Research Article Cloud-Assisted Scalable Video Delivery ...downloads.hindawi.com/journals/ijdsn/2015/205106.pdfcommunity-based video streaming systems have attracted increasing research

Research ArticleCloud-Assisted Scalable Video Delivery Solution overMobile Ad Hoc Networks

Lujie Zhong1 and Shijie Jia2

1 Information Engineering College Capital Normal University Beijing 100048 China2Academy of Information Technology Luoyang Normal University Luoyang 471022 China

Correspondence should be addressed to Shijie Jia shjjiagmailcom

Received 14 October 2014 Accepted 17 December 2014

Academic Editor Liang Zhou

Copyright copy 2015 L Zhong and S Jia 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

Cloud computing is promising avenue for supporting high-performance and interactive streaming serviceMaking use of the cloudsto flexibly increase the scale of video service system and provide fast search are key determinants for scalablemobile video streamingin order to ensure smooth playback experience In this paper we propose a novel Cloud-assisted Scalable Video Delivery Solutionover mobile ad hoc networks (CSVD) CSVD makes use of the clouds to share responsibility for the load of resource managementof media server which supports fast resource searching and enhances system scalability CSVD designs a new estimation model ofresourcemaintenance scale in terms of quality of service- (QoS-) oriented dynamic balance between supply and demand in order toeconomically use the clouds A novel supplier scheduling algorithm that assigns resource suppliers for fast responding user requestin terms of their load and serving capacity is proposed Extensive tests show how CSVD achieves much better performance resultsin comparison with other state-of-the-art solutions

1 Introduction

The mobile ad hoc networks (MANETs) are the significantnetwork technologies for the next generation Internet whichhave extensive application areas [1ndash3] The new wirelesscommunication protocol such as IEEE 80211 can meet highbandwidth requirement of multimedia services to enableprovision of rich visual content for the mobile users inMANETs [4] Video streaming is a significant one of themultimedia services which provides rich content in multi-ple network environment such as MANETs VANETs andwireless sensor networks (WSN) [5ndash10] P2P technologiesare well known for supporting large-scale video streamingsystem deployment [11ndash15] However provision of P2P-basedhigh-quality video streaming service with efficient contentsharing over MANETs is a challenging issue Due to limitedcapacities of energy and storage the mobile nodes only cacherelatively short video clip so as to frequently replace the datain the playback buffer for watching desired content Search-ing requested video content from fragmentary distributedresources in P2P networks leads to long start delay andhigh-cost network bandwidth which cannot ensure smooth

playback and meet the demand of green communication Onthe other hand the pursuit of popular video content resultsin high load of system due to process request and scheduleresources reducing system scalabilityTherefore a light-dutysolution which efficiently maintains and schedules resourcescarried by themobile nodes and supports fast search for videocontent should be considered for video streaming service inMANETs

Numerous researchers have shown great interest in high-efficiency resource sharing for video streaming system inwireless networks For instance QUVoD in [16] a Chord-based video sharing solution over VANETs groups peers intoa chained Chord structure in 4G networks in terms of thesimilarity of stored video chunks which can achieve reliablesupply and fast location of video resources SURFNet in [17]is a tree-based video sharing solution in which the peers withlong online times are grouped into an AVL tree and connectwith an attached holder-chainwhose items have similar videocontent However with increasing number of nodes the highmaintenance cost for structured topology (Chordtree) limitsthe scalability of these solutions For uncertain blowout ofuser access for video resources enlarging the scale of server

Hindawi Publishing CorporationInternational Journal of Distributed Sensor NetworksVolume 2015 Article ID 205106 10 pageshttpdxdoiorg1011552015205106

2 International Journal of Distributed Sensor Networks

Internet

Media server

Cloud

Figure 1 CSVD architecture

cluster increases the cost of system deployment SPOONin [18] is a community-based file sharing solution overMANETs SPOON groups the peers into multiple commu-nities in terms of the interest similarity which can achievehigh efficiency of content sharing However the stability ofcommunity structure relies on the capacities of communitycoordinator and determines system performance and main-tenance cost of communities Making use of the server tocompensate insufficient bandwidth further limits the systemscale

Recently cloud computing has become the most popu-lar computing paradigm [19] Providing on-demand serverresources to users relies on the shared pool of servers indatacenters [20] With rapid growth of mobile devices usagemobile cloud streaming becomes a promising avenue for sup-porting high-performance and interactive streaming service[21 22]Themost of cloud-based multimedia systems use theclouds to compensate insufficient capacities of upload band-width computing and storage in the systems However therandom request behavior of users for the video content leadsto the increase in the complexity of cloud usage and the costof interactivity between server and clouds Moreover makinguse of the clouds to meet the requirement of bandwidthbrings expensive monetary cost

In this paper we propose a novel Cloud-assisted ScalableVideo Delivery solution over mobile ad hoc networks (CSVD)

As Figure 1 shows the clouds assist the media server tomanage the hotspot resources to address the large-scaleintensive request when the server cannotmeet the demand ofusers for video resources A novel estimation model ofresource maintenance scale based on QoS-oriented dynamicbalance between supply and demand is proposed reducingmonetary cost from rental resources in the clouds A novelsupplier scheduling mechanism that schedules suppliers interms of their load and estimated processing capacity is pro-posed reducing the lookup delay and balancing the load ofsuppliers Simulation results show how CSVD achieves muchbetter performance results in comparisonwith other state-of-the-art solutions

2 Related Work

There have been numerous studies on P2P-based man-agement and lookup optimization of resources for videostreaming services in recent years

The solutions based on structured content distributiontopology are well known for resource lookup efficiency Forinstance QUVoD in [16] employs a group-based Chordstructure to uniformly distribute the video content where thepeers which store similar content in the Chord overlay form agroup By using this structure the request for sequential video

International Journal of Distributed Sensor Networks 3

chunks can be addressed in a group reducing chunk seekingtraffic and balancing peer load However the increase in thenumber of nodes leads to high system load for maintainingthe Chord overlay and limits QUVoDrsquos scalability SURFNetin [17] groups stable peers which have long online time andstore superchunk-level video content into an AVL tree Aholder-chain which is composed of peers with similar videocontent is attached to a peer in the AVL tree in terms of thesimilarity of stored content SURFNet makes use of the tree-based overlay topology to obtain nearly constant and loga-rithmic lookup time for seeking in a video stream or betweendifferent videos The maintenance cost of peers relies on thestability of the AVL tree However the long online time can-not ensure the state stability of nodes in the tree Thereforewith the increase in the number of peers the maintenanceof the tree structure also increases very much so that thesystem scalability and lookup efficiency are highly restrictedThe structured overlay such as QUVoD and SURFNet canachieve fast lookup of resources and make full use ofpeersrsquo upload bandwidth but the high maintenance costlimits the systemrsquos scale and wastes network bandwidth

The proposed mesh-based solutions with an unstruc-tured topology have high system scalability For instancethe authors of [23] proposed a mesh-based P2P stream-ing solution Each peer selects the nodes as its neighborsaccording to different predefined policies Because mutualcontact between these nodes form a random graph thesystem can perceive dynamic distribution process of videochunk and utilize created cluster of large-bandwidth peersto address intensive request of hotspot resources Chang andHuang [24] proposed a mesh-based interleaved video framedistribution scheme to support user interactivity Howeverthe resource search in the mesh-based solutions employsgossip scheme which does not support fast resource lookupThe low performance of resource lookup does not ensuresmooth playback Moreover the dissemination of gossipmessages consumes mass network bandwidth

Recently some P2P file sharing solutions based on virtualcommunities have been proposed For instance SPOON[18] groups mobile nodes into a community in terms ofcommon interest and frequent interaction between usersSPOON designs a role assignment for the community mem-bers to handle the file lookup both intracommunity andintercommunity and an file searching scheme for high-efficiency resource search in terms of user interest HoweverSPOON makes use of files stored to estimate the inter-est similarity between nodes which does not obtain highaccuracy of interest similarity The fragile community struc-ture results in increasing maintenance cost of communitymembers and a negative influence on file search efficiencyThe mobility of nodes is not mentioned in SPOON sothat the dynamic geographical distance between communitymembers brings negative influence for content delivery C5[25] groups the peers which request the same content andare near to each other into a community and collaborativelyfetches content The community members use a WLANto deliver local resources with other members Makinguse of WLAN interfaces to communicate with internalmembers can improve the delivery efficiency However the

deployment environment of C5 relies on the premise that anumber of mobile nodes have close location with each otherduring a long period and subscribe the same content so C5 isdifficult to be implemented in mobile networks The increasein the community scale introduces the highmaintenance costfor community members so the capacities of communitycoordinator become the bottleneck of system scalability

Inspired by community-based file sharing solutions thecommunity-based video streaming systems have attractedincreasing research interests from various researchers Forinstance AMCV in [26] proposed a mini-community-basedvideo sharing scheme in wireless mobile networks AMCVgroups the peers into a community in terms of the sim-ilarity of requested video chunks and uses an ant colonyoptimization-based community communication strategy thatdynamically bridges communities to support fast search forresources The interest-based peer groups can ensure highaccuracy for predicting the resource demand of users toreduce the start delay however because the broker nodesin the communities need to maintain information of com-munity members and handle the request messages frominternal or external members With increasing number ofpeers AMCVrsquos scale relies on the capacities of broker nodesnamely the broker nodes in the communities cannot bearhigh load for the management of community members so asto limit systemrsquos scalability PMCV in [27] proposed a novelperformance-aware mobile community-based video deliverysystem over vehicular ad hoc networks PMCV employs amobile community detection scheme to group peers into amobile community in terms of the similarity of playbackand movement behavior between users This scheme canobtain high stability of community structure and efficientvideo delivery Moreover PMCV makes use of a communitymember management mechanism to achieve high efficiencyof resource lookup and low maintenance cost

3 CSVD Detailed Design

31 Media Server The media server stores several videoresources to provide original video data for all mobile nodesin MANETs namely a video files set is 119878video = (V

1 V2

V119899) When a mobile node 119899

119894joins the system or a system

member requests a new video it sends a request messageto the server The server uses a system member list torecord the information of these request nodes namely119878member = ((119899

1VID 119905119901

1) (1198992VID 119905119901

2) (119899

119898VID 119905119901

119898))

whose items include the node ID requested video ID andtimestamp After the server receives the request message itselects a supplier which has the minimum value of requestedtimestamp with the requester ensuring stable logical linkbetween supplier and requester and reducing the number ofrepetitive requestmessagesMoreover in order to balance theload between suppliers and enhance the video delivery effi-ciency the server considers the serving capacities andmovingbehavior of suppliers The supplier scheduling algorithm isdetailed in Section 33 When the requester receives thereturn message containing the information of the supplier itconnects with the supplier and fetches video content If any


member leaves the system it sends the quit message to theserver After the server receives a quit message of the systemmembers it removes the member information from 119878member

With increasing number of maintained members andrequest messages the server cannot shoulder the load ofmaintaining node state and processing messages The serverrequires the clouds to maintain the state of members whichhave the video content of high-frequency access When thesystem members or mobile nodes request a popular videofile their request messages are redirected to the clouds whichare responsible for assigning the appropriate supplier for therequesters

32 Maintenance Scale Renting cloud resources to maintainand schedule the video resources stored by the nodes in P2Pnetworks brings monetary cost In order to economically useclouds the scale of maintained members should be keptwithin appropriate level in terms of the balance betweensupply and demand Due to the variation of cached videocontent the maintenance scale is dynamically regulated interms of the resource requirements change Let 119878V119894 hArr (119899

119886 119899119887

119899119905) be the set of system members which cache V

119894 where

the items in 119878V119894 are considered as candidate suppliers (CSs)The CSs receive the request message forwarded by the cloudsand deliver the video content for the requesters Let NR(119899

119896)

be the number of request messages which are received by amember 119899

119896carrying V

119894during a period time 119879

119899119896 The request

arrival rate (RAR) of 119899119896is defined as

120582119899119896=NR (119899

119896)

119879119899119896

(1)

where 120582119899119896is the number of request messages received by 119899

119896

in unit time The sum of RAR of all items in 119878V119894 is obtainedaccording to

|119878V119894 |

sum

119888=1

120582119899119888= 120582 (2)

where |119878V119894 | returns the number of items in 119878V119894 and 120582 denotesthe number of requestmessages received by the clouds in unittime andmeets the Poisson distribution [20] Because there isthe mutual independence relationship between RAR of itemsin 119878V119894 RAR meets the Poisson distribution When 119899

119896receives

a requestmessage the event it connects with the request nodeand successfully delivers to the video data is considered as 119899

119896

handling the request messageThe time of handling a requestmessage can be defined as

119879(119901)

119899119896= (1 + 120572) 119879

(119897)

119899119896+ 119879(119905)

119899119896 120572 gt 0 (3)

where 120572 is a degeneration factor of denoting decrease inthe handling capacities such as decreasing energy With thedecrease in the handling capacities of nodes the value of 119879(119901)

119899119896

increases 119879(119897)119899119896

is the time of local handling request messageand 119879(119905)

119899119896is the time of successful delivery of first video data

Let NH(119899119896) be the number of request messages which are

handled by 119899119896during a period of time 119879


processing rate of 119899119896can be obtained according to

120583119899119896=NH (119899

119896)

119879119899119896

(4)

where 120583119899119896

does not meet a specific distribution due to theunreliable link in the mobile environment and differentperformance of mobile devices Each CS handles the requestmessage according to the first-come-first-service rule Inorder to ensure high QoS of system the relationship between120582119899119896and 120583

119899119896needs to meet the following equation

min119899119896isin119878V119894

119906119899119896minus 120582119899119896 gt 0 (5)

The stay delay of request message in themessage queue ofsupplier is defined as119879(119904)

119899119896= 119879(119897)

119899119896+119879(119908)

119899119896 where119879(119908)

119899119896is the delay

of request message waiting to be processed If the requestprocessing rate is higher than the request arrival rate 119879(119908)

119899119896

is 0 s The suppliers can fast handle the request message anddeliver requested video data Otherwise if 119879(119908)

119899119896gt 0 the

suppliers need to handle other request messages so that theincrease in the startup delay of request nodes leads to thelow level of user experience In order to ensure the quality ofuser experience if119879(119908)

119899119896gt 0 the clouds need tomaintainmore

resources to meet the demand of request nodes Because therandom quit of nodes results in decreasing the number ofitems in 119878V119894 the scale of items in 119878V119894 shouldmeet the followingrule

Rule 1 If the decreased number of items in 119878V119894 is greater thanthe increment of items in 119878V119894 in unit time or the requestprocessing rate and request arrival rate of all items in 119878V119894cannot meet (5) the clouds enlarge the scale of 119878V119894

As we know once the request nodes receive the video datafrom the suppliers they cache the video content and are con-sidered as CSs When the clouds need to enlarge the scale of119878V119894 these request nodes which have reliable state shouldpreferentially be added into 119878V119894 If the clouds cannot obtainmoremembers (eg mass nodes request other video contentsor quit the system) the clouds provide the video streaming forthe request nodes

33 Supplier Scheduling Mechanism As Figure 2 shows thesupplier scheduling mechanism architecture relies on theserving capacity of CSs and the similarity of moving behaviorbetween CSs and requester to select appropriate supplier (1)Serving capacity of CSs each CS makes use of the length ofhandling queue to calculate the expectation time of handlinga request message and delivering video dataThis expectationtime is considered as the serving capacity of CS namely thelow time of handling and delivery denotes strong servingcapacity (2) Similarity of moving behavior between CSs andrequester the CSs report the information of serving timeand moving trace The cloudsserver use(s) the similarity ofmoving trace of CSs and requester as weight value of serving


Supplier scheduling mechanism

Calculates length of queue supplier and requesterCollects moving behavior of

moving behaviorCalculates similarity value ofObtains expectation time of

Calculates weighted serving capacity of candidate suppliers

serving

Figure 2 Supplier scheduling mechanism architecture

capacity of CSsThemeasurement of weighted serving capac-ity can ensure high efficiency of handling requestmessage anddelivering video data

When the cloudsserver receive(s) the request messagesof members it selects the appropriate CSs from 119878V119894 Thisis because each mobile device is limited by the energybandwidth computation and storage The cloudsserver notonly forward(s) these request messages in terms of thecapacities of mobile devices but also balance(s) the loadbetween CSs Each CS handles the request message accordingto the first-come-first-service rule namely the process ofhandling request message meets the1198721198661 queuing modelWhen the request message of a member is forwarded to 119899

119896

in terms of Pollaczek-Khintchine (P-K) formula the numberof items in the handling queue of 119899

119896can be defined as

119873119876(119899119896) = 120588V119894 +

1205882

V119894 + 1205822

V119894120590V119894

2 (1 minus 120588V119894) (6)

where 120590119888119894is the variance of time of 119899

119896handling received

request messages during 119879119899119896 120588V119894 denotes the time of 119899

119896

handling the request messages and is defined as

120588V119894 = 120582V119894119864 (119879(119901)

119899119896)V119894 (7)

where 119864(119879(119901)119899119896)V119894 is the expectation value of time of 119899

119896process-

ing all request messages during 119879119899119896 We make use of Little

formula [28] to obtain the average residence time of requestmessage in the queue of message processing of 119899

119896according

to

119882119904=119871119904

120582V119894= 119864 (119879

(119901)

119899119896)V119894+

120582V119894119864 (119879(119901)

119899119896)2

V119894+ 1205822

V119894120590V119894

2 (1 minus 120582V119894119864 (119879(119901)

119899119896)2

V119894)

(8)

Further the expectation time of serving a requester isobtained according to

119879(119904)

119899119896= 119882119904+ 119864 (119879

(119889)

119899119896)V119894 (9)

where 119879(119889)119899119896

is the time of 119899119896delivering first data of 119888

119894 119864(119879(119889)119899119896)V119894

is the expectation value of 119879(119889)119899119896

and 119879(119904)119899119896

also is considered

as the serving capacity of 119899119896 The clouds can obtain a set of

serving capacities of all items in 119878V119894 namely 119878SC = (119879(119904)

119899119886 119879(119904)

119899119887

119879(119904)

119899119905) Moreover we consider the similarity of moving

trace between supplier and requester and transform (9) to(10)

SC119899119896=

1

119879(119904)

119899119896+ cos (119908

119899119896) + 1

(10)

where119908119899119896is a similarity of moving trace of nodes Each node

119899119894periodically updates the information of one-hop neighbor

nodes and considers them as encountered nodes namely119871119890= (1198991 1198992 119899

119896) Let 119891

119896denote the number of encounter

of 119899119894and 119899119896Themean value of encounter number of all nodes

in 119871119890is defined as

119891 =sum119896

119888=1119891119888

119896 (11)

If 119891119896gt 119891 119899

119896is a frequently encountered node of 119899

119894 119899119894

extracts the information of frequently encountered nodes andconstructs a vector 119898119905

119894= (119899119886 119899119887 119899V) to denote moving

trace The similarity value of moving trace between 119899119894and 119899119896

is defined as

119908119899119896=

1003816100381610038161003816119898119905119894 cap 1198981199051198961003816100381610038161003816

max [10038161003816100381610038161198981199051198941003816100381610038161003816 1003816100381610038161003816119898119905119896

1003816100381610038161003816] (12)

The new system members which request a video contenthave not served other nodes namely 119879(119904) = 0 The itemin 119878V119894 with maxSC

119899119886 SC119899119887 SC

119899119905 is considered as a CS

which has the strongest serving capacity and is selected as thesupplier Because 119879(119904) of new systemmembers is 0 the valuesof their SCs are less The probability of new system membersbecoming suppliers is higher than other members which canbalance the load of system members

34 Model Implementation In the process of scheduling therequest the cloudsserver need(s) to balance the load ofCSs The cloudsserver remove(s) the CSs whase surplusbandwidth cannot meet the playback rate or request process-ing rate (RPR) is lower than the request arrival rate from119878V119894 If the removed CSs recover sufficient bandwidth andRPR they are added into 119878V119894 When the cloudsserverreceive(s) a request message 119903

119909 theyit select(s) a CS 119899

119896with

maxSC119899119886 SC119899119887 SC

119899119905 as the supplier The cloudsserver

make(s) the decision of adding the requester into 119878V119894 in termsof Rule 1 In the process of handling each request message119899119896records receive the message number and the time of

processing each message during a period time 119879119899119896 namely

NR(119899119896) NH(119899

119896)119879(119897)119899119896 and119879(119905)

119899119896 119899119896further obtains the values of

120582119899119896 120583119899119896 and 119879(119901)

119899119896 After 119899

119896finishes the delivery of video data

for the requester 119899119896records the total time 119879(119889)

119899119896of data trans-

mission and sends a message containing 120582119899119896 120583119899119896 119879(119901)119899119896

119879(119889)119899119896

and moving trace to the cloudsserver The cloudsserveruse(s) these parameter values to update the 119879(119904)

119899119896of 119899119896 The

cloudsserver make(s) use of moving traces of requester


(1) receives request message of requester 119899119906

(2) for (119894 = 0 119894 lt 119905 119894++)(3) if bandwidth and RPR of 119878V119895 [119894]meet the demand(4) calculates serving expectation time of 119878V119895 [119894] by (9)(5) calculates moving similarity of 119878V119895 [119894] and 119899119906 by (11)(6) obtains measurement value SC

119894of capacity of 119878V119895 [119894]

(7) adds SC119894into result set 119877

(8) end if(9) end for(10) forwards message to supplier 119899V with minimum in 119877(11) 119899V returns response message to 119899

119906

(12) 119899V sends video data to 119899119906

Algorithm 1 Search process of video file V119895

and CSs to calculate similarity of moving behavior betweenthem By investigating the similarity of moving behavior andserving capacity of CS the system can achieve fast responsefor the request and high-efficiency delivery of video dataensuring smooth playback experienceThepseudocode of theprocess of resource search is detailed in Algorithm 1 whosecomputation complexity is 119874(119899)

4 Testing and Test Results Analysis

We investigate the performance of the proposed CSVD incomparison with SPOON [18] a state-of-the-art P2P-basedfile sharing solution The number of video files is 100 andthe length of each file is 60 s CSVD was modeled andimplemented in NS-2 as described in the previous sections

41 Testing Topology and Scenarios Table 1 lists some NS-2simulation parameters of the MANET for the two solutionsWe created 100 user viewing logs namely each user randomlyaccesses 20 video files and the viewing period time is setto random value Moreover when the users have watcheda video in terms of the random playback period time theycontinue to request new video file 100 mobile nodes playvideo file following 100 logs and uniformly join the systemfollowing the Poisson distribution from 0 s to 360 s After thenodes arrive at the target location they continue to move tonew target location in new assigned speed In SPOON 100nodes randomly store 20 files with different keywords beforethey join system and the number of intersection of their localfiles is 100 The requested files corresponding to 100 logs arenot included in their local files We define some parametervalue for SPOON 120587 = 1 Ω = 1 119878max = 1 ℎ1 = 30 ℎ2 = 30and 119879

119866= 1

42 Performance Evaluation The performance of CSVD iscompared with that of SPOON in terms of average file seekdelay (AFSD) packet loss rate (PLR) throughput videoquality and overlay maintenance cost respectively

(1) AFSDThe difference value between the time when a noderequests a video file and the time when the node receives first

Table 1 Simulation parameter setting for MANET

Parameters ValuesArea 800 times 800m2

Channel ChannelWirelessChannelNetwork interface PhyWirelessPhyExtMAC interface Mac802 11Number of mobile nodes 400Number of mobile nodes playingvideo 100

Mobile speed range of nodes [0 30]msSimulation time 600 sSignal range of mobile nodes 200mDefault distance between serverand nodes 6 hops

Default distance between cloudsand nodes 6 hops

Transmission protocol UDPWireless routing protocol DSRBandwidth of server 20MbsBandwidth of mobile nodes 10MbsTransmission rate of video data 128 kbsTravel direction of mobile nodes RandomPause time of mobile nodes 0 s

video data is considered as the file seek delayThemean valueof the cumulative sum of delay values during a time intervaldenotes AFSD

As Figure 3 shows SPOONrsquos AFSD curve experiencesslow increase from 119905 = 300 s to 119905 = 540 s after fast decreasefrom 119905 = 60 s to 119905 = 240 s with the delay range between 26 sand 39 s SPOONrsquos curve finally decreases to roughly 36 sfrom 119905 = 540 s to 119905 = 600 s CSVDrsquos curve shows slow risewith slight fluctuation which slowly decreases to 14 s from119905 = 120 s to 119905 = 180 s The curve has a slow increase from119905 = 210 s to 119905 = 510 s and decreases to roughly 2 s at 119905 = 600 swhich maintains relatively low level (the values are roughly35 lower than those of SPOON)

International Journal of Distributed Sensor Networks 7Av

erag

e file

seek

ing

delay

(s)

0

1

2

3

4

5

SPOONCSVD

60 120 180 240 300 360 420 480 540 600

Simulation time (s)

Figure 3 AFSD against simulation time

Aver

age fi

le se

ekin

g de

lay (s

)

0

1

2

3

4

5

Number of nodes20 30 40 50 60 70 80 90 1000

SPOONCSVD

Figure 4 AFSD against number of nodes

Figure 4 presents the AFSD variation with increasingnumber of nodes which have joined the system The bluecurve corresponding to the SPOONrsquos results has both highervalues and larger fluctuation with the increase in the numberof nodes (the values are between 26 s and 46 s) The CSVDresults illustrated with red curve have values between 15 sand 23 s with lower variations than SPOONrsquos results

Initially the number of members in communities isrelatively less so SPOON only uses the interest-orientedrouting algorithm (IRA) to search the video files fromforeign communities The request messages are continuallyforwarded between the mobile nodes which leads to thehigh AFSD With the increase in the number of nodes thecommunity members require the community coordinatorsto search desired files If the requested files are in theintracommunity the intracommunity search speeds up theprocess of resource location so SPOONrsquos AFSD values canfast decrease and keep the relatively low level Howeverwhen the members do not fetch the requested files from theintracommunity they rely on the community ambassadorsto search the resources from foreign communitiesTherefore

05

04

03

02

01

0

Pack

et lo

ss ra

te

Number of nodes20 30 40 50 60 70 80 90 1000

SPOONCSVD

Figure 5 PLR against number of nodes

SPOONrsquos AFSD values show fast rise with increasing numberof search messages Moreover SPOON does not consider themobility of nodes so that the dynamic change of geographicallocation between the requesters and the suppliers bringsnegative influence for the delay of data delivery In CSVDthe server and clouds are responsible for handling the requestmessages and scheduling the available resources for therequesters so the fast response at the server and clouds sidereduces the delay of video file seeking Moreover CSVDinvestigates the similarity ofmoving trace between requestersand suppliers The stable mobility between requesters andsuppliers enhances the efficiency of video data delivery Onaverage CSVDrsquos results are better than those of SPOON

Packet Loss Rate (PLR) The ratio between the number ofpackets lost in the process of video data transmission and thetotal number of packets of video data sent is defined as PLR

As Figure 5 shows the curves corresponding to CSVDand SPOON show a rise trend with increasing number ofnodes The results of CSVD and SPOONmaintain low levelswhen the number of nodes increases to 60 and represent fastrise from 70 to 100 However CSVDrsquos PLR is roughly 20better than the values associated with SPOON

Figure 6 shows the variation of PLR values with theincrease in mobility speed of mobile nodes in MANETs TheCSVD results have both low values and slight increase from[0 5] to (25 30] and are between 015 and 022 The blue barscorresponding to the SPOONresultsmaintain high levels andare between 016 and 024The CSVD PLR values are roughly10 lower than the results of SPOON

Small scale system members only consume the smallnumber of network bandwidth so the PLR curves of twosystems keep slight rise With increasing number of systemmembers the members require more network bandwidth sothat the network congestion results in high PLR On the otherhand the lowmobility ofmobile nodes brings slight variationin the PLR due to the slow change in the geographicaldistance With the increase in the mobility of mobile nodesthe fast variation of geographical distance between requesters


03

025

02

01

015

005

0

SPOONCSVD

[0 5] (5 10] (15 20](10 15] (20 25] (25 30]Range variation in mobility speed (ms)

Pack

et lo

ss ra

te

Figure 6 PLR against range variation in the mobility speed ofmobile nodes

and suppliers leads to high probability of packet loss Thecommunity coordinators and ambassadors in SPOONdo notconsider the mobility of nodes in the process of the assign-ment of suppliers for the requesters The efficiency of datadelivery in SPOON is influenced by the increase in thegeographical distance between requesters and suppliersnamely the communication with long distance increases theprobabilities of wireless link break and packet loss In CSVDthe server and cloudsmatch the similarity between requestersand suppliers and assign the suppliers which have the mostsimilarmoving behavior with requesters to provide requestedvideo streaming The stable geographical distance betweenthem ensures high transmission performance such as lowdelay and reduced PLR Therefore CSVD PLR is kept at lowlevel

Average Throughput The total number of packets received inthe overlay during a certain time period divided by the lengthof this time period is defined as the average throughput

Figure 7 shows the variation of average throughput ofSPOON and CSVD with increasing simulation time Thecurves corresponding to two systems show similar risetrajectory namely they experience a fast rise from 119905 = 60 s to119905 = 180 s and a decreasing trend from 119905 = 210 s to 119905 = 600 sThe increment of CSVD results is higher than that of SPOONand the peak value of SPOONrsquos curve at 360 s is larger thanthat of CSVD

The mobile nodes request the video content following aPoisson distribution from 119905 = 0 s to 119905 = 360 s Theincrease in the number of nodes leads to numerous trans-mitted video data packets in network The two systems havefast rise trend from 119905 = 0 s to 119905 = 180 s and slowincrease from 119905 = 210 s to 119905 = 360 s and reach peakvalues at 119905 = 360 s When the data traffic of transmittingvideo content is larger than the bandwidth provided by thenetwork numerous data packets are discarded reducing thethroughput In SPOON the delivery without consideringmobility of requesters and suppliers is subjected to serious

10000

9000

8000

7000

6000

5000

4000

3000

2000

1000

0

Thro

ughp

ut (K

Bs)

SPOONCSVD

60 120 180 240 300 360 420 480 540 600

Simulation time (s)

Figure 7 Throughput against simulation time

negative influence namely the communication quality in thetransmission path decreases due to the network congestionThe values of SPOON throughput maintains low level InCSVD the requesters receive video data from the suppliersassigned by the server and clouds in terms of similarity oftheir moving behavior This ensures that the delivery perfor-mance of CSVD is positive including high throughput lowPLR and low delay Therefore the throughput values ofCSVD are larger than those of SPOON

Video Quality The peak signal-to-noise ratio (PSNR) [29] isused to denote the video quality measured in decibels (dB)and is estimated according to (13)

PSNR = 20 sdot log10(

MAX Bit

radic(EXP Thr minus CRT Thr)2) (13)

where EXP Thr is the average throughput expected from thedelivery of the video content MAX Bit is the average bitrateof the video stream as resulted from the encoding processand CRT Thr denotes the actual throughput measured dur-ing delivery MAX Bit and EXP Thr are 128 kbs in terms ofsimulation settings respectively We calculate PSNR of singlevideo streaming corresponding to every node according tothe throughput with increasing number of nodes

Figure 8 shows the average video quality of single videostreaming corresponding to each node with increasing num-ber of the nodes The results of SPOON and CSVD show thefall trendThe red bars corresponding to CSVDrsquos results havethe range [8 30] and are roughly 10 higher than SPOONThe blue bars corresponding to SPOONrsquos results have a fastdecrease from the peak value 26 dB to a minimum of 7 dB

PSNR reflects the video quality perceived by users Thedelivery of video content relies on the retransmission ofmobile nodes inMANETsThe small number of systemmem-bers do not consume toomany bandwidths of the forwardingnodes so that the PLR and delay maintain low level (PSNRalso keeps high level) With increasing number of nodes


35

30

25

20

15

10

5

0

PSN

R

CSVDSPOON

20 40 60 80 100

Number of nodes

Figure 8 PSNR against number of nodes

SPOONCSVD

Simulation time (s)60 120 180 240 300 360 420 480 540 600

1

2

3

4

5

Mai

nten

ance

ove

rhea

d (K

Bs)

Figure 9 Maintenance overhead against simulation time

the requirement of high bandwidth for the video streamingconsumes the network bandwidth and triggers the networkcongestion At themoment the high PLR and low throughputbring low PSNR of single video streaming corresponding toeach node SPOON neglects the mobility of requesters andsuppliers so that the transmission performance of video datais subjected to serious influence from network congestionIn CSVD the requesters and suppliers have similar movingbehavior by the assignment of the server and clouds Thehigh-efficiency delivery of video data can obtain respectivelylow PLR The PSNR values of CSVD are better than those ofSPOON

Maintenance OverheadThe average bandwidth which is usedby the sent messages for maintaining the overlay topology isconsidered as the maintenance overhead

As Figure 9 shows the overlay maintenance overheadvalues of two systems have similar changing trend withincreasing number of mobile nodes SPOONrsquos results fastincrease from 119905 = 240 s to 119905 = 600 s after a slow risefrom 119905 = 60 s to 119905 = 240 s The curve correspondingto CSVD maintains a slow increasing trend and has a low

increment with values roughly 30 lower than those ofSPOON

The nodes in SPOON are grouped into multiple commu-nities in terms of the interest The community coordinatorsare responsible for maintaining the state and stored resourcesof members and handling the request messages of filesAlthough SPOON employs a periodical state maintenancemechanism the increasing scale ofmaintained nodes leads tohigh load for the coordinators Mass messages of statemaintenance and files request consume a lot of energy andbandwidth of coordinators so the capacities of coordinatorsbecome the bottleneck of system scalability Moreover thefrequent exchange of membersrsquo state messages and broadcastmessages of coordinatorsrsquo replacement increase the mainte-nance cost of communities SPOONrsquos maintenance overheadvalues fast increase with increasing number of nodes UnlikeSPOON (all nodes are maintained by the coordinators) theserver and clouds in CSVD only maintain the playback stateof nodes reducing the number of exchanging messages Theclouds also dynamically regulate the number of maintainednodes in terms of the balance between supply and demandTherefore CSVDrsquos maintenance overhead for the overlaytopology keeps lower level than that of SPOON

5 Conclusion

In this paper we propose a novel Cloud-assisted Scal-able Video Delivery solution over mobile ad hoc networks(CSVD) CSVD improves the scalability of light-duty videostreaming system with the help of the clouds by maintainingthe peers which carry hotspot resources in P2P networks toensure smooth playback experience of users The estimationmodel of resource maintenance scale can regulate the utiliza-tion of cloud resources in terms of dynamic balance betweensupply and demandThe supplier scheduling mechanism canassign resource suppliers in terms of their load and servingcapacity The results show how CSVD ensures lower averagefile seek delay lower packet loss rate higher throughputhigher video quality and lower overlaymaintenance cost thanSPOON

Conflict of Interests

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

Acknowledgments

This work was supported in part by the National Natural Sci-ence Foundation ofChina (NSFC) underGrant nos 61402303and 61303053 in part by the Project of Beijing MunicipalCommission of Education under Grant KM201510028016and in part by the Beijing Natural Science Foundation underGrant 4142037


References

[1] M Qi and J Hong ldquoAAPP an anycast based AODV routingprotocol for peer to peer services in MANETrdquo InternationalJournal of Distributed Sensor Networks vol 5 no 1 p 48 2009

[2] S Jia C Xu G-M Muntean J Guan and H Zhang ldquoCross-layer and one-hop neighbour-assisted video sharing solution inmobile Ad Hoc networksrdquo China Communications vol 10 no6 pp 111ndash126 2013

[3] Q Guan F R Yu S Jiang V C M Leung and H MehrvarldquoTopology control inmobile AdHoc networks with cooperativecommunicationsrdquo IEEEWireless Communications vol 19 no 2pp 74ndash79 2012

[4] S Jin and S Choi ldquoA seamless handoff with multiple radios inIEEE 80211 WLANsrdquo IEEE Transactions on Vehicular Technol-ogy vol 63 no 3 pp 1408ndash1418 2014

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

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

[7] C Xu Z Li J Li H Zhang and G-M Muntean ldquoCross-layer fairness-driven concurrent multipath video delivery overheterogenous wireless networksrdquo IEEE Transactions on Circuitsand Systems for Video Technology no 99 2014

[8] W Zhang Z Li and Q Zheng ldquoSAMP supporting multi-source heterogeneity in mobile P2P IPTV systemrdquo IEEE Trans-actions on Consumer Electronics vol 59 no 4 pp 772ndash7782013

[9] M A Hoque M Siekkinen and J K Nurminen ldquoEnergyefficient multimedia streaming to mobile devices-a surveyrdquoIEEE Communications Surveys and Tutorials vol 16 no 1 pp579ndash597 2014

[10] S Jia C Xu A V Vasilakos J Guan H Zhang and G-M Muntean ldquoReliability-oriented ant colony optimization-based mobile peer-to-peer VoD solution in MANETsrdquoWirelessNetworks vol 20 no 5 pp 1185ndash1202 2014

[11] Y Chen B Zhang C Chen and D M Chiu ldquoPerformancemodeling and evaluation of peer-to-peer live streaming systemsunder flash crowdsrdquo IEEEACM Transactions on Networkingvol 22 no 4 pp 2428ndash2440 2014

[12] L Zhou Y Zhang K Song W Jing and A V VasilakosldquoDistributed media services in P2P-based vehicular networksrdquoIEEE Transactions on Vehicular Technology vol 60 no 2 pp692ndash703 2011

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

[14] Y Sun Y Guo Z Li et al ldquoThe case for P2P mobile videosystem over wireless broadband networks a practical study ofchallenges for a mobile video providerrdquo IEEE Network vol 27no 2 pp 22ndash27 2013

[15] H Shen Z Li Y Lin and J Li ldquoSocialTube P2P-assisted videosharing in online social networksrdquo IEEETransactions on Paralleland Distributed Systems vol 25 no 9 pp 2428ndash2440 2014

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

[17] D Wang and C K Yeo ldquoSuperchunk-based efficient search inP2P-VoD systemrdquo IEEE Transactions onMultimedia vol 13 no2 pp 376ndash387 2011

[18] K Chen H Shen and H Zhang ldquoLeveraging social networksfor P2P content-based file sharing in disconnected MANETsrdquoIEEE Transactions on Mobile Computing vol 13 no 2 pp 235ndash249 2014

[19] F Liu S Shen B Li and H Jin ldquoCinematic-quality VoD in ap2p storage cloud design implementation andmeasurementsrdquoIEEE Journal on Selected Areas in Communications vol 31 no9 pp 214ndash226 2013

[20] Y Wu C Wu B Li X Qiu and F C M Lau ldquoCloudMediawhen cloud on demandmeets video on demandrdquo in Proceedingsof the 31st International Conference on Distributed ComputingSystems (ICDCS rsquo11) pp 268ndash277 July 2011

[21] L Zhou Z Yang J J P C Rodrigues and M GuizanildquoExploring blind online scheduling for mobile cloud multime-dia servicesrdquo IEEE Wireless Communications vol 20 no 3 pp54ndash61 2013

[22] S Wang and S Dey ldquoAdaptive mobile cloud computing toenable richmobile multimedia applicationsrdquo IEEE Transactionson Multimedia vol 15 no 4 pp 870ndash883 2013

[23] A P C Da Silva E Leonardi M Mellia and M Meo ldquoChunkdistribution in mesh-based large-scale P2P streaming systemsa fluid approachrdquo IEEE Transactions on Parallel and DistributedSystems vol 22 no 3 pp 451ndash463 2011

[24] C-L Chang and S-P Huang ldquoThe interleaved video frame dis-tribution for P2P-based VoD system with VCR functionalityrdquoComputer Networks vol 56 no 5 pp 1525ndash1537 2012

[25] L Tu and C-M Huang ldquoCollaborative content fetching usingMAC layer multicast in wireless mobile networksrdquo IEEE Trans-actions on Broadcasting vol 57 no 3 pp 695ndash706 2011

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

[27] C Xu S Jia M Wang L Zhong H Zhang and G-MMuntean ldquoPerformance-aware mobile community-based VoDstreaming over vehicular Ad Hoc networksrdquo IEEE Transactionson Vehicular Technology 2014

[28] F J Beutler ldquoMean sojourn times in Markov queueing net-works littlersquos formula revisitedrdquo IEEE Transactions on Informa-tion Theory vol 29 no 2 pp 233ndash241 1983

[29] S-B Lee G-M Muntean and A F Smeaton ldquoPerformance-aware replication of distributed pre-recorded IPTV contentrdquoIEEE Transactions on Broadcasting vol 55 no 2 pp 516ndash5262009

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



Internet

Media server

Cloud

Figure 1 CSVD architecture

cluster increases the cost of system deployment SPOONin [18] is a community-based file sharing solution overMANETs SPOON groups the peers into multiple commu-nities in terms of the interest similarity which can achievehigh efficiency of content sharing However the stability ofcommunity structure relies on the capacities of communitycoordinator and determines system performance and main-tenance cost of communities Making use of the server tocompensate insufficient bandwidth further limits the systemscale

Recently cloud computing has become the most popu-lar computing paradigm [19] Providing on-demand serverresources to users relies on the shared pool of servers indatacenters [20] With rapid growth of mobile devices usagemobile cloud streaming becomes a promising avenue for sup-porting high-performance and interactive streaming service[21 22]Themost of cloud-based multimedia systems use theclouds to compensate insufficient capacities of upload band-width computing and storage in the systems However therandom request behavior of users for the video content leadsto the increase in the complexity of cloud usage and the costof interactivity between server and clouds Moreover makinguse of the clouds to meet the requirement of bandwidthbrings expensive monetary cost

In this paper we propose a novel Cloud-assisted ScalableVideo Delivery solution over mobile ad hoc networks (CSVD)

As Figure 1 shows the clouds assist the media server tomanage the hotspot resources to address the large-scaleintensive request when the server cannotmeet the demand ofusers for video resources A novel estimation model ofresource maintenance scale based on QoS-oriented dynamicbalance between supply and demand is proposed reducingmonetary cost from rental resources in the clouds A novelsupplier scheduling mechanism that schedules suppliers interms of their load and estimated processing capacity is pro-posed reducing the lookup delay and balancing the load ofsuppliers Simulation results show how CSVD achieves muchbetter performance results in comparisonwith other state-of-the-art solutions

2 Related Work

There have been numerous studies on P2P-based man-agement and lookup optimization of resources for videostreaming services in recent years

The solutions based on structured content distributiontopology are well known for resource lookup efficiency Forinstance QUVoD in [16] employs a group-based Chordstructure to uniformly distribute the video content where thepeers which store similar content in the Chord overlay form agroup By using this structure the request for sequential video









1 V2




1VID 119905119901

1) (1198992VID 119905119901

2) (119899

119898VID 119905119901

119898))






119886 119899119887


119894 where


119896)


119896carrying V




120582119899119896=NR (119899

119896)

119879119899119896

(1)


119896


|119878V119894 |

sum

119888=1

120582119899119888= 120582 (2)


119896receives


119896


119879(119901)

119899119896= (1 + 120572) 119879

(119897)

119899119896+ 119879(119905)

119899119896 120572 gt 0 (3)


119899119896

increases 119879(119897)119899119896







120583119899119896=NH (119899

119896)

119879119899119896

(4)

where 120583119899119896



min119899119896isin119878V119894

119906119899119896minus 120582119899119896 gt 0 (5)


119899119896= 119879(119897)

119899119896+119879(119908)

119899119896 where119879(119908)



119899119896


119899119896gt 0 the












serving




119896



119873119876(119899119896) = 120588V119894 +

1205882

V119894 + 1205822

V119894120590V119894

2 (1 minus 120588V119894) (6)




119896


120588V119894 = 120582V119894119864 (119879(119901)

119899119896)V119894 (7)


119896process-



119896according

to

119882119904=119871119904

120582V119894= 119864 (119879

(119901)

119899119896)V119894+

120582V119894119864 (119879(119901)

119899119896)2

V119894+ 1205822

V119894120590V119894

2 (1 minus 120582V119894119864 (119879(119901)

119899119896)2

V119894)

(8)


119879(119904)

119899119896= 119882119904+ 119864 (119879

(119889)

119899119896)V119894 (9)

where 119879(119889)119899119896


119894 119864(119879(119889)119899119896)V119894


and 119879(119904)119899119896

also is considered



119899119886 119879(119904)

119899119887

119879(119904)



SC119899119896=

1

119879(119904)

119899119896+ cos (119908

119899119896) + 1

(10)




119896) Let 119891




119891 =sum119896

119888=1119891119888

119896 (11)

If 119891119896gt 119891 119899


119894 119899119894




is defined as

119908119899119896=

1003816100381610038161003816119898119905119894 cap 1198981199051198961003816100381610038161003816

max [10038161003816100381610038161198981199051198941003816100381610038161003816 1003816100381610038161003816119898119905119896

1003816100381610038161003816] (12)


119899119886 SC119899119887 SC





119896with

maxSC119899119886 SC119899119887 SC




NR(119899119896) NH(119899

119896)119879(119897)119899119896 and119879(119905)


120582119899119896 120583119899119896 and 119879(119901)

119899119896 After 119899





119879(119889)119899119896


119899119896of 119899119896 The





119894of capacity of 119878V119895 [119894]



119906







119866= 1












erag

e file

seek

ing

delay

(s)

0

1

2

3

4

5

SPOONCSVD

60 120 180 240 300 360 420 480 540 600

Simulation time (s)


Aver

age fi

le se

ekin

g de

lay (s

)

0

1

2

3

4

5

Number of nodes20 30 40 50 60 70 80 90 1000

SPOONCSVD




05

04

03

02

01

0

Pack

et lo

ss ra

te

Number of nodes20 30 40 50 60 70 80 90 1000

SPOONCSVD








03

025

02

01

015

005

0

SPOONCSVD


Pack

et lo

ss ra

te






10000

9000

8000

7000

6000

5000

4000

3000

2000

1000

0

Thro

ughp

ut (K

Bs)

SPOONCSVD

60 120 180 240 300 360 420 480 540 600

Simulation time (s)





MAX Bit






35

30

25

20

15

10

5

0

PSN

R

CSVDSPOON

20 40 60 80 100

Number of nodes


SPOONCSVD


1

2

3

4

5

Mai

nten

ance

ove

rhea

d (K

Bs)







5 Conclusion




Acknowledgments



References
































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation

















1 V2




1VID 119905119901

1) (1198992VID 119905119901

2) (119899

119898VID 119905119901

119898))






119886 119899119887


119894 where


119896)


119896carrying V




120582119899119896=NR (119899

119896)

119879119899119896

(1)


119896


|119878V119894 |

sum

119888=1

120582119899119888= 120582 (2)


119896receives


119896


119879(119901)

119899119896= (1 + 120572) 119879

(119897)

119899119896+ 119879(119905)

119899119896 120572 gt 0 (3)


119899119896

increases 119879(119897)119899119896







120583119899119896=NH (119899

119896)

119879119899119896

(4)

where 120583119899119896



min119899119896isin119878V119894

119906119899119896minus 120582119899119896 gt 0 (5)


119899119896= 119879(119897)

119899119896+119879(119908)

119899119896 where119879(119908)



119899119896


119899119896gt 0 the












serving




119896



119873119876(119899119896) = 120588V119894 +

1205882

V119894 + 1205822

V119894120590V119894

2 (1 minus 120588V119894) (6)




119896


120588V119894 = 120582V119894119864 (119879(119901)

119899119896)V119894 (7)


119896process-



119896according

to

119882119904=119871119904

120582V119894= 119864 (119879

(119901)

119899119896)V119894+

120582V119894119864 (119879(119901)

119899119896)2

V119894+ 1205822

V119894120590V119894

2 (1 minus 120582V119894119864 (119879(119901)

119899119896)2

V119894)

(8)


119879(119904)

119899119896= 119882119904+ 119864 (119879

(119889)

119899119896)V119894 (9)

where 119879(119889)119899119896


119894 119864(119879(119889)119899119896)V119894


and 119879(119904)119899119896

also is considered



119899119886 119879(119904)

119899119887

119879(119904)



SC119899119896=

1

119879(119904)

119899119896+ cos (119908

119899119896) + 1

(10)




119896) Let 119891




119891 =sum119896

119888=1119891119888

119896 (11)

If 119891119896gt 119891 119899


119894 119899119894




is defined as

119908119899119896=

1003816100381610038161003816119898119905119894 cap 1198981199051198961003816100381610038161003816

max [10038161003816100381610038161198981199051198941003816100381610038161003816 1003816100381610038161003816119898119905119896

1003816100381610038161003816] (12)


119899119886 SC119899119887 SC





119896with

maxSC119899119886 SC119899119887 SC




NR(119899119896) NH(119899

119896)119879(119897)119899119896 and119879(119905)


120582119899119896 120583119899119896 and 119879(119901)

119899119896 After 119899





119879(119889)119899119896


119899119896of 119899119896 The





119894of capacity of 119878V119895 [119894]



119906







119866= 1












erag

e file

seek

ing

delay

(s)

0

1

2

3

4

5

SPOONCSVD

60 120 180 240 300 360 420 480 540 600

Simulation time (s)


Aver

age fi

le se

ekin

g de

lay (s

)

0

1

2

3

4

5

Number of nodes20 30 40 50 60 70 80 90 1000

SPOONCSVD




05

04

03

02

01

0

Pack

et lo

ss ra

te

Number of nodes20 30 40 50 60 70 80 90 1000

SPOONCSVD








03

025

02

01

015

005

0

SPOONCSVD


Pack

et lo

ss ra

te






10000

9000

8000

7000

6000

5000

4000

3000

2000

1000

0

Thro

ughp

ut (K

Bs)

SPOONCSVD

60 120 180 240 300 360 420 480 540 600

Simulation time (s)





MAX Bit






35

30

25

20

15

10

5

0

PSN

R

CSVDSPOON

20 40 60 80 100

Number of nodes


SPOONCSVD


1

2

3

4

5

Mai

nten

ance

ove

rhea

d (K

Bs)







5 Conclusion




Acknowledgments



References
































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation













119886 119899119887


119894 where


119896)


119896carrying V




120582119899119896=NR (119899

119896)

119879119899119896

(1)


119896


|119878V119894 |

sum

119888=1

120582119899119888= 120582 (2)


119896receives


119896


119879(119901)

119899119896= (1 + 120572) 119879

(119897)

119899119896+ 119879(119905)

119899119896 120572 gt 0 (3)


119899119896

increases 119879(119897)119899119896







120583119899119896=NH (119899

119896)

119879119899119896

(4)

where 120583119899119896



min119899119896isin119878V119894

119906119899119896minus 120582119899119896 gt 0 (5)


119899119896= 119879(119897)

119899119896+119879(119908)

119899119896 where119879(119908)



119899119896


119899119896gt 0 the












serving




119896



119873119876(119899119896) = 120588V119894 +

1205882

V119894 + 1205822

V119894120590V119894

2 (1 minus 120588V119894) (6)




119896


120588V119894 = 120582V119894119864 (119879(119901)

119899119896)V119894 (7)


119896process-



119896according

to

119882119904=119871119904

120582V119894= 119864 (119879

(119901)

119899119896)V119894+

120582V119894119864 (119879(119901)

119899119896)2

V119894+ 1205822

V119894120590V119894

2 (1 minus 120582V119894119864 (119879(119901)

119899119896)2

V119894)

(8)


119879(119904)

119899119896= 119882119904+ 119864 (119879

(119889)

119899119896)V119894 (9)

where 119879(119889)119899119896


119894 119864(119879(119889)119899119896)V119894


and 119879(119904)119899119896

also is considered



119899119886 119879(119904)

119899119887

119879(119904)



SC119899119896=

1

119879(119904)

119899119896+ cos (119908

119899119896) + 1

(10)




119896) Let 119891




119891 =sum119896

119888=1119891119888

119896 (11)

If 119891119896gt 119891 119899


119894 119899119894




is defined as

119908119899119896=

1003816100381610038161003816119898119905119894 cap 1198981199051198961003816100381610038161003816

max [10038161003816100381610038161198981199051198941003816100381610038161003816 1003816100381610038161003816119898119905119896

1003816100381610038161003816] (12)


119899119886 SC119899119887 SC





119896with

maxSC119899119886 SC119899119887 SC




NR(119899119896) NH(119899

119896)119879(119897)119899119896 and119879(119905)


120582119899119896 120583119899119896 and 119879(119901)

119899119896 After 119899





119879(119889)119899119896


119899119896of 119899119896 The





119894of capacity of 119878V119895 [119894]



119906







119866= 1












erag

e file

seek

ing

delay

(s)

0

1

2

3

4

5

SPOONCSVD

60 120 180 240 300 360 420 480 540 600

Simulation time (s)


Aver

age fi

le se

ekin

g de

lay (s

)

0

1

2

3

4

5

Number of nodes20 30 40 50 60 70 80 90 1000

SPOONCSVD




05

04

03

02

01

0

Pack

et lo

ss ra

te

Number of nodes20 30 40 50 60 70 80 90 1000

SPOONCSVD








03

025

02

01

015

005

0

SPOONCSVD


Pack

et lo

ss ra

te






10000

9000

8000

7000

6000

5000

4000

3000

2000

1000

0

Thro

ughp

ut (K

Bs)

SPOONCSVD

60 120 180 240 300 360 420 480 540 600

Simulation time (s)





MAX Bit






35

30

25

20

15

10

5

0

PSN

R

CSVDSPOON

20 40 60 80 100

Number of nodes


SPOONCSVD


1

2

3

4

5

Mai

nten

ance

ove

rhea

d (K

Bs)







5 Conclusion




Acknowledgments



References
































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation














serving




119896



119873119876(119899119896) = 120588V119894 +

1205882

V119894 + 1205822

V119894120590V119894

2 (1 minus 120588V119894) (6)




119896


120588V119894 = 120582V119894119864 (119879(119901)

119899119896)V119894 (7)


119896process-



119896according

to

119882119904=119871119904

120582V119894= 119864 (119879

(119901)

119899119896)V119894+

120582V119894119864 (119879(119901)

119899119896)2

V119894+ 1205822

V119894120590V119894

2 (1 minus 120582V119894119864 (119879(119901)

119899119896)2

V119894)

(8)


119879(119904)

119899119896= 119882119904+ 119864 (119879

(119889)

119899119896)V119894 (9)

where 119879(119889)119899119896


119894 119864(119879(119889)119899119896)V119894


and 119879(119904)119899119896

also is considered



119899119886 119879(119904)

119899119887

119879(119904)



SC119899119896=

1

119879(119904)

119899119896+ cos (119908

119899119896) + 1

(10)




119896) Let 119891




119891 =sum119896

119888=1119891119888

119896 (11)

If 119891119896gt 119891 119899


119894 119899119894




is defined as

119908119899119896=

1003816100381610038161003816119898119905119894 cap 1198981199051198961003816100381610038161003816

max [10038161003816100381610038161198981199051198941003816100381610038161003816 1003816100381610038161003816119898119905119896

1003816100381610038161003816] (12)


119899119886 SC119899119887 SC





119896with

maxSC119899119886 SC119899119887 SC




NR(119899119896) NH(119899

119896)119879(119897)119899119896 and119879(119905)


120582119899119896 120583119899119896 and 119879(119901)

119899119896 After 119899





119879(119889)119899119896


119899119896of 119899119896 The





119894of capacity of 119878V119895 [119894]



119906







119866= 1












erag

e file

seek

ing

delay

(s)

0

1

2

3

4

5

SPOONCSVD

60 120 180 240 300 360 420 480 540 600

Simulation time (s)


Aver

age fi

le se

ekin

g de

lay (s

)

0

1

2

3

4

5

Number of nodes20 30 40 50 60 70 80 90 1000

SPOONCSVD




05

04

03

02

01

0

Pack

et lo

ss ra

te

Number of nodes20 30 40 50 60 70 80 90 1000

SPOONCSVD








03

025

02

01

015

005

0

SPOONCSVD


Pack

et lo

ss ra

te






10000

9000

8000

7000

6000

5000

4000

3000

2000

1000

0

Thro

ughp

ut (K

Bs)

SPOONCSVD

60 120 180 240 300 360 420 480 540 600

Simulation time (s)





MAX Bit






35

30

25

20

15

10

5

0

PSN

R

CSVDSPOON

20 40 60 80 100

Number of nodes


SPOONCSVD


1

2

3

4

5

Mai

nten

ance

ove

rhea

d (K

Bs)







5 Conclusion




Acknowledgments



References
































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation












119894of capacity of 119878V119895 [119894]



119906







119866= 1












erag

e file

seek

ing

delay

(s)

0

1

2

3

4

5

SPOONCSVD

60 120 180 240 300 360 420 480 540 600

Simulation time (s)


Aver

age fi

le se

ekin

g de

lay (s

)

0

1

2

3

4

5

Number of nodes20 30 40 50 60 70 80 90 1000

SPOONCSVD




05

04

03

02

01

0

Pack

et lo

ss ra

te

Number of nodes20 30 40 50 60 70 80 90 1000

SPOONCSVD








03

025

02

01

015

005

0

SPOONCSVD


Pack

et lo

ss ra

te






10000

9000

8000

7000

6000

5000

4000

3000

2000

1000

0

Thro

ughp

ut (K

Bs)

SPOONCSVD

60 120 180 240 300 360 420 480 540 600

Simulation time (s)





MAX Bit






35

30

25

20

15

10

5

0

PSN

R

CSVDSPOON

20 40 60 80 100

Number of nodes


SPOONCSVD


1

2

3

4

5

Mai

nten

ance

ove

rhea

d (K

Bs)







5 Conclusion




Acknowledgments



References
































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










erag

e file

seek

ing

delay

(s)

0

1

2

3

4

5

SPOONCSVD

60 120 180 240 300 360 420 480 540 600

Simulation time (s)


Aver

age fi

le se

ekin

g de

lay (s

)

0

1

2

3

4

5

Number of nodes20 30 40 50 60 70 80 90 1000

SPOONCSVD




05

04

03

02

01

0

Pack

et lo

ss ra

te

Number of nodes20 30 40 50 60 70 80 90 1000

SPOONCSVD








03

025

02

01

015

005

0

SPOONCSVD


Pack

et lo

ss ra

te






10000

9000

8000

7000

6000

5000

4000

3000

2000

1000

0

Thro

ughp

ut (K

Bs)

SPOONCSVD

60 120 180 240 300 360 420 480 540 600

Simulation time (s)





MAX Bit






35

30

25

20

15

10

5

0

PSN

R

CSVDSPOON

20 40 60 80 100

Number of nodes


SPOONCSVD


1

2

3

4

5

Mai

nten

ance

ove

rhea

d (K

Bs)







5 Conclusion




Acknowledgments



References
































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










03

025

02

01

015

005

0

SPOONCSVD


Pack

et lo

ss ra

te






10000

9000

8000

7000

6000

5000

4000

3000

2000

1000

0

Thro

ughp

ut (K

Bs)

SPOONCSVD

60 120 180 240 300 360 420 480 540 600

Simulation time (s)





MAX Bit






35

30

25

20

15

10

5

0

PSN

R

CSVDSPOON

20 40 60 80 100

Number of nodes


SPOONCSVD


1

2

3

4

5

Mai

nten

ance

ove

rhea

d (K

Bs)







5 Conclusion




Acknowledgments



References
































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










35

30

25

20

15

10

5

0

PSN

R

CSVDSPOON

20 40 60 80 100

Number of nodes


SPOONCSVD


1

2

3

4

5

Mai

nten

ance

ove

rhea

d (K

Bs)







5 Conclusion




Acknowledgments



References
































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










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









Documents

Research Article Cloud-Assisted Scalable Video Delivery ...downloads.hindawi.com/journals/ijdsn/2015/205106.pdfcommunity-based video streaming systems have attracted increasing research