter We have modeled the layerlevel of SVC video by using the Gilbert Elliot diagram [37]which is based on a two-state Markov model as a shown inFigure 4 where state 119871

119894represents the state of base layer

while state 119871119894+1

represents the state of enhancement layerThe transition from 119871

119894+1to 119871119894represents the Quantization

Parameter QP119894 In contrast The transition from 119871

119894to 119871119894+1

represents the Quantization Parameter QP119894+1

326 Complexity Adaptation The complexity adaptationcomponent uses a complexity model following the approachof [38] that works by mapping every set of quality levels (spa-tial temporal and SNR) into processor cycles required fordecoding the SVC coded video stream Based on definitionsin Table 2 decoding complexity of an SVC stream can be

Table 2 Symbols for analytical complexity model

Notation Description

119862119868119862119875119862119861

Average macroblock decoding complexity of119868-119875-119861 picture

CSCQAverage macroblock decoding complexity atspatialquality enhancement layers

119879119863119876 Total layer number for temporalspatialqualityscalability

119905119889119902 Layer index for temporalspatialquality scalability1198720 Number of macroblocks per picture120588 Portion of key pictures coded as 119868-pictures

Table 3 Simulation setup

Parameter ValueSimulation duration 10 minutesNumber of peers 180Number of servers 1Server upload capacity 4086KbpsVideo length 100 frames

calculated The complexity for decoding scalable streams isgiven by

119862GOP Dec = 1198720 (120588119862119868 + (1minus120588)119862119875 + (2119879(0)minus 1) 119862

119861)

+8119863+1 minus 1

72119879(0)1198720119876119862119876

+ 48119863+1minus 1

72119879(0)1198720 (119862119904 +119862119861)

(3)

4 Simulation

41 Experimental Setup We have implemented the proposedquality adaptive streaming in simulator PSIM [15] usingJava language Our implementation was validated by usingactual video streamTo conduct rigorous quantitative analysisof the proposed algorithms under wide range of workingconditions we implemented a testing application to emulatethe characteristics of realistic P2PTV systems This testingapplication enables us to conduct controllable and repeatableexperiments with different parameters and large number ofpeers The setup of our experiments is as follows Our simu-lation lasts for 10min with varied cross traffic to present thedynamic end-to-end resources We create a highly dynamicP2P streaming system with 180 heterogeneous peers that arerandomly and continually changing In addition we considerhaving one server with 4086Kbps The basic setup usedfor the performance evaluation is shown in Table 3 Theupload bandwidth values of peers are chosen according to thedistribution given in Table 4

Without losing generality we consider one video source[39] with length of 100 frames By using JSVM [40] the videosource is encoded into a total of 17 layers which contains1 layer for both spatial and quality scalability and 4 layersfor temporal scalability And the specific layer bitstream


Table 4 Resource configuration for the peers

Set 1 Set 2 Set 3Number 60 60 60Screen size 176 times 144 352 times 288 704 times 576Upload speed 128Kbps 320Kbps 800KbpsDownload speed 256 kbps 560Kbps 1200Kbps

Table 5 SVC bitstream information

Layer Resolution Frame rate Bitrate DTQ0 176 times 144 1875 3020 (0 0 0)1 176 times 144 375 4100 (0 1 0)2 176 times 144 75 5190 (0 2 0)3 176 times 144 15 622 (0 3 0)4 176 times 144 1875 7810 (0 0 1)5 176 times 144 375 10060 (0 1 1)6 176 times 144 75 12340 (0 2 1)7 176 times 144 15 14400 (0 3 1)8 352 times 288 1875 16690 (1 0 0)9 352 times 288 375 21940 (1 1 0)10 352 times 288 75 27620 (1 2 0)11 352 times 288 15 33180 (1 3 0)12 352 times 288 30 36940 (1 4 0)13 352 times 288 1875 33550 (1 0 1)14 352 times 288 375 42870 (1 1 1)15 352 times 288 75 53090 (1 2 1)16 352 times 288 15 63070 (1 3 1)17 352 times 288 30 70540 (1 4 1)

information is depicted in Table 5 As models for evaluatingthe perceived quality by end users for the resized anddisturbed video sequence we used PSNR metric and SSIMAn efficient implementation of these metrics is provided bythe MSU Video Quality Measurement Tool [41]

42 Experimental Results Now we evaluate how our pro-posed adaptation algorithms improve the performance of theP2PTV system and we simulate changing parameters to seehow the LLA reacts to themWe analyze the impact of severalsystem parameters on the performance and robustness ofour mechanisms especially in presence of heterogeneousnetwork terminal devices that have distinct characteristicsand high peer churn rates Also we simulate the scenariowithquality adaptation by LLI andwithout it in terms of subjectivequality of playback MPlayer subjective QoE estimation withMOS and objectivemetric with SSIMMoreover we estimatethe performance and efficiency of module LLA by comparingit with module PQA on objective QoE (PSNR metric)

421 Quality Adaptive by LLI Figure 5 [42] shows thereceived video quality at network terminal device in eachscenario that is with our proposed LLI mechanism andwithout applying it along with the expected video layersquality when using the four different quality layers Weobserved that LLImechanism improves the subjective quality

Table 6 Quality adaptation with the LLA compared to the PQA

Layer Impact of module LLAon PSNR

Impact of module PQAon PSNR

3 3422 3427 3516 343517 4072 365313 2784 270714 2713 26710 265 262712 3964 363411 2687 26659 2706 2668 2768 2708

as compared to the scenario without quality adaptation andprovides considerably a better quality Also we can notice inFigures 6 and 7 that the MOS values (sQoE) and the SSIMmeasurement (oQoE) correspond to the number of layers atthe peers in the scenario with quality adaptation that is muchhigher than that of without quality adaptation This can beexplained by the adaptation of Quantization Parameter (QP)with SVC layer for the end usersThe obtained results show anoticeable improvement in the overall QoE for the perceivedvideo at receiver end

422 Performance of LLA In this section we evaluate theperformance of LLA by comparing it with PQA [11] on thesystem P2PTV by varying the Quantization Parameter (QP)in the scenario of LLA but it fixed in the scenario of PQAfor each layer with fluctuation curve of peer throughput(Figure 8) The quality adaptation with the LLA compared tothe PQA is presented in Table 6 As shown in these results(Figures 9 and 10) we noticed that LLA module achievesbetter average PSNR value than PQA module in terms ofadapted QP for each layer with vibration of throughput

423 Impact of Churn Rate on Quantization Parameter (QP)In this scenario we enable the module LLA for adapting thereal time resources with the video stream in Table 5 and wemeasure the average streaming rate during live streamingsessions In this evaluation we study the impact of the churnrate on the Quantization Parameter (QP) of layer level duringstreaming and we consider a highly dynamic peer-to-peernetwork with highly frequent arrivals and departures ofpeers In highly dynamic peer-to-peer systems some peersjoin the system start streaming and also contribute theirresources to others At the same time other peers maybe leaving the system which will result in loss of uploadresources and perhaps disruption of some ongoing streamingsessions As the churn rate increases the network becomesmore dynamic We measure Quantization Parameter (QP) ofquality level perceived across all peers for each churn rateFor example a churn rate of 2 means that if 120593 number ofpeers leaves the system during the simulation time 2120593 newpeers will arrive during that period From Figure 11 we can


(a) Scenario without quality adaptation (b) Scenario with quality adaptation

Figure 5 Subjective quality in the scenario without quality adaptation and with quality adaptation by LLI

0

1

2

3

4

5

6

MO

S

14 17 7 13 15Layer

Scenario with quality adaptation Scenario without quality adaptation

Figure 6 MOS in the scenario without quality adaptation and withquality adaptation by LLI

see that the adaptation of bitrate average streaming rate andQuantization Parameter (QP) by LLA module with churnrate which varies between 1 and 10 When the churn rateincreases we observe that the bitrate and average streamingrate decrease But the important factor of QuantizationParameter (QP) increases and vice versa

The deployment and evaluation show that our mecha-nisms are adaptable with heterogeneous end terminals ofpeers and available real-time resources Indeed the LLI canprovide very good quality adaptation in terms of subjectivequality of playback MPlayer subjective QoE estimation withMOS and objective QoE by mean of the SSIM Furthermorethe LLA makes a remarkable increase in video quality valueunder fluctuation throughput available at the peer

075

08

085

09

095

1

105

SSIM

val

ue

14 17 7 13 15Layer

Scenario with quality adaptationScenario without quality adaptation

Figure 7 SSIM values in the scenario without quality adaptationand with quality adaptation by LLI

5 Conclusions

In this paper we have developed the quality adaptationmechanisms required for SVC-based P2PTV systems Thosemechanisms can adapt the video quality to both static anddynamic peer and system resources

The distinction between LLI and LLA is crucial in sep-arating adaptation stages of a streaming session The LLIassures that static resources of a peer are considered andmatched to prevent overloading which means it can adaptvideo quality with different capabilities of clientsrsquo devicesBut the LLA adapt SVC layer of a peer according to the real-time resources and fluctuations of network conditions so that


0

500

1000

1500

2000

2500

3000

1 2 3 4 5 6 7 8 9 10

Thro

ughp

ut (k

bps)

Time (min)

Figure 8 Fluctuation curve of peer throughput

0

10

20

30

40

50

36 34 28 28 28 30 30 30 30 30

PSN

R (d

B)

QP

Figure 9 Evaluation of the LLA impact of the Quantization Parameter (QP) on PSNR

0

10

20

30

40

36 36 36 36 36 36 36 36 36 36

PSN

R (d

B)

QP

Figure 10 Evaluation of the PQA impact of the Quantization Parameter (QP) on PSNR

the playback does not need to stop and wait for unavailableblocks Consequently the number of stalls can be reducedduring the playback

The simulation results showed that ourmechanisms reactquickly to various system changes while providing best qual-ity adaptation of scalable streaming with available resourcesand unpredictable network We have shown that LLI canhelp P2PTV system to achieve a better quality adaptationfor subjective quality of playback MPlayer subjective QoEwith MOS and objective of QoE assessment method withSSIM Furthermore we have demonstrated the effectiveness

of the LLA module compared to the PQA one It can be con-cluded that the adaptation using LLA enables a remarkableincrease in objective QoE of PSNR value compared to theexisting adaptation PQA mechanism while considering theQuantization Parameter (QP)

Conflict of Interests

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


0100200300400500600700800

1 5 4 3 6 7 8 2 9 10

Bitr

ate (

kbps

)

Churn rate(a) Bitrate adaptation versus churn rate

0

200

400

600

800

1 2 3 4 5 6 7 8 9 10

Aver

age s

tream

rate

(k

bits

)

Time (min)

(b) Average streaming rate adaptation by LLA module

05

10152025303540

1 2 3 4 5 6 7 8 9 10

QP

Time (min)

(c) Quantization Parameter (QP) adaptation by LLAmodule

02468

1012

1 2 3 4 5 6 7 8 9 10Ch

urn

rate

Time (min)

(d) Churn rate of P2PTV system

Figure 11 Bitrate average streaming rate andQuantization Parameter (QP) adaptation by LLAwith churn rate of peers in the P2PTV system

References

[1] Cisco Visual Networking Index Forecast and Methodology2013ndash2018 httpwwwciscocomcenussolutionscollateralservice-providerip-ngn-ip-next-generation-networkwhitepaper c11-481360html

[2] Y Lu J D D Mol F A Kuipers and P Van Mieghem ldquoAna-lytical model for mesh-based P2PVoDrdquo in Proceedings of the10th IEEE International Symposium on Multimedia (ISM rsquo08)pp 364ndash371 IEEE Berkeley Calif USA December 2008

[3] BitTorrent 2015 httpwwwbittorrentcom[4] SopCast 2015 httpwwwsopcastcom[5] PPLive 2015 httpwwwpplivecomenindexhtml[6] PPStream 2015 httpwwwppstreamcom[7] UUSee 2015 httpwwwuuseecom[8] R Rejaie and A Ortega ldquoPALS peer-to-peer adaptive layered

streamingrdquo in Proceedings of the 13th ACM International Work-shop on Network and Operating Systems Support for DigitalAudio and Video (NOSSDAV rsquo03) 2003

[9] A T Nguyen B Li and F Eliassen ldquoChameleon adaptive peer-to-peer streaming with network codingrdquo in Proceedings of theIEEE INFOCOM pp 1ndash9 IEEE San Diego Calif USA March2010

[10] M Alhaisoni A Liotta and M Ghanbari ldquoImprovingP2P streaming methods for IPTVrdquo International Journal onAdvances in Intelligent Systems vol 2 no 3 pp 354ndash365 2009

[11] O Abboud K Pussep A Kovacevic and R Steinmetz ldquoQualityadaptive peer-to-peer streaming using scalable video codingrdquo inWired-Wireless Multimedia Networks and ServicesManagement12th IFIPIEEE International Conference on Management ofMultimedia and Mobile Networks and Services MMNS 2009Venice Italy October 26-27 2009 Proceedings vol 5842 ofLecture Notes in Computer Science pp 41ndash54 Springer BerlinGermany 2009

[12] Mederic Blestel and Mickael Raulet Open SVC Decoder 2015httpsourceforgenetprojectsopensvcdecoder

[13] Y Lahbabi E H Elhaj and A Hammouch ldquoAdaptive P2PTVwith scalable video codingrdquo in Proceedings of the 4th IEEEInternational Colloquium in Information Science and Technology(CIST rsquo14) pp 364ndash369 IEEE TetouanMoroccoOctober 2014

[14] MPlayer 2015 httpwwwmplayerhqhu[15] 2014 httpnslcssfucaprojectsnc-svcnc-svctargz[16] V Menkovski and A Liotta ldquoQoE for mobile streamingrdquo in

Mobile MultimediamdashUser and Technology Perspectives chapter2 InTech Rijeka Croatia 2012

[17] T Wiegand G Sullivan J Reichel H Schwarz and M WienldquoJoint draft 9 of SVC amendment (revision 2)rdquo DocumentJVTV201 2007

[18] ITU-T Rec Advanced Video Coding for Generic AudiovisualServices Version 3 H264 amp ISOIEC 14496-10 AVC 2005

[19] A Liotta and L Lin ldquoThe operatorrsquos response to P2P service

demandrdquo IEEECommunicationsMagazine vol 45 no 7 pp 76ndash83 2007

[20] J Liu S G Rao B Li and H Zhang ldquoOpportunities and chal-lenges of peer-to-peer internet video broadcastrdquo Proceedings ofthe IEEE vol 96 no 1 pp 11ndash24 2008

[21] A Biernacki ldquoSimulating performance of a BitTorrent-basedP2P TV systemrdquo in Computer Networks vol 160 of Commu-nications in Computer and Information Science pp 448ndash458Springer Berlin Germany 2011

[22] M Alhaisoni and A Liotta ldquoCharacterization of signalling andtraffic in joostrdquo Peer-to-Peer Networking and Applications vol2 no 1 pp 75ndash83 2009

[23] V Adzic H Kalva and B Furht ldquoOptimizing video encodingfor adaptive streaming over HTTPrdquo IEEE Transactions onConsumer Electronics vol 58 no 2 pp 397ndash403 2012


[24] S Xiang L Cai and J Pan ldquoAdaptive scalable video streaminginwireless networksrdquo in Proceedings of the 3rd ACMMultimediaSystems Conference (MMSys rsquo12) Chapel Hill NC USA 2012

[25] V Menkovski G Exarchakos A C Sanchez and A LiottaldquoA quality of experience management modulerdquo InternationalJournal on Advances in Intelligent Systems vol 4 no 1-2 pp 13ndash19 2011

[26] B Staehle A Binzenhofer D Schlosser and B Boder ldquoQuanti-fying the influence of network conditions on the service qualityexperienced by a thin client userrdquo in Proceedings of the 14thGIITG Conference on Measuring Modelling and Evaluation ofComputer and Communication Systems (MMB rsquo08) DortmundGermany April 2008

[27] T Zinner O Abboud O Hohlfeld T Hossfeld and P Tran-Gia ldquoTowards QoE management for scalable video streamingrdquoin Proceedings of the 21st ITC Specialist Seminar on Multi-media ApplicationsmdashTraffic Performance and QoE ProgramMiyazaki Japan March 2010

[28] M Alhaisoni M Ghanbari and A Liotta ldquoLocalized mul-tistreams for P2P streamingrdquo International Journal of DigitalMultimedia Broadcasting vol 2010 Article ID 843574 12 pages2010

[29] International Telecommunication Union ldquoDefinition of qualityof experiencerdquo ITU-T Delayed Contribution D197 2004

[30] Q Huynh-Thu and M Ghanbari ldquoScope of validity of PSNR inimagevideo quality assessmentrdquo Electronics Letters vol 44 no13 pp 800ndash801 2008

[31] ZWang A C Bovik H R Sheikh and E P Simoncelli ldquoImagequality assessment from error visibility to structural similarityrdquoIEEE Transactions on Image Processing vol 13 no 4 pp 600ndash612 2004

[32] Z Ma and Y Wang ldquoComplexity modeling of scalable videodecodingrdquo in Proceedings of the IEEE International Conferenceon Acoustics Speech and Signal Processing (ICASSP rsquo08) pp1125ndash1128 Las Vegas Nev USA March 2008

[33] Y Lahbabi E H Ibn Elhaj and A Hammouch ldquoQualityadaptation using scalable video coding (SVC) in peer-to-peer(P2P) video-on-demand (VoD) Streamingrdquo in Proceedings ofthe 4th International Conference on Multimedia Computing andSystems (ICMCS rsquo14) vol 7 Marrakesh Morocco April 2014

[34] S Lanka ldquoOptimise UI power using wake-on-approach (part1)rdquo EE Times vol 4 2014

[35] C Feng and B Li Network Coding for Content Distribution andMultimedia Streaming in Peer-to-Peer Networks Department ofElectrical and Computer Engineering University of Toronto2012

[36] H Luan K-W Kwong X Hei and D H K Tsang ldquoAdaptivetopology formation for peer-to-peer video streamingrdquo Peer-to-PeerNetworking andApplications vol 3 no 3 pp 186ndash207 2010

[37] M Mushkin and I Bar-David ldquoCapacity and coding forthe Gilbert-Elliot channelsrdquo IEEE Transactions on InformationTheory vol 35 no 6 pp 1277ndash1290 1989

[38] Y Lahbabi E H Ibn Elhaj and A Hammouch ldquoQuality ad-aptation using scalable video coding (SVC) in peer-to-peer(P2P) video-on-demand (VoD) streamingrdquo in Proceedings ofthe 4th International Conference on Multimedia Computing andSystems (ICMCS rsquo14) Marrakesh Morocco April 2014

[39] 2014 httpevalsvcgooglecodecomsvntrunkjsvmbinOut-put20svc20videos

[40] JSVM ldquoJSVM Softwarerdquo CVS Repository of JSVM pserv-erjvtusergarconientrwth-aachendecvsjvt

[41] MSU Quality Measurement Tool httpcompressionruvideoquality measureinfo enhtml

[42] 2015 httpwwwmediafirecomdownloadj8zt896stsdhw74scenario+without+and+with+quality+adaptationrar

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Active and Passive Electronic Components

Control Scienceand Engineering

Journal of



RotatingMachinery


Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics


Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation



DistributedSensor Networks



adaptive playback in P2P video streaming system HoweverPALs only consider single-dimensional scalability whichcannot adapt to heterogeneous static resources of peers andnetwork condition Chameleon [9] is a new adaptive P2Pstreaming protocol that combines the advantages of networkcoding (NC) and Scalable Video Coding (SVC) The core ofchameleon is studied including neighbor selection qualityadaptation receiver-driven peer coordination and senderselection with different design options But it cannot adaptto dynamic bandwidth and catastrophic failures Authors in[10] propose simple modifications to existing protocols thathave the potential to lead to significant benefits in terms oflatency jitter throughput packet loss and PSNR Abboudet al [11] developed quality adaptive in P2P VoD systemsbased on SVC Quality adaptation mechanisms are mainlycomposed of two stages The IQA is used for adapting thehighest possible layerwith static resources (Screen resolutionbandwidth and processing power) Moreover the PQA [11]is responsible for adjusting the layer according to the changesof dynamic resources (network condition and throughput)However it cannot adapt the Quantization Parameter (QP)with instantaneous throughput available and peer churn

In contrast to thementioned pieces of work in this paperwe present our P2PTV system with full support for qualityadaptation using SVC Therefore we use three-dimensionalscalability as defined by theH264SVC standard [12] to adaptto different peer resources and available real-time resources

Our results show that our proposed algorithms can pro-vide the best quality level adaptation with available resourcesand contribute to giving more robustness against heteroge-neous peers and high churn rates LLI [13] is demonstrated viathe effectiveness of quality adaptationwith the heterogeneousnetwork terminal devices that have distinct characteristics interms of subjective quality for the actual playback we usea modified version of the MPlayer [14] that supports SVC[12] scaling method MOS and objective QoE metric likeSSIM By the same token LLA [13] is evaluated and comparedwith PQA [11] on PSNR metric of objective QoE using PSIMsimulator [15] Deployment and evaluation show that theLLA can achieve a remarkable video streaming PSNR valueincrease

The rest of this paper is organized as follows Backgroundon SVC P2PTV streaming adaptive streaming and Qualityof Experience are given in Section 2 The proposed of ourquality adaptation algorithms that use SVC is described indetail in Section 3 Section 4 illustrates the simulation resultsas well as our analysis andwe conclude the paper in Section 5

2 Background

21 Scalable Video Coding Scalable Video Coding (SVC)[16 17] is an extension of the H264MPEG-4 AdvancedVideo Coding standard [18] which introduces a speciallayered encoding this is similar to progressive JPEG usedfor image compression and transmission over the Web Avideo encoded in SVC format is created composing multiplesubstreams derived from the original video signal Eachsubstream can be transmitted independently from the others

However in order to reconstruct the video the client hasto start from the base layer and then sequentially use theavailable improvement layers This way starting from asingle encoded video stream is possible to achieve a multibitrate streaming service In contrast other currently usedtechnologies require separate encoding of each individualstream at different bit rate This operation needs to be doneonce if the streams are stored as independent files or eachtime the content is required if done on the fly In the firstcase for a typical streaming service between 3 and morethan 10 copies of each video are created and stored onthe transmitting source introducing a great redundancyWhile the storage costs are becoming less of a factor thisapproach greatly increases the initial efforts and the ongoingmaintenance complexity of the system In the second case weremove the storage requirements but the encoding processhas significantly higher computing cost SVC tries to reducethose costs while keeping encoding efficiency and an efficientgranularity

22 P2PTV Streaming Traditional Internet TV servicesbased on a simple unicast approach are restricted tomoderatenumbers of clientsThe overwhelming resource requirementsmake these solutions impossible when the number of usersgrows to millions By multiplying servers and creating acontent distribution network (CDN) the solution will scaleonly to a larger audience with regard to the number ofdeployed servers which may be limited by infrastructurecosts Finally the lack of widespread deployment of IP-multicast limits the availability and scope of this solution fora TV service on the Internet scale Therefore the use of P2Poverlay networks [19] to deliver live television in the Internet(P2PTV) is achieving popularity and has been considered asa promising alternative to IP unicast and multicast models[20] The raising popularity of this solution is confirmedby the amount of new P2PTV [21] applications that havebecome available including PPLive SOPCast Tvants TVU-Player Joost [22] Babelgum and Zattoo and by constantlyincreasing amount of their users As the P2PTV is notwithoutdrawbacks currently the popularity is gaining solutionscombiningmulticast CDN and P2P approaches However incertain performance evaluation scenarios the components ofsuch hybrid systems can be considered separately The nodesin a P2PTV network called peers self-organize themselvesto act both as clients and servers to exchange TV contentbetween themselves

23 Adaptive Streaming Adaptive streaming [23 24] has agood potential to replace widely used progressive downloadAdaptive streaming can dynamically adjust the video bitrateto the varying available bandwidth and prevent prefetchingtoo much future video data when the extra bandwidthis available but the data are eventually left unused Foradaptive streaming video servers need to maintain multiplecopies of the same video with different bitrates for differentclients and clients with different kinds of connectivity whichrequires additional server storage and reduces cache hit ratioRecently Scalable Video Coding (H264SVC) [17 18] is










Tracker

Streaming



MPlayerQoE

Peer selection




0= (1198890 1199050 1199020) with base


119897=







Screen resolution

Bandwidth

Device power

User preference

Display

Frame rate

SNR level

Spatial adaptation

Temporal adaptation



Bitrate adaptation

Final decision


dl

tl

ql

Ld119897t119897 q119897

LLI



Block availability

Churn rate

Throughput

Peer resources

Device power

Status adaptation



Bitrate adaptation

Final decision

Video SVC

Frame rate


PSNR level


dl

t998400

l

q998400

l

Ld119897t

998400

119897q998400

119897

LLA


the 1198711198891198971199051015840

1198971199021015840

119897






Li Li+1

QPi+1

QPi



119877 = (119906119901+119906119904

119873) (1)




streaming server


120588 =120582

120583 (2)




while state 119871119894+1




119894to 119871119894+1





119862119868119862119875119862119861









119861)

+8119863+1 minus 1

72119879(0)1198720119876119862119876

+ 48119863+1minus 1

72119879(0)1198720 (119862119904 +119862119861)

(3)

4 Simulation














3 3422 3427 3516 343517 4072 365313 2784 270714 2713 26710 265 262712 3964 363411 2687 26659 2706 2668 2768 2708







0

1

2

3

4

5

6

MO

S

14 17 7 13 15Layer





075

08

085

09

095

1

105

SSIM

val

ue

14 17 7 13 15Layer



5 Conclusions




0

500

1000

1500

2000

2500

3000

1 2 3 4 5 6 7 8 9 10

Thro

ughp

ut (k

bps)

Time (min)


0

10

20

30

40

50

36 34 28 28 28 30 30 30 30 30

PSN

R (d

B)

QP


0

10

20

30

40

36 36 36 36 36 36 36 36 36 36

PSN

R (d

B)

QP








0100200300400500600700800

1 5 4 3 6 7 8 2 9 10

Bitr

ate (

kbps

)


0

200

400

600

800

1 2 3 4 5 6 7 8 9 10

Aver

age s

tream

rate

(k

bits

)

Time (min)


05

10152025303540

1 2 3 4 5 6 7 8 9 10

QP

Time (min)


02468

1012

1 2 3 4 5 6 7 8 9 10Ch

urn

rate

Time (min)



References










































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation


















Tracker

Streaming



MPlayerQoE

Peer selection




0= (1198890 1199050 1199020) with base


119897=







Screen resolution

Bandwidth

Device power

User preference

Display

Frame rate

SNR level

Spatial adaptation

Temporal adaptation



Bitrate adaptation

Final decision


dl

tl

ql

Ld119897t119897 q119897

LLI



Block availability

Churn rate

Throughput

Peer resources

Device power

Status adaptation



Bitrate adaptation

Final decision

Video SVC

Frame rate


PSNR level


dl

t998400

l

q998400

l

Ld119897t

998400

119897q998400

119897

LLA


the 1198711198891198971199051015840

1198971199021015840

119897






Li Li+1

QPi+1

QPi



119877 = (119906119901+119906119904

119873) (1)




streaming server


120588 =120582

120583 (2)




while state 119871119894+1




119894to 119871119894+1





119862119868119862119875119862119861









119861)

+8119863+1 minus 1

72119879(0)1198720119876119862119876

+ 48119863+1minus 1

72119879(0)1198720 (119862119904 +119862119861)

(3)

4 Simulation














3 3422 3427 3516 343517 4072 365313 2784 270714 2713 26710 265 262712 3964 363411 2687 26659 2706 2668 2768 2708







0

1

2

3

4

5

6

MO

S

14 17 7 13 15Layer





075

08

085

09

095

1

105

SSIM

val

ue

14 17 7 13 15Layer



5 Conclusions




0

500

1000

1500

2000

2500

3000

1 2 3 4 5 6 7 8 9 10

Thro

ughp

ut (k

bps)

Time (min)


0

10

20

30

40

50

36 34 28 28 28 30 30 30 30 30

PSN

R (d

B)

QP


0

10

20

30

40

36 36 36 36 36 36 36 36 36 36

PSN

R (d

B)

QP








0100200300400500600700800

1 5 4 3 6 7 8 2 9 10

Bitr

ate (

kbps

)


0

200

400

600

800

1 2 3 4 5 6 7 8 9 10

Aver

age s

tream

rate

(k

bits

)

Time (min)


05

10152025303540

1 2 3 4 5 6 7 8 9 10

QP

Time (min)


02468

1012

1 2 3 4 5 6 7 8 9 10Ch

urn

rate

Time (min)



References










































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation











Screen resolution

Bandwidth

Device power

User preference

Display

Frame rate

SNR level

Spatial adaptation

Temporal adaptation



Bitrate adaptation

Final decision


dl

tl

ql

Ld119897t119897 q119897

LLI



Block availability

Churn rate

Throughput

Peer resources

Device power

Status adaptation



Bitrate adaptation

Final decision

Video SVC

Frame rate


PSNR level


dl

t998400

l

q998400

l

Ld119897t

998400

119897q998400

119897

LLA


the 1198711198891198971199051015840

1198971199021015840

119897






Li Li+1

QPi+1

QPi



119877 = (119906119901+119906119904

119873) (1)




streaming server


120588 =120582

120583 (2)




while state 119871119894+1




119894to 119871119894+1





119862119868119862119875119862119861









119861)

+8119863+1 minus 1

72119879(0)1198720119876119862119876

+ 48119863+1minus 1

72119879(0)1198720 (119862119904 +119862119861)

(3)

4 Simulation














3 3422 3427 3516 343517 4072 365313 2784 270714 2713 26710 265 262712 3964 363411 2687 26659 2706 2668 2768 2708







0

1

2

3

4

5

6

MO

S

14 17 7 13 15Layer





075

08

085

09

095

1

105

SSIM

val

ue

14 17 7 13 15Layer



5 Conclusions




0

500

1000

1500

2000

2500

3000

1 2 3 4 5 6 7 8 9 10

Thro

ughp

ut (k

bps)

Time (min)


0

10

20

30

40

50

36 34 28 28 28 30 30 30 30 30

PSN

R (d

B)

QP


0

10

20

30

40

36 36 36 36 36 36 36 36 36 36

PSN

R (d

B)

QP








0100200300400500600700800

1 5 4 3 6 7 8 2 9 10

Bitr

ate (

kbps

)


0

200

400

600

800

1 2 3 4 5 6 7 8 9 10

Aver

age s

tream

rate

(k

bits

)

Time (min)


05

10152025303540

1 2 3 4 5 6 7 8 9 10

QP

Time (min)


02468

1012

1 2 3 4 5 6 7 8 9 10Ch

urn

rate

Time (min)



References










































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










Li Li+1

QPi+1

QPi



119877 = (119906119901+119906119904

119873) (1)




streaming server


120588 =120582

120583 (2)




while state 119871119894+1




119894to 119871119894+1





119862119868119862119875119862119861









119861)

+8119863+1 minus 1

72119879(0)1198720119876119862119876

+ 48119863+1minus 1

72119879(0)1198720 (119862119904 +119862119861)

(3)

4 Simulation














3 3422 3427 3516 343517 4072 365313 2784 270714 2713 26710 265 262712 3964 363411 2687 26659 2706 2668 2768 2708







0

1

2

3

4

5

6

MO

S

14 17 7 13 15Layer





075

08

085

09

095

1

105

SSIM

val

ue

14 17 7 13 15Layer



5 Conclusions




0

500

1000

1500

2000

2500

3000

1 2 3 4 5 6 7 8 9 10

Thro

ughp

ut (k

bps)

Time (min)


0

10

20

30

40

50

36 34 28 28 28 30 30 30 30 30

PSN

R (d

B)

QP


0

10

20

30

40

36 36 36 36 36 36 36 36 36 36

PSN

R (d

B)

QP








0100200300400500600700800

1 5 4 3 6 7 8 2 9 10

Bitr

ate (

kbps

)


0

200

400

600

800

1 2 3 4 5 6 7 8 9 10

Aver

age s

tream

rate

(k

bits

)

Time (min)


05

10152025303540

1 2 3 4 5 6 7 8 9 10

QP

Time (min)


02468

1012

1 2 3 4 5 6 7 8 9 10Ch

urn

rate

Time (min)



References










































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation




















3 3422 3427 3516 343517 4072 365313 2784 270714 2713 26710 265 262712 3964 363411 2687 26659 2706 2668 2768 2708







0

1

2

3

4

5

6

MO

S

14 17 7 13 15Layer





075

08

085

09

095

1

105

SSIM

val

ue

14 17 7 13 15Layer



5 Conclusions




0

500

1000

1500

2000

2500

3000

1 2 3 4 5 6 7 8 9 10

Thro

ughp

ut (k

bps)

Time (min)


0

10

20

30

40

50

36 34 28 28 28 30 30 30 30 30

PSN

R (d

B)

QP


0

10

20

30

40

36 36 36 36 36 36 36 36 36 36

PSN

R (d

B)

QP








0100200300400500600700800

1 5 4 3 6 7 8 2 9 10

Bitr

ate (

kbps

)


0

200

400

600

800

1 2 3 4 5 6 7 8 9 10

Aver

age s

tream

rate

(k

bits

)

Time (min)


05

10152025303540

1 2 3 4 5 6 7 8 9 10

QP

Time (min)


02468

1012

1 2 3 4 5 6 7 8 9 10Ch

urn

rate

Time (min)



References










































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation












0

1

2

3

4

5

6

MO

S

14 17 7 13 15Layer





075

08

085

09

095

1

105

SSIM

val

ue

14 17 7 13 15Layer



5 Conclusions




0

500

1000

1500

2000

2500

3000

1 2 3 4 5 6 7 8 9 10

Thro

ughp

ut (k

bps)

Time (min)


0

10

20

30

40

50

36 34 28 28 28 30 30 30 30 30

PSN

R (d

B)

QP


0

10

20

30

40

36 36 36 36 36 36 36 36 36 36

PSN

R (d

B)

QP








0100200300400500600700800

1 5 4 3 6 7 8 2 9 10

Bitr

ate (

kbps

)


0

200

400

600

800

1 2 3 4 5 6 7 8 9 10

Aver

age s

tream

rate

(k

bits

)

Time (min)


05

10152025303540

1 2 3 4 5 6 7 8 9 10

QP

Time (min)


02468

1012

1 2 3 4 5 6 7 8 9 10Ch

urn

rate

Time (min)



References










































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










0

500

1000

1500

2000

2500

3000

1 2 3 4 5 6 7 8 9 10

Thro

ughp

ut (k

bps)

Time (min)


0

10

20

30

40

50

36 34 28 28 28 30 30 30 30 30

PSN

R (d

B)

QP


0

10

20

30

40

36 36 36 36 36 36 36 36 36 36

PSN

R (d

B)

QP








0100200300400500600700800

1 5 4 3 6 7 8 2 9 10

Bitr

ate (

kbps

)


0

200

400

600

800

1 2 3 4 5 6 7 8 9 10

Aver

age s

tream

rate

(k

bits

)

Time (min)


05

10152025303540

1 2 3 4 5 6 7 8 9 10

QP

Time (min)


02468

1012

1 2 3 4 5 6 7 8 9 10Ch

urn

rate

Time (min)



References










































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










0100200300400500600700800

1 5 4 3 6 7 8 2 9 10

Bitr

ate (

kbps

)


0

200

400

600

800

1 2 3 4 5 6 7 8 9 10

Aver

age s

tream

rate

(k

bits

)

Time (min)


05

10152025303540

1 2 3 4 5 6 7 8 9 10

QP

Time (min)


02468

1012

1 2 3 4 5 6 7 8 9 10Ch

urn

rate

Time (min)



References










































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation









research article adaptive streaming of scalable videos...

Documents