Interference-aware Multipath Video Streaming for Quality Delivery in Vehicular Environments

Transcription

1 Page of Iterferece-aware Multipath Video Streamig for Quality Delivery i Vehicular Eviromets Ahmed Aliyu, Abdul Haa Abdullah, Member IEEE, Nauma Aslam, Member IEEE, Raja Zahilah Raja Mohd Radzi, Ayma Altameem, Rupak Kharel, Member IEEE, Usma Mohammed Joda Abstract The multipath trasmissio has bee regarded as the most suitable trasmissio cocept for high data rate packets such as video data. The video packets are split ito differet frames for trasmissio via differet paths. The trasmissio is carried out cocurretly, thus oe path may iterfere with aother path. Cosiderig the multipath video streamig, the issue of route couplig effect has bee oe of the major challeges of multipath trasmissio, especially i a vehicular etwork. The route couplig causes wireless cotetio ad video packet collisio. I this regard, this paper proposes a multipath video trasmissio protocol that cosiders path s route couplig effect i order to miimize iterferece betwee multiple paths. The route couplig miimizatio strategy is based o selectig two dispersed vehicles based o a agle by the source vehicle withi it etwork coverage. The lik ad ode disjoit cocepts are icorporated ito the dispersed vehicle selectio. I order to further ascertai vehicles with zero or low iterferece, the lik sigal power ad badwidth capability have bee employed. Cocerig the dispersed vehicle selectio ad lik sigal power, mathematical derivatios are preseted with their umerical aalysis. I additio, performace evaluatio of the proposed scheme has bee performed based o the simulatio carried out. The various results obtaied from the simulatio are bechmarked with the baselie research works. The results demostrate that the proposed Iterferece-aware Multipath Solutio with Lik ad Node Disjoit (IMSLND) protocol for video streamig offers higher video quality whe compared to the baselie research works. Idex Terms Video streamig, multipath, vehicular etwork, iterferece, multimedia, Route couplig, VANETs. R I. INTRODUCTION ECENTLY the advacemet i vehicular commuicatio improves o o-road safety ad ifotaimet services. The Itelliget Trasportatio Systems (ITS) are desiged systems that miimize o-road accidet ad to ehaces mechaisms for emergecy respose. This had led to several cotributios by both idustry ad researchers to improve o protocols ad mechaism that ehace o-road safety ad ifotaimet services. The research ad developmet is supported by the KTP grat at the School of Egieerig, Machester Metropolita Uiversity, UK. A. Aliyu, AH. Abdullah, RZRM Radzi ad UM. Joda, are with the Uiversiti Tekologi Malaysia, Johor Bahru, Malaysia. aahmed@live.utm.my; haa@utm.my; zahilah@utm.my umjoda@gmail.com. N. Aslam is the Northumbria Uiversity, Newcastle upo Tye, NE ST, UK auma.aslam@orthumbria.ac.uk A. Altameem is with the Kig Saud Uiversity, Saudi Arabia. aaltameem@ksu.edu.sa R. Kharel is the Machester Metropolita Uiversity, M BH, UK r.kharel@mmu.ac.uk I most of the recet cotributios, text a message-based ad beaco sigal-based data are the most cosidered. However, the ature of the data does ot provide a more realistic iformatio o o-road accidet ad ifotaimet []. Therefore, streamig of video for o-road safety ad ifotaimet has bee cosidered i some research work [-]. Video data provides iformatio that is more appealig, comprehesive, iteractive ad uderstadig to vehicle users []. The O-Board-Uit icludig Dedicate Short Rage Commuicatios (DSRC) device ad RoadSide Uits (RSUs) supports video streamig amog vehicles. The streamig video ca be related to a pedestria crossig the road or accidet occurrece ahead o the road. For the ifotaimet, advertisemet of o-road grocery shops ad gas statios ca be displayed o OBU of users vehicle. The video streamig i vehicular commuicatio could be Vehicle-to-Vehicle (VV) or Vehicleto-Ifrastructure (VI). VV is the commuicatio betwee vehicles, which is facilitated by the OBU. Meawhile, VI is the commuicatio betwee the vehicle ad o-road device aided by RSUs. Thus, the video streamig is a importat aspect of vehicular commuicatio which improves users oboard experiece. Video streamig i vehicular eviromet ecouters several challeges due to the high data rate of video packets, the dyamic topology of VANETs ad costraied resources. The challeges icrease whe tryig to achieve high-quality video streamig due to a large amout of video data. cosiderig the aforemetioed challeges, we propose a protocol based o geographical routig that is capable of achievig high-quality video streamig i a vehicular etwork. Protocols icludig Forward Error Correctio (FEC) ad multipath solutios have bee employed. Both the FEC ad multipath solutios are ofte cross-layer based approach. May research works that are based o FEC techiques geerates duplicate packets durig trasmissio, this lead to redudat packets ad large badwidth cosumptio [-]. I recet research work, which is based o multipath video streamig [, -]. It is a approach, which is based o partitioig video frames i order to trasmit it through multiple paths (See Fig. ). This approach miimizes the high data rate issues i video trasmissio [0]. Eve though, i the multiple paths formatio, the sigal coverage of the odes i differet paths are ot cosidered. Hece, this may lead to cotetio, collisio, ad cogestio of video packets, which i tur causes video packet loss. The loss of the video packets affects the quality of the video streamig. Therefore, i order to have a quality video streamig, the sigal coverage of odes i the multipath ad most suitable routig

2 Page of protocol must be take ito cosideratio durig paths formatio. Thus, the iterferece i multiple paths will be avoided or miimized. Video source ode Iterferece Coverage Trasmissio Coverage Path Path Path Path Fig.. The Multipath Video Streamig Sceario Video destiatio ode Further, oe of the most suitable routig protocol for vehicular commuicatio is geographic-based routig protocol [, ]. It is based o the exploitatio of a geographical positio of a vehicle for makig routig decisio [, ]. The routig decisios are ofte based o parameters such as directio speed ad, directio ad/or static forwardig regio [-]. Several research studies have focused o directio ad distace such as Mobility-Aware which is a improvemet o Greedy Forwardig protocol (MAGF) [], forwardig decisio based o Directioal Greedy Routig (DGR) [0] ad data forwardig based o Greedy Stateless Perimeter Routig cosiderig Motio Vector (GSPR-MV) []. Some techiques which are based o the static geographic regio have also bee suggested icludig Segmet of vehicle ode, quality of Lik ad Degree of coectivity based Geographic DIstace Routig (SLDGDIR) [] ad Vorooi Diagram-based Geographic Distace Routig (V-GEDIR) []. Therefore, i our proposed protocol a fixed forwardig regio with greedy forwardig approach is cosidered based o multipath video trasmissio. Therefore, the article proposes a multipath video trasmissio protocol that cosiders path s route couplig effect i order to miimize iterferece betwee multiple paths. Precisely, the cotributios i this article are highlighted as follows. ) The route couplig miimizatio strategy desiged ad developed based o selectio of two dispersed vehicles cosiderig agle of separatio by the source vehicle withi it etwork coverage. The lik ad ode disjoit cocept is also icorporated for the dispersed vehicle selectio. ) I order to further ascertai vehicles with zero or miimal iterferece, the lik sigal power ad badwidth capability have bee employed. ) Cocerig the dispersed vehicle selectio ad lik sigal power, mathematical derivatios are preseted with their umerical aalysis. ) I additio, performace evaluatio of the proposed scheme has bee performed based o the simulatio carried out. The various results obtaied from the simulatio are bechmarked with the baselie research works. Cosequetly, related questios that are aimed to be aswered i this paper are as follows. ) How to select two dispersed vehicles withi the etwork coverage of a source vehicle that eables multiple trasmissios with zero or miimal iterferece, which is due to route couplig? ) How to form a multipath trasmissio that is lik ad ode disjoited i order to miimize route couplig? ) How to estimate the lik sigal power ad badwidth capability of vehicle s lik i order to determie the quality of the lik for video streamig i multipath vehicular etwork? ) How to mathematically formulate the dispersio of the vehicles, the lik sigal power ad badwidth capability of the ext forwardig vehicle s lik? The remaiig parts of the paper are structured as follows. I sectio, we preset a comprehesive review of related literature. Sectio, suggests our proposed mathematical model ad algorithms. Sectio, presets the simulatio results ad their aalysis, ad fially, sectio cocludes the paper. II. RELATED WORK I this sectio, a qualitative review of video streamig i vehicular eviromets is preseted focusig o MAC ad codig orieted video streamig. Sectio. ad. discusses the MAC orieted ad codig orieted video streamig respectively. A. MAC Orieted Video Streamig These approaches are based o lik layer techology, the use of MAC layer i etwork optimizatio offers sigificat beefits. The layer is usually adjusted to modify frame sizes cosiderig physical rules i order to attai a optimal balace betwee the higher delay of smaller frames ad the potetial distortio of losig larger video frames []. At the MAC layer, parameters are maipulated icludig retrasmissio i order to achieve robust ad qualitative video trasmissio []. The FEC approach performs recovery ad correctio of the loss ad damaged video packets durig video trasmissio. However, FEC adds some redudat video packets i order to compesate the loss video packet, this leads to icrease i badwidth utilizatio of the etwork, hece creatig aother challege. Asefi, et al. [] suggested a adaptatio scheme that employs multi-objective optimizatio structure, which cocurretly reduces the likelihood of start-up delay ad playback freeze of a streamig video at the destiatio vehicle. The start-up delay ad playback freeze are reduced by turig the MAC retrasmissio limit i relatio to chael delay packet trasmissio rate. However, delay due to packet loss has ot bee adequately cosidered. WAVE-based Hybrid Coordiatio Fuctio (W-HCF) employs the cotrolled access capabilities istead of the basic cotetio-based access of the IEEE 0.p. Also, it uses vehicle locatio iformatio ad

3 Page of coordiatio amog WAVE providers i order to ehace the performace of delay costraied ad loss-aware ifotaimet applicatios []. Further, a selective Rebroadcast mechaism for Video streamig over VANETs (ReViV) is proposed to relieve overloaded chaels ad assist i deliverig video cotet i sparse etwork settigs []. The mechaism chooses a fewer subset of rebroadcastig vehicles so as to reduce iterferece ad attai higher video quality. Error recovery video streamig protocol that uses multi-chael to address packet is suggested []. The multi-chael is categorized ito the reliable ad oreliable chael. However, chael cotetio has ot bee cosidered. Bucciol, et al. [], suggested a solutio, which is based o FEC ad Iterleavig Real-time Optimizatio (FIRO) approach to improve video streamig quality. However, i the MAC ad FEC approaches, the challeges of the high data rate of the video data have ot bee adequately cosidered. Further, the issue of iterferece i MAC layer based o route couplig effect i multipath trasmissio has ot bee cosidered i previous work. The overlay approach for video streamig is based o creatig a replicate of the real etwork for faster video packet forwardig from source to destiatio. The forwardig vehicles are cosidered as the relay vehicles. The relay vehicles are selected alog the path of the destiatio vehicle. I Hsieh ad Wag [], a robust ad dyamic overlay multicast for multimedia streamig i vehicular commuicatio has bee suggested. The idea is based o hadlig o-grouped ad ocooperatig vehicles i commuicatio. Aother approach is based o probabilistic replica placemet approach for video streamig i the vehicular delay-tolerat etwork []. A overlay based o clusterig scheme for Mobile-IP system has bee proposed to tackle the frequet iterruptio ad dissemiatio of ivalid video fragmets. The clusterig strategy is based o groupig vehicle odes that have the same movig characteristics ad video supply requiremets. The clustered odes have the ability to lear ad take a decisio based o deploy-ability of a stored video [0]. A flexible cooperative streamig system over a cooperative vehicle fleet cosiderig mobile badwidth aggregatio strategy has bee proposed []. The study addresses the issues i K-hop cooperative streamig. I Rezede, et al. [], a solutio that employs reactive ad scalable uicast has bee preseted to address the striget requiremet of video streamig i vehicular commuicatio. However, route couplig effect i overlay approaches for video streamig. However, the above discussio focusses more o the ability to select a vehicle ode from replicated odes i the overlay. Meawhile, due to the dyamic ature of vehicular etwork frequet update of the overlay structure causes high commuicatio overhead ad ca also lead to high eergy cosumptio. I additio, the high data rate of video data is ot cosidered i the overlay trasmissio, hece cogestio i the etwork might occur which i tur lead to less video quality. B. Codig Orieted Video Streamig It is a approach that is ceter o the itegratig video compressio techiques with the best ode ad route selectio techiques. This techique is desiged to guaratee optimal video streamig quality. Both the video compressio by partitioig ad route selectio ad, formatio for video streamig must cosider Quality of Service (QoS) of the video based o stadard ad user perceptio. I geerality, both the striget requiremets of the video streamig ad VANETs limitatios eed to be cosidered to achieve qualitative video streamig delivery. The QoS/QoE based approach cosiders major QoS factors icludig delay, jitter, packet loss ad efficiet badwidth utilizatio i the video codig ad video trasmissio. I QoS/QoE approach, the target is ofte to derive peak result that will be acceptable to users of the streamed video. A Seamless quality-drive multi-hop data delivery scheme for video streamig i urba VANETs settigs []. It icorporates etwork layer scheme for seamless delivery of video stream packets i VANETs settigs. The routig scheme cosiders quality-drive parameters i order to deliver video streams from a dedicated etwork to a fixed destiatio through multi-hop commuicatio. A QoE-drive user-cetric video-o-demad service i urba multi-homed PP-based vehicular etworks is suggested to provide a ew service to achieve better QoE. Better QoE ca be achieved by cosiderig badwidth issues []. I these services, vehicles create a lower layer VANETs through wireless access i vehicular eviromet iterfaces. It geerates a upper layer PP chord overlay o top of the cellular etwork. I aother approach, a QoE-cetric codig ad routig are employed to achieve better path selectio by cosiderig mea opiio score for QoE []. This approach is hadled i four differet categories icludig choosig of path ad cotrol packet, followed by evet activatio topology cotrol packets ad, baed liks, ad the estimatio of packet loss ad mea loss burst size. Further, a QoE-drive ad likquality receiver based trasmissio is proposed for improvig the quality of video while cosiderig VANETs challegig eviromet []. I additio, a geographical receiver-based beacoless strategy is proposed as a solutio for trasmittig video streams i VANETs. However, this approach lacks the ability to segmet the video high data rate ad create load balacig i the etwork. The multipath codig is a approach, which forwards substreams through differet paths from seder vehicle to the receiver vehicle. Multipath codig-cetric routig cosiders the compressio of video while at the same time chooses most suitable ad reliable paths for video stream forwardig. I multipath video scheme, video streamig flow is partitioed ito distict paths durig trasmissio. It reduces the high video data rate, by achievig load balacig durig video trasmissio. Multipath video streamig maily focusses o path selectio algorithm. It usually employs lik/ode disjoit approaches for efficiet routig of video streams. The multipath supports attaimet of QoS i the followig ways icludig fault tolerace, load balacig ad, badwidth ad delay

4 Page of aggregatio. Aother of its kid is multisource video trasmissio. A Multipath Video Streamig solutio for vehicular etworks with Lik disjoit ad Node-disjoit (MSLND) is suggested to address FEC issues of video trasmissio i VANETs []. MSLND esures retrasmissio, rather tha the FEC. I additio, a multipath solutio cetered o the disjoit algorithm is also suggested to decrease the iterferece ad cotetio, leadig to a acceptable delay ad a higher trasmissio rate. I this, iter-frames are trasmitted through the UDP protocol while oly I-frames are trasmitted through the TCP protocol. To ehace the delay of TCP trasmissios, a ETX-TCP algorithm is itegrated to select the best ad suitable paths for TCP trasmissio. However, despite its stregth i retrasmissio of video streamig, the solutio assumes that oce there are lik ad ode disjoit strategy i the multipath selectio, the iterferece is avoided, this is ot always true because odes havig iterferece betwee each order ca be selected as ode disjoit or lik disjoit. Hece, a adequate solutio that cosiders the vehicle positio ad estimates the level of the dispersed vehicle i order to miimize route couplig is required. I aother study, a Locatio-cetric multipath approach for streamig video over VANETs (LIAITHON) has bee proposed to avoid route couplig effects []. The approach is cetered o locatio parameters to select the best multiple paths for video stream forwardig. Further, it uses forwardig zoe scheme for reducig collisio ad cogestio problem. The approach used is based o calculatig the degree of closeess of vehicle ode i order miimize the route couplig effect. However, the vehicle is very dyamic i ature, hece they chage positio. Therefore, a more dyamic solutio for miimizig route couplig, which miimizes iterferece eed to be explored. A multiple path solutios with error correctio for video streamig over VANETs (LIAITHON+) is preseted. The aim is to reduce collisio ad packet loss i high data rate etworks []. LIAITHON+ employs multiple paths approach to distribute the high data rate traffic ito a set of paths. However, the forwardig strategy cosidered is ot realistic for multiple path selectios sice the agular geometry is less the degrees, most of the odes at this rage of agle ormally iterfered. De Felice, et al. [] suggested a Distributed Beacoless Dissemiatio (DBD) routig protocol for prerecorded video Data trasmissio over VANETs. It is a itegrated framework that hadles QoE of video services ad routig protocol. DBD, further advaces the performace of IEEE 0.p/WAVE MAC layer, by resolvig the spurious forwardig problem. Li, et al. [] proposed a joit codig/routig optimizatio usig Distributed Video Codig (DVC) ad etwork codig (NC). The optimizatio is betwee video quality ad etwork lifetime, which is cetered o the iformatio theory of wireless visual sesor etwork. Similarly, Zou, et al. [] suggested a priority-based flow optimizatio i multipath ad etwork codig based routig. Further, a Fieldbased Aycast Routig (FAR) routig for pre-recorded video, it cetered o rapid multipath routig dyamics of a electrostatic potetial field model based o Poisso s equatio []. A aalysis of the probabilistic multipath trasmissio of video streamig i the multi-radio wireless etwork has bee preseted []. A Probability Geeratio Fuctio (PGF) is geerated to assess the delay metrics, such that the least chael data rate to support a video sub-stream is obtaied. Further [0], proposed a multi-path provisioig strategies cosiderig cloud assisted scalable codig video streamig with QoS requiremets. The strategies improve the performace of Scalable Video Codig (SVC). Also, a multipath strategy based o etwork proxy for video streamig has bee proposed for vehicular commuicatio []. The multipath cocept employs cocurret trasmissio, which leads to iterferece due to route couplig effect. Some solutios have bee proposed as metioed i the literature but are ot adequate. Hece, there is eed to desig ad develop a multipath video trasmissio that cosiders the route couplig effect i order to miimize iterferece. The ext is Sectio, which presets ad discuss the proposed protocol. III. INTERFERENCE-AWARE MULTIPATH VIDEO STREAMING The desig ad developmet of the iterferece-aware multipath video streamig protocol cosiderig vehicle separatio, lik ad ode disjoit, ad lik sigal power with badwidth capability. The multipath etwork model is explaied i Sectio.. A. Multipath Network Model A vehicular commuicatio is created by a set of N odes (vehicles) where N =,, ad each vehicle are equipped with a sigle radio iterface. The chaels available i the etwork is deoted as C, where C =,, c ad C Max is the highest badwidth of each chael. The IEEE 0.p/WAVE protocol provides Mbps for Japa, 0 Kbps for Europe ad Mbps for the USA. All odes are assumed to be dyamic with varyig speed. Further, odes operate with differet trasmissio power thus, havig creatig differet trasmissio coverage. Therefore, a lik l betwee two odes is active if ad oly if, it is fuctioig o a sigle chael. The lik l is iterfered if there exists a certai coverage area of collisio domai where aother lik l lies i the same chael assiged to l. The iterfered area is a shared physical coverage area of the trasmissio regio of seder ad receiver. Cosiderig coectivity, a graph G cosist of Poits ad Edges(P, E). The elemets of P are called odes ad elemets of E are the coectig liks betwee poits of the graph. We cosider vehicular etwork topology as a dyamic graph structure. Let G be a graph with a set of paths M. I geeral, a path i a graph costitute of series of distict poits p that is, set of odes p, p, p,, p k p i p i+ with a edge E, which is the lik betwee two differet poits i =,, k. The legth l of a sigle path is the sum of all edges i the path. Hece, we ifer that the agle θ betwee the selected multiple paths is iversely proportioal to the iterferece of the coverage area of each poit i the multiple paths M. M θ I () C

5 Page of Therefore, if the agle θ is assumed ad calculated before trasmittig a video stream through the selected paths cosiderig lie of directio of relayig ode. Hece, iterferece betwee multiple paths would be avoided ad qualitative video streamig delivery would be achieved. The sigificat challeges of video streamig i vehicular commuicatio is how to trasmit video data with fewest video frame loss ad miimum trasmissio delay. Due to the aforemetioed challeges, a multipath video trasmissio is employed to achieve qualitative video streamig. The video frames are split ito differet paths i order to achieve fewer frame loss ad miimum trasmissio delay. I most of the existig studies, video streamig usig multipath maily emphasizes o path selectio algorithm without cosiderig the ature of data trasmitted. There is eed to extesively cosider the ature of video data trasmitted i a oe part ad the type of protocol to trasmit distict video data. I this work, we have developed ad desiged a two paths video streamig scheme by splittig video stream ito two distict flows amely, refereceframe ad eighbor-frame. The routig protocol cosidered is geographical routig protocol, which does ot icur high etwork overhead whe compared to M-AODV. Cosiderig the MPEG compressio stadard, video frames are categorized ad defied as I-frames, P-frames, ad B- frames (see Fig. ). I-frames ormally cotais ad ecoded with importat iformatio of a etire frame. It ca be ecoded self-reliat without referece frame to retrieve frames of the video streamig. P-frames are decoded by cosiderig either I- frame or P-frame, which eed to be decoded with the referece frame of the video. Meawhile, B-frames relies o both previous ad the ext frame followig the I-frame or P-frame. Cosequetly, both P-frames ad B-frames are depedet frames based o referece frames. B I P B B B B P B B B I B Group of Pictures Fig.. Represetatio of MPEG video frames format. A Group of Picture (GOP) is a combiatio of I-frames, P- frames, ad B-frames (see Fig. ). The I-frame is usually the idirect ad direct referece frame of the P-frames ad B- frames i the GOP. P-frames ca also serve as a referece frame or predicted frame based o the other referece frame. If the source of the predictio is traced, a I-frame will be reached which does ot deped o ay referece frame. Thus, wheever a I-frame is lost or damaged, the etire GOP might be lost or damaged. Nevertheless, oce trasmissio of I-frame is guarateed, the quality of the etire GOP ca be ehaced. I order to maitai the quality of video at the period of trasmissio, priority level eeds to be assiged based o the importace of the type of video data. For example, the I-frame is essetial i predictig both P-frames ad B-frames, therefore I-frames would have a higher priority o accessig ad utilizig etwork resources. While P-frames ad B-frames will have lesser priority i accessig ad utilizig the same etwork resources. I this study, we partitioed the video streamig trasmissio ito two amely, referece-frames, which represet I-frames, ad eighbor-frames, which represets both P-frames ad B-frames. Referece-frames ad eighbor-frames are trasmitted o primary ad secodary paths respectively, which is based geographical routig protocol (greedy routig). Hece, the primary path has higher priority because of the I- frame compared to the secodary path for P-frames ad B- frames. Sice the aim is to miimize iterferece due route couplig i multipath setup. There is eed to estimate iterferece based o some parameters i the ext hop vehicle of the multipath, the followig parameters are cosidered for avoidig route iterferece icludig i) agle betwee the two first forwardig vehicles, which are eighbors to the source vehicle ad ii) the lik quality. The lik quality is measured cosiderig lik sigal power, badwidth capacity (BC), packet error rate ad the sigal to oise ratio of the lik. The parameters have bee assiged with same weight fuctio sice every parameter is importat for achievig qualitative lik. The sum of the total weight score is oe. The weight associated with each parameter is represeted as follows. 0. Dispersed Agle 0. PER 0. SNR Weight fuctio = 0. BC 0. LSP { B. Iterferece i Multipath Video Streamig The iterferece level of odes i a multipath setup ca be symmetrically reduced if the agle betwee the correspodig two odes ca be wideed such that iterferece coverage of each ode does ot overlap with oe aother. I order to mathematically formulate the cocept of the agle. We cosider a lie with a distict edpoit P where P serve as a source vehicle ode SVN ad P is the itermediary ode (relay ode). Sice we are cosiderig a two paths trasmissio, we cosider aother lie P coectig from P that is P, hece, a agle is formed betwee two lies with the same edpoit which is calculated i degree ad is amed agle of the multipath (vertex), that is P P P (see Fig. ). I multipath video trasmissio, the agle betwee the SVN ad the two relay odes from the correspodig two paths eed to be cosidered.

6 Page of P P h P0 P θ P P P P θ P P P P P P0 P θ P P p h p p p P θ P P P P Fig.. Vehicular Commuicatio Sceario Forms a Obtuse Triagle The agle betwee the SVN ad the two relay odes of the selected paths is proportioal to the iterferece coverage area of each ode i the two paths. The suitable separatig agle betwee P ad P, P is a obtuse agle, sice P P P > 0 ad P P P < 0 which has the ability of reducig iterferece i the multipath commuicatio. First, let fid the area of the obtuse triagle cosiderig P as the base of the triagle (see Eq. ). Area of P P P = [P 0 P P ] [P 0 P P ] () Where breadth of the obtuse triagle is P P = b. Therefore, we deduced that area of the triagle is expressed as i Eq. : O area = P P P = h b () To estimate a agle of the multipath video packet forwardig, we eed to calculate the obtuse agle where 0 > θ < 0. Usig cosie rule, a obtuse triagle with side dimesios p p p ca be used to calculate the multipath suitable agle, we cosider θ for agle P, which is opposite side p as follows: cos θ = p +p p p p θ = Cos ( p +p p p p ) () A agle is said to be obtuse, if ad oly if cos θ < 0. Hece, a obtuse triagle fulfils p + p < p, p + p < p, ad p + p < p. C. Probabilistic Model for Video Streamig I this sectio, the circular trasmissio coverage area of the vehicle ode is cosidered. A SVN P is assumed to be at the ceter poit of diameter of the circular coverage area with two other vehicle odes P P, which they serve as relay odes. They also form a obtuse agle with P i order to reduce iterferece while creatig two paths trasmissio for video streamig. The existece of three vehicle odes that forms a obtuse triagle i the coverage area relies o obtuse agle θ, the vehicle ode desity λ ad the trasmissio coverage, which are the two p Radii R p. The aim is to ivestigate the impact of p parameters θ, λ ad R p o the probability of fidig at least two vehicles odes, which forms a obtuse triagle. I order to achieve a obtuse triagle, a rage of θ values is give as 0 > θ < 0 util two vehicle odes are foud. The vehicle odes are avigatig i a two dimesioal etwork regio ad presece of two vehicles i the etwork regio strictly follows Poisso Distributio Fuctio (PDF) cosiderig vehicle ode desity λ. Cosiderig the average desity of vehicle odes i a etwork coverage, the frequecy of vehicle odes available to form a obtuse agle is calculated by employig Poisso distributio. I additio, each vehicle ode is idepedet ad vehicle odes are selected to serve as a relay ode, which are chose at radom cosiderig obtuse agle requiremet. Several research works have bee coducted i order to miimize iterferece i data packet trasmissio i vehicular commuicatio. However, few studies of multipath video data trasmissio have focused o iterferece i the routig process. The studies i Wag, et al. [] ad Schmidt, et al. [] are basically o usig received sigal stregth as the estimatig factor to measure iterferece level of a lik, which is ot adequate to have qualitative video streamig trasmissio due to dyamic ature of VANET odes. Therefore, we use a geometric agle estimatio, which ca assist i miimizig iterferece i a multipath video streamig trasmissio. The ivestigatio deduced that large dispersio of agle θ that is 0 > θ < 0 coected to the two paths reduces multipath iterferece. I additio, if the desity of vehicles is high, there is eed for smaller trasmissio coverage i order to do away with iterferece, which leads video data collisio. Hece, we cosider a value of radius (00 m) for the coverage area i this study. Let assume Y represets the radom variable which is the frequecy of vehicle odes that ca form a obtuse triagle, the the probability of the availability of g vehicle odes that forms a obtuse triagle area i a No-Shadowig Settig (NSS) P NSS Oarea (Y = g) is calculated as show i Eq. (): P NSS Oarea (Y = g) = (λ O area )g e (λ O area ) g! By substitutig O area give i Eq. (), the we have Equatio (): P NSS Oarea (Y = g) = [λ( h(b))]g g! () e λ( h(b)) () If we substitute g = 0, probability P NSS Oarea (Y = 0) of o vehicle available i the obtuse triagle area cosiderig NSS, is expressed i Eq. () as follows: P NSS Oarea (Y = 0) = e λ( h(b)) ()

7 Page of The probability P NSS Oarea (Y = ) of the presece of at least oe vehicle ode i the obtuse triagle area cosiderig NSS is preseted as follows i Eq. (): P NSS Oarea (Y = ) = e λ( h(b)) () D. Impact of Shadowig o Video Trasmissio To achieve a more realistic probabilistic aalysis of the presece of more tha oe vehicle i a obtuse triagle area, shadowig settigs must be cosidered. Shadowig is caused due to obstructio of huge vehicles, buildigs, ad other physical objects. These lead to o-circular trasmissio coverage. Therefore, o-circular trasmissio coverage is employed for itegratig shadowig model cosiderig obtuse triagle area. Trasmissio coverage is usually varied i terms of directio due to the impact of shadowig o the received sigal power []. The received sigal power is expressed as i Eq. (): PS r = PS t {0 log 0 K 0 ω log 0 d d 0 τ} () Costat K represets chael atteuatio ad atea characteristics, path loss expoet is represeted as ω. Distace betwee odes ad referece distace for odes atea are deoted as d ad d 0 respectively. Where τ is the Gaussia ocetralized radom variable cosidered. R θ R r Fig.. Effect of Shadowig Circular Trasmissio Coverage π R UB area = P(PS πr r (r) PS mi ) rdr dθ 0 0 (0) PS r (r) is the received sigal power i sa at certai distace r. We employ Log-ormal distributio, because it precisely ad accurately models the differece i received sigal power, which is due to shadowig []. Hece, by employig Logormal distributio, the probability of PS r at r beig higher tha PS mi, which is represeted as P(PS r (r) PS mi ) ad is further mathematically modeled as i Eq. () P(PS r (r) PS mi ) = φ ( PS mi (PS t +0 log 0 K 0ω log 0 ( r d )) o ) () σ τ Where, φ(t) = y t π e dy, ad σ τ is the variace of τ. By cosiderig Eq. (), UB area ca be expressed as give i Eq. (): UB area = R φ (y + z log r ) r dr () R 0 R Where PS r mea deotes the average received sigal power at certai distace R, z = 0ω log 0(e) ad y = PS mi PS r σ τ Eq. () is further simplified as i equatio (): UB area = φ(y) + e ( yz z ) φ( yz ) z mea (R). σ τ () Further, we assume that PS r mea (R) = PS mi, Eq. () ca be further simplified as i Eq. (): UB area = + e z ( z φ) () Eq. is modified by itroducig UB area to fid the probability of availability of oe or more vehicles i a obtuse triagle area SS cosiderig shadowig settigs P Oarea,, which is represeted as follows i Eq. (): SS P Oarea (Y ) = e λ( h(b)) ( UB area ) () πr E. Lik Quality Model for Video Trasmissio I vehicular commuicatio, vehicles have geographical iformatio through the use of GPS. The Lik Quality (LQ) betwee a seder ad a receiver vehicle ca be approximated by cosiderig the lik sigal power of the receiver, the badwidth capacity, the packet error rate ad the sigal-to-oise ratio. The estimatio of LQ has the ability to give a idea of the Blocked sigal area iterferece level of a ext hop vehicle. The predictio of the iterferece level will assist i selectig the best vehicles i the multiple paths for the video packet trasmissio. To estimate Small area (sa) the LQ, the receive sigal power with the most widely Ublocked sigal area (UBarea) acceptable two-ray groud reflectio model has bee employed. Further, a shadowig model which is more appropriate for vehicular commuicatio eviromet is cosidered to predict actual LQ of the selected multiple paths i order to avoid paths with iterferece. The lik received sigal power betwee a trasmitter ad receiver vehicles are based o the two-ray groud reflectio model, which is give as i Eq. (): PS r = PS tg t G r H t Hr () ( d l ) S l PS r ad PS t are the received sigal power of the receiver ad trasmitter respectively, G t ad G r are the atea gai of trasmittig ad receivig ode, the H t ad H r represets the height of trasmittig ad receivig odes ateas, d l is the distace of the lik betwee seder ad receiver ode, ad S l is the multipath system loss. Meawhile, i practicality, the received sigal power is ot a sufficiet parameter to determie the LQ ad viability of the lik for the relay ode. Therefore, the badwidth capacity, packet error rate ad Sigal to Noise Ratio (SNR) eed to be estimated. Video data is ormally large i size thus, large size of badwidth is required for efficiet ad qualitative video streamig trasmissio. I video trasmissio, badwidth estimatio is regarded as the whole quatity of video data trasmitted divided by the playback period. Thus, the

8 Page of Badwidth Capacity cosiderig Video Data (BC V D ) ca be mathematically expressed i Eq. () as follows: Q BC V D = VD T () PB T Where VD Q T is the quatity of video data trasmitted ad PB T is the playback period durig video data trasmissio. The Sigal to Noise Ratio (SNR) of the lik is cosidered i respect to video streamig. As previously stated, qualitative video streamig trasmissio requires zero or miimum oise i the trasmissio lik. The SNR is a essetial parameter for lik quality predictio, which is mathematically depicted as follows i Eq. (): ap SNR l = PS r () PS th +ap PS if The ap represets the amplitude of the fadig chael usig Rayleigh distributio, thermal oise sigal power is assumed as PS th ad PS if represet the iterferece sigal power of the lik. I order to cosider the packet error rate, we start from the Bit Error Rate of the lik (BER l ), we use biary phase shift keyig modulatio which depicted i Eq. () as follows: BER l = ( SNR l ) +SNR l () I the case of Packet Error Rate of the lik (PER l ) cosiderig a sigle lik, trasmissio is computed as demostrated i Eq. (0): PER l = ( ( BER l ) LT ) (0) By cosiderig vehicle odes dyamic fuctios for lik breakage, the we preset Eq. () as show: PER l = ( ( BER l ) LT ) + {f q (w)} () The legth of the packet i bits is represeted as LT ad f q (w) is the vehicle ode dyamic fuctio cosiderig the striget delay requiremet for video delivery. Eq. () is the geeric formula for PER l caused due to lik breakage, which does ot iclude lik breakage due to dyamicity of the vehicles. The secod part of Eq. () that is, {f q (w)} is the empirical fuctio used to estimate PER l probability because of abrupt route chagig of vehicular odes. Based o the fuctio f q (w), it is assumed that it has previous kowledge compoet ad ca forecast future heuristic compoet. The mathematical represetatio of the fuctio f q (w) is show i Eq. (): f q (w) = ( ) () {y(w)+z(w)} Where y(w) is the frequecy of differet route chage take ad speed rate by a ode i the previous avigatio. z(w) represets the umber of route chages ad speed rate expected by the ode i future to reach the destiatio usig path with miimum cost. By usig the aforemetioed fuctio, wheever there is frequecy icrease i either chage of routes, speed rate or both, the the {y(w) + z(w)} icreases. The value of vehicle mobility fuctio also icreases withi the rage of 0 f q (w). Video packets are retrasmitted through multiple paths, wheever a trasmissio failure occur. A packet ca be effective at least oce i retrasmissios through multiple paths. The probability of effective trasmissio ca be mathematically represeted as i=( PER l ) PER i l. The retrasmissio attempt is idicated as i. Cosequetly, the PER l over a sigle lik based o multipath with retrasmissio ca be expressed as i Eq. () as follows: PER i l = i=0 ( PER l )PER l () Packet Error Rate PER l of a multiple path with retrasmissio i a sigle lik, which is made up of k umber of odes is expressed as i Eq. (): PER path = ( PER l ) k () For k umber of odes i two paths lik is mathematically formulated as preseted i Eq. () PER Mpath = (( PER l ) k ) () Thus, we employ all the aforemetioed derived parameters i order to select a qualitative lik for video trasmissio i vehicular commuicatio. By cosiderig all the parameters, qualitative video streamig delivery ca be attaied. I the ext sectio, we preset some algorithms developed for the video streamig routig ad further discuss their fuctioality ad viability. F. IMSLND Algorithm I this subsectio, the Iterferece-aware Multipath Video Streamig Solutio with Lik ad Node disjoit (IMSLND) algorithm is developed based o the geographical routig protocol. The IMSLND algorithm icludes ode disjoit protocol, ext forwardig vehicle protocol, ad the multipath cocept. The algorithm is aimed at reducig iterferece betwee multipath trasmissios. It also miimizes forwardig overhead ad improves the NFV selectio criteria. The criteria are to avoid paths with iterferece while selectig the lik with the best quality. The algorithm cosiders multipath agle that avoids iterferece durig path selectio, lik quality, ad ext forwardig vehicle selectio decisio. The IMSLND is preseted as follows, startig with ode disjoit algorithm, followed by ext forwardig vehicle algorithm ad the the mai IMSLND algorithm. The multipath video trasmissio cocept cosiders ode disjoit as i []. The ode disjoit strategy employs two paths, such that there is o commo ode betwee the paths durig video trasmissio. It has a low collisio possibility with striget requiremet whe merged with lik disjoit strategy. Cosequetly, ode disjoit path selectio strategy is suitable for collisio-aware trasmissio such as video trasmissio i vehicular commuicatio. The complexity of Algorithm is

9 Page of preseted as follow; sice two paths are cosidered, the we have path as p with m legth ad path as p with legth. The Algorithm complexity is to decide ad select two paths that are ode disjoit ad which ode has higher agle of dispersio. Although, the agle of dispersio is oly cosidered for the first two odes which are selected by the SVN. It is by comparig all the odes that exist i the two paths, which is O(m). I additio, sice the compariso iclude sortig of the two possible paths by employig Quicksort, the mea complexity is O( log ). Cosiderig the sorted paths ad the geometric agle relatioship betwee odes of the two paths, we compare the odes i the two paths based o Algorithm. The mea complexity of Algorithm is O( log ) + O(m + ), > m. The worst-case situatio of Algorithm is whe all odes of the two selected paths are scaed, which is O( + m). I this situatio, computatio complexity of the worst case of Quicksort process is O( ). Although, Quicksort computatio complexity of the worst case ca be avoided, where > m. Hece, the worst case computatio complexity is O( ). The best-case situatio of the Algorithm occurs if there are fewer umber of odes i oe path ad the agle betwee the closest selected ode is greater tha 0 compared to the other path. The best-case computatio complexity is O(), for the reaso that oly oe path is scaed. I additio, computatio complexity of the best-case situatio of Quicksort is O( log ). Therefore, the computatio complexity of the best-case situatio of the ode disjoit algorithm is O( log ). Algorithm Fuctio Notatio Iput Process : : : : : : : : : 0: : : : : : : : Node Disjoit Vehicle Selectio Algorithm p : Legth of the first path p : Legth of the secod path i: Nodes i the first path j: Nodes i the secod path θ: The agle betwee the two ode disjoit paths p, p, i, j Iitializatio p > 0 p > 0 i, j = 0 Idetify (p ) ad sort (p ) Idetify (p ) ad sort (p ) While i < p or j < p do If θ of the first two odes: p {i} ad p {j} > 0 < 0 the Retur (true) Else if Agle betweem p {i} ad p {j} is 0 0 the Retur (false) If p {i} = p {j} the Retur (false) Else if p {i} < p {j} the Icreamet i If p {i} > p {j} the Icreamet j : : 0: : : Output Else Execute lie Ed if Ed while Retur (true) Two Paths Node Disjoit I this subsectio, the cocept of itermediate ode selectio after the two qualified odes for the multipath are chose based o the Azimuth triagle coordiates positio of the selected ode that is, ext forwardig ode. Each ode calculates it the relative agle of directio to the eighbor odes ad selects a ode that has the same coordiate positio ad satisfies the aforemetioed parameters. This ode is made as VRN, the process is cotiued i both paths util video packets get to the DVN. Note that, NFN is the same as the VRN. The complexity of this algorithm is related to that of the compariso complexity i Algorithm that is the ode disjoit algorithm. Cosiderig the fact that, at the ode selectio oly compariso is made based o the coordiate positio ad the parameters of the odes. Hece, the complexity of the compariso is O(m), further, sice the compariso iclude sortig the the mea complexity of the sortig is O( log ). Therefore, the mea complexity of Algorithm is O( log ) + O(m + ). The worst case sceario of Algorithm is whe all the eighbor odes of a PFN of the two paths are scaed which is O( + m). For the worst-case sceario of quicksort process, the complexity is O( ). Eve though, the computatioal complexity of the worst-case sceario for Quicksort ca be avoided if the umber of eighbor ode is oe or two ad whe the first scaed NFN is the most suitable ode based o the coordiate positio ad parameters. Hece, i that situatio the worst case computatioal complexity is O(). The best-case sceario of the Algorithm occurs if there are fewer umber of eighbor odes to PFN of the two paths. The best-case computatio complexity is O(), for the reaso that oly oe or few eighbor odes are, scaed from the two paths. I additio, computatio complexity of the best-case situatio of Quicksort is O( log ). Therefore, the computatio complexity of the best-case situatio of the ext forwardig ode selectio algorithm is O( log ). Algorithm Fuctio Iput Process : : : : : : : : : Next Hope Vehicle Selectio Algorithm p, p, i, j Iitializatio p > 0 p > 0 i, j = NFN eighbor odes of PFN While VRN = PFV do Calculate PFN relative coordiate directio to NFN ad parameters If the NFN == VRN the Retur (true)

10 Page 0 of : : : : : : Output Else if NFN == DVN the Retur (false) Forward to DVN without calculatig coordiate directio ad metrics Ed if Ed while Retur (true) NFN amog the itermediate odes I algorithm, the complete process of the IMSLND protocol is logically preseted. The video packet is forwarded from SVN through the itermediate odes of the multiple paths, the to the DVN. The detailed discussio of the video streamig routig process is show after the algorithm. Algori thm Notati os IMSLND DVN: Destiatio Vehicle Node SVN: Source Vehicle Node PFV: Preset Forwardig Vehicle VSN: Video Source Node SVOT: Set of Vehicles i Obtuse Triagle NFV: Next Forwardig Vehicle θ: The agle betwee the two ode disjoit paths Q lik : Quality of sigle lik of i th vehicles i obtuse QV: Qualified Vehicle is vehicle that fulfills all re SRV: Set of Reachable Vehicles p : Legth of the first path p : Legth of the secod path PS r, PS t, H t, H r, G t, G r, d l, S l, b, h, λ Iput Proce ss. Iitializatio SVOT = ull VRN = ull SVN = PFV θ =. SVOT. = {vehicles i the trasmissio rage of PFV} SVN is eighbor of DVN ad If ode disjoit ad ( SVN == VSN ) the Forward the video packet directly to DVN usig two QV from VSN Exit. Else While (DVN SRV ad SVN VSN = NFV) Forward the video packet to DVN usig two qualified lik Ed while. Else While (SVOT = ull) a. Calculate obtuse triagle area usig Eq. ( SVOT = {vehicles i O area } b. If (SVOT = ull & 0 > θ < 0 ) the icremet θ by Else Wait for radom quatity of time Ed while Ed if. For each vehicle SVOT V Calculate badwidth capacity BC D of each Eq. (): BC V D = VD T PB T Ed for. For each vehicle SVOT Calculate packet error rate PER Mpath of eigh PER path = ( ( PER l ) k ) Ed for. For each vehicle SVOT Calculate SNR of the eighbor ode lik usig Eq. () Ed for. p (BC V D + PER path + SNR) = Max{BC V D (liks(p )) + PER path (liks(p )) + SNR} 0. p (BC V D + PER path + SNR) = Max{BC V D (liks(p )) + PER path (liks(p )) + SNR}. QV == NFV. Trasmit the video packet to NFV up to DVN cosiderig NFV selectio process. Exit Outpu t NFV for SVN = VSN NFV for SVN VSN G. Explaatio of IMSLND Algorithm The IMSLND algorithm executes steps -, wheever the vehicle source ode SVN wats to trasmit video packet to a certai destiatio vehicle ode (DVN) i the etwork. The step, is the iitializatio of variables. I the d step, the SVOT acquires iformatio about the positios of their immediate eighbors ode positio with reply timestamp. This iformatio are used by the preset forwardig vehicle PFV. I the rd step, the PFV ispect for whether DVN is i SVOT ad if source vehicle ode SVN is the same as video source ode VSN, ad if DVN is foud amog the SVOT set ad SVN is the same as VSN, the PFV forward the video packet to NFN usig available two qualified vehicle QV liks. I step, if SVN is ot the same as VSN ad DVN are foud amog the SVOT set, the forward the video packet to NFV usig available QV lik. I the case where step ad are ot foud, the algorithm executes step, i which a segmet formed a obtuse triagle with sector usig agle 0 > θ < 0 is determied. The badwidth capacity of each vehicle lik i SVOT is computed i the th step. I the th step, the quality of each vehicle lik i the SVOT Q

11 Page of based o PER is calculated. Also i the th step, the SNR is estimated to kow the distace of the ode ad it sigal quality. I the th-0th steps, the Next Forwardig Vehicle (NFV) is determied for the two paths based o Azimuth coordiate system i order to forward the video packet to the ext ode cosiderig iterferece route couplig. I the th step, the NFV is the same as the QV, sice the qualified vehicle is always chose as the relay vehicle. I the th step, the video packet is delivered to the NFN which becomes the PFV. Meawhile, i the th step, the video packet trasmissio is termiated. Step - ad - are employed at vehicle hop util the video packet is delivered to DVN. Figure is preseted i order to aid uderstadig of the steps ad logical flow i the algorithm. The computatioal complexity of the IMSLND algorithm is the sum of the total computatio complexity of either worst case ad or best-case sceario of Algorithm ad. Thus, the computatio complexity etails both for compariso ad sortig process. IV. CASE STUDY-BASED EXPERIMENT I this sectio, experimetal results obtaied to examie the performace of the basic mathematical formulatios ad the suggested approach have bee preseted. The sectio is distictively categorized ito two subsectios amely, subsectio. ad.. Subsectio., etails umerical results obtaied for validatig the mathematical formulatios. While subsectio., is the discussio of simulatio results obtaied ad the bechmarkig coducted. A. Numerical Results I this subsectio, the umerical results have bee geerated usig MATLAB to examie the effect of parameter variatios o the mathematical formulatios. The correspodig set of values of differet parameters eeded to geerate the results have bee stated i the various plots. The effect of parameter variatios o the probability of availability of oe or more odes i a obtuse triagle area cosiderig o-shadowig NSS settigs ( P Oarea (Y )) ad shadowig settigs SS ( P Oarea (Y )) are depicted i Fig.. Cosiderig the results show i Fig. (a), it demostrates that for the offset agle θ =, the probability of availability of oe or more vehicles i the obtuse agle area cosiderig o-shadowig settigs is 0. for vehicle desity λ = vhc/m. The value θ = is cosidered to be the least threshold value to examie the performace of our proposed approach. The result show i Fig. (b) demostrates that for each of the vehicle desities cosidered, the NSS probability P Oarea (Y )of availability of oe or more vehicles i the obtuse agle area cosiderig o-shadowig settigs is greater tha 0. for the trasmissio rage of 0 m to 00 m. The result is used to aalyze the performace of our proposed approach. I Fig. (c), the result shows that, for NSS obtuse agle θ =, the propability P Oarea (Y )of availability of oe or more vehicles i the obtuse agle area cosiderig o-shadowig settigs is greater tha 0. for vehicle desity λ = vhc/m. Specifically, the NSS probability P Oarea (Y ) rises with the icrease i obtuse agle for ay precise value of the desity cosidered. For NSS istace, whe θ = 0, the probability P Oarea (Y ) attai a value of.0 at desity λ = vhc/m, which is the highest probability value. Further, the effect of shadowig o the probability of availability of oe or more vehicles i the obtuse triagle area has bee depicted i the result Fig. (d). The result demostrates that, shadowig has great effect o a smaller obtuse triagle agle. For example, whe θ < 0, but with the rise i obtuse triagle agle θ > 0 the effect is

12 Page of miimized sigificatly. I the ext diagram (see Fig. ), we depict the probability of packet error rate i oe hop coverage cosiderig o-shadowig ad shadowig settigs. (a) (c) (d) Fig.. The probability of availability of oe or more vehicle i the obtuse triagle area (a) ad (b) represets the availability of vehicle cosiderig agle ad trasmissio coverage respectively, (c) ad (d) represets vehicle desity ad agle NSS ad SS. The results of the probability of packet error rate PER path i both sigle ad multiple paths PER Mpath with retrasmissio is show i Fig. (a, b, c ad d). The result preseted i Fig. (a) depicts that packet error rate is at lowest whe oe-hop coverage is 00-0m for both sigle ad multipath trasmissio at various values of θ. However, packet error rate for multipath is lower compared to that of the sigle path (see Fig. (a ad b)) due to achievig load balacig, path diversity ad miimizatio of iterferece betwee two paths. The whole of the observatios have bee employed for the selectio of ext forwardig vehicle with the best lik quality for video streamig. Further, the result i Fig. (c) shows that the effect of shadowig o packet error rate is highly oticeable for sigle path, but lesser for multipath ext forwardig vehicle trasmissio. B. Simulatio ad Results Aalysis The results of simulatios coducted to examie the performace of IMSLND are preseted i this subsectio. The performace is tested cosiderig dyamicity ad frequet positio chages of vehicles i the etwork topology due to the high mobility of vehicle odes. I additio, the performace of IMSLND is tested cosiderig varied desities i a urba traffic settig. I the simulatio, Peak Sigal to Noise Ratio (PSNR), Structural Similarity (SSIM) idex, Data Receivig (b) Rate (DRR) ad delay i the etwork have bee measured. The results achieved for IMSLND are compared with two baselie protocols amely, MSLND ad FEC. We first discuss the simulatio eviromet ad setup. (a) (c) (d) Fig.. Represet impact of parameters o PER path (a) ad (b) sigle ad multipath i a o-shadowig settigs, (c) ad (d) sigle ad multipath i shadowig settigs. ) Simulatio Settigs IMSLND has bee implemeted usig the etwork simulator NS-. [], Evalvid [] ad mobility model geerator for VANETs (MOVE) from Simulatio of Urba MObilty (SUMO) []. NS- is a stadard etwork simulator, which has the capability of mimickig etwork traffic ad commuicatio scearios for ormal data ad multimedia data. Evalvid is a acceptable video quality evaluatio tool, which offers tool-sets of video files ad framework for the assessmet of video trasmissio. MOVE has the capability of geeratig realistic mobility model i a urba traffic settig. MOVE is developed o the upper layer of a ope source micro traffic simulator. The ecessary features of vehicle mobility traffic settigs icludig a umber of laes ad roads, umber of directio flow i each road laes, umber of traffic lights ad juctios, acceleratios, the speed of vehicles, the probability of turig right or left of a vehicle at a specific juctio have bee put ito cosideratio ad implemeted. By usig the two key modules of MOVE icludig vehicle movemet maager ad road map maager. I additio, the mobility traces created usig MOVE with the aid of SUMO is directly employed i NS- (See Fig. ). Two scearios of vehicular traffic settigs are cosidered icludig simple lae urba sceario ad high-desity urba sceario. I the simple lae sceario, all vehicles are o (b)

13 Page of multiple laes i the same directio of the road. The aim of usig simple lae sceario is to examie the performace of IMSLND i low dese urba settigs. Forty (0) vehicles are distributed across three () laes of the road, which are avigatig i the same directio. Durig avigatio i the simulatio sceario, a video is trasmitted from source vehicle through multipath itermediate vehicles, the to the destiatio vehicle. The speed cosidered for each vehicle rage from. to. m/s (0 to 0 km/h). The legth ad breadth of the simple lae sceario are,000,00 m. I the high-desity urba sceario, map-based setup is cosidered, it is based o road etwork of Johor Bahru (Jala Abu Bakar) Malaysia (see Fig. ). A OpeStreetMap (OSM) satellite image of the city is geerated ad imported ito SUMO that icorporates mobility ad etwork iformatio with the map. Afterward, desig ad cofiguratio of trace files are geerated i relatio to vehicle traffic flow timig i the Johor Bahru map, which is produced to examie the performace of IMSLND i a simple lae ad high-desity urba traffic settigs. The whole cocept of buildig Johor Bahru city map o SUMO is based o OSM. I the high-desity urba sceario, a umber of vehicles i the simulatio setup is varied from 00 to 00 i order to examie the performace of differet etwork desity. The simulatio results are geerated based o the mea average of all vehicles i the etwork. The simulatio area covered by the high-desity urba sceario is m. Fig.. City Map of Johor Bahru Malaysia The selected video for trasmissio is the well-kow bridge-far_cif, which has a streamig duratio of secods, rather tha silet_qcif ad akiyo_cif with streamig duratio of oly secods, the bridge-far_cif is used so as to evaluate the log duratio effect of differet protocols. The cosidered metrics for examiig the performace of the simulatio iclude peak sigal to oise ratio (PSNR), Structural SIMilarity (SSIM) idex data receivig rate ad delay. These metrics measure the quality of the trasmitted video. Due to the fact that, video quality is defied by the trasmissio rate of the seder, hece, Data Receivig Rate (DRR) has also bee measured. I the simulatio, real-time video streamig is evaluated. Thus, the fixed data rate for the video trasmissio has bee cosidered. All the differet phases of the simulatios are executed -0 times, which gives the advatage of takig the mea average of the results of the simulatio. I order to attai reliable mea average results, % cofidece level has bee cosidered for the cofidece iterval. However, the complete procedure of the simulatio has bee show i Fig. Table, presets simulatio parameters ad criteria cosidered for implemetatio of IMSLND protocol. Prepare Satellite Image of Johor Bahru, Malaysia Via OSM Trace File Via MOVE as.sumo.tr Tcl Simulatio Script Video Trace Files Cotaiig Packet Size Vehicular Traffic Flow Via Traffic Model Creator of MOVE as.tcl Network Simulator Tool (NS-) C++ Simulatio Objects TclCL SUMO NS- Shell Executable Commad Geerate Usig MP-Files OTcl Simulatio Objects Evalvid Produce Result Cofigure Johor Bahru Implicit Attributes of the Map (.od.xml ad.edg.xml) Johor Bahru Map Via Netcovert as.et.xml (johorbahru.et.xml) Simulatio Trace File Geerate Seder Trace Files i NS- Fig.. Simulatio Tools ad Procedure Table. Simulatio Parameters NAM (Aimatio) Xgraph (Plottig) Produce a Receiver Trace Files Parameters Values Simple lae area,000,00 m Urba simulatio area m Simulatio time 00 s Vehicle speed. to. m/s (0 to 0 km/h) Number of vehicles 00 to 00 MAC protocol IEEE 0.p Video resolutio Video play duratio s Trasmissio rage Frequecy 00 m. GHz Propagatio model Shadowig Atea model Omi-directioal Chael type Packet type Hello packet timeout Wireless TCP ad UDP secod Scearios -Simple Laes ad

14 Page of Bechmarked protocol Metrics -High-desity urba sceario -IMSLND -MSLND -FEC PNSR, SSIM idex, Data Receivig Rate, Delay I the simulatio setup, IEEE 0.p has bee cosidered, because is the stadard Wireless Access i Vehicular Eviromet (WAVE) protocol. For the propagatio model, shadowig model has bee employed which is the most realistic model. Sigal coverage of each ode i the simulatio has bee set to 00 m. Three protocols have bee compared icludig proposed protocol, MSLND, Forward Error Correctio (FEC). For each sceario simulatio, 00 s has bee set because the time is greater tha the whole time required for video trasmissio. The PSNR ad SSIM have bee evaluated at the receivers ed i simple lae sceario. The Receivig data rate is estimated based o overall received video packets divided by overall trasmissio time. Delay is the summatio of startup delay, propagatio delay, trasmittig delay, queuig delay ad processig delay ecoutered durig trasmissio. Cosiderig simulatio of urba sceario, the results are based o average outcomes of all odes that received the video streams. The evaluatio of all metrics is the same as that of the simple lae sceario. ) Simulatio Settigs I this subsectio, the followig video quality metrics are employed to compare the performace of the IMSLND with the baselie protocol are preseted. The metrics iclude PSNR, SSIM idex, received data rate ad ed-to-ed delay. i) Peak Sigal to Noise Ratio: is a expressio of the ratio betwee the maximum probable value of a sigal ad the power of distortig oise that affects the quality of its represetatio. Because, may sigals have a very wide dyamic rage, (ratio betwee the largest ad smallest probable values of a variable quatity) [,, ]. The PSNR is geerally expressed i terms of the logarithmic decibel scale, see Eq. () ad (). PSNR = 0 log 0 ( MAX f MSE ) () The Mea Square Error (MSE) ca be expressed as: MSE = m f(i, j) g(i, j) m 0 0 () Where fdeotes matrix data of the origial image, g represets matrix data of a degraded image i questio. m Represets umbers of rows of pixels of the images ad i represets the idex of that row represets the umber of colums of pixels of the image ad j represets the idex of that colum. MAX f is the maximum sigal value that exists i the origial kow to be good image. ii) Structural Similarity idex is a approach used to calculate the perceived similarity betwee the trasmitted video images ad the origial video images. The calculatio of the SSIM idex is grouped ito three icludig cotrast, lumiace, ad structural assessmet. Cotrast assessmet [, 0] Ct(a, b) is the differece of σ a ad σ b the, we have Ct(a, b) = σ aσ b +K σ a +σ b +K while the Lumiace assessmet is deoted as L(a, b) = μ aμ b +K μ a +μ where K b+k ad K are costat. The structural assessmet is carried out as St(a, b) o the ormalized sigals a μa σaad (b μ b ) σ b Thus, S(a, b) = f(ct(a, b), L(a, b), St(a, b)) the fially, Eq. () ad () are depicted as follows: (μ SSIM(a, b) = a μ b +K )(σ ab +K ). () (μ a +μ b+k )(σ a +σ b+k ) MSSIM(a, b) = M SSIM(a M j= j, b j ) () Where μ a, μ b represets local meas, σ a ad σ b is the stadard deviatios, while σ ab is the cross-covariace for video images a, b. iii) Data Receivig Rate is the ratio betwee a umber of data packets i video streamig geerated at source vehicle ad a umber of the data packet i video streamig delivered to vehicles at destiatio []. The statistical formula used to calculate Data Received Rate (DRR) i terms of percetage ca be expressed i Eq. (0) as follows: DRR% = { ( PS i= ) PR N } 00 (0) Where PS deotes umber of data packet i video streamig set i i th simulatio ru ad PR represets umber of data packet i video streamig received i i th simulatio ru. iv) Delay: is the total time take for a video stream to be trasmitted from source to a specified destiatio []. The total delay time ecompasses five major steps i video stream trasmissio icludig startup delay, propagatio delay, trasmissio delay, queuig delay ad processig delay. It ca be represeted mathematical i Eq. () as follows. DL = N D St, D Pr, D Tr, D Qu, D Prs () Where N is the umber of liks (hops) i the etwork, D St is the startup delay of the video stream. D Pr is the propagatio delay for trasmittig a sigle bit from source to destiatio, D Tr is the trasmissio delay of a video streamig packet trasmitted from source to destiatio. D Qu is the queuig delay for video stream before trasmissio ad D Prs is the processig delay ecoutered durig video streamig from source to destiatio.

15 Page of ) Aalysis of the Results The simulatio results achieved for the proposed algorithm, which has % cofidece iterval have bee preseted. This subsectio has bee categorized ito two () amely, subsectio A ad B. Subsectio A cosist of bechmark aalysis of results achieved for simple lae sceario. Meawhile, subsectio B cosist of bechmark aalysis of results achieved for the urba sceario. i) Simple lae sceario: I the case of simple lae sceario, the video trasmissio starts at approximately secods after the simulatio starts. The playtime duratio of the video is secods, which is the same as the time take for the source ode to trasmit the video. Forty (0) vehicles are simulated i the simple lae sceario. The simulatio results are depicted i Figs. 0(a) to 0(d). Peak Sigal to Noise Ratio (PSNR) has bee oe of the most commoly employed metrics for evaluatig the quality of a video after trasmissio. Compariso of PSNR results from the three differet protocols is preseted i Fig. 0(a). Further, the results for Structural SIMilarity (SSIM) idex are preseted i Fig. 0(b). SSIM is idetified to have higher sesitivity to image degradatio ad stable with huma eye perceptio whe compared to PSNR. Cosiderig both PSNR ad SSIM results, it is clear that IMSLND offers a better video quality tha that of MSLND ad FEC. This is because IMSLND hadles iterferece i the multipath trasmissio. Also, TCP is employed to guaratee the trasmissio of the I-frames, beig that they are the most importat frames i a sigle Group Of Picture (GOP). It also helps i maiteace of quality of other oticeable frames which are geerated by the predicted frames icludig P-frames ad B-frames. Furthermore, Forward Error Correctio (FEC) also has the ability to guaratees the trasmissio of I-frames, by way of replicatig the I-frames. Hece, FEC ca realize higher video quality whe compared with User Datagram Protocol (UDP). Because i UDP delivery ad retrasmissio is ot guarateed. Coversely, FEC experieces more packet loss durig video trasmissio due to a burst of trasmitted packets which is caused because of replicated packets of FEC. Cosiderig VANETs, FEC drawback is higher because of frequet chage i vehicle positio ad costraied etwork resources. The received data rate is aother metric that estimates receivig capabilities at the receivers ed. I Fig. 0(c), IMSLND has a higher received data rate whe compared with MSLND ad FEC. The simple reaso is that IMSLND cosiders iterferece at each selected ode ad also uses TCP protocol i order to esure trasmissio, which miimizes the umber of packet collisio ad cotetio, hece it reduces packet loss. Additioally, the lik quality estimated at each ode provides best ode selectio for video streamig, which leads to higher video delivery rate. The I- frames are specifically studied i the simulatio, which shows that there is higher delivery of I-frame packets compared to that of MSLND ad FEC. The delay latecy of the video trasmitted is also a vital metric i the real-time video streamig. Delay i video trasmissio is uavoidable, however, it must be withi a acceptable rage of huma eyes perceptio. The mea delay of IMSLND compared with MSLND ad FEC are preseted i Fig. 0(d). The result demostrates that IMSLND achieves slightly lower mea delay compared to the MSLND. Eve though, the delay is relatively high but is lower tha that of the MSLND. However, the high delay is caused due to the use of TCP for I-frames trasmissio. I the simulatio, it is observed that most of the delayed packets are TCP packets. It is a established fact that the major drawback of TCP is a higher delay. FEC has lower mea delay compared to that of the MSLND. Despite the high delay, IMSLND has ot exceeded the allowed maximum delay of 0. secods, which is realistic to the huma eye perceptio. Meawhile, further research could be coducted i order to miimize the mea delay. (a) (c) (d) Fig. 0. The result of simple lae sceario. (a) ad (b) represets the result of PSNR ad SSIM idex respectively, (c) ad (d) represets the result of receivig data rate ad delay respectively. ii) Dese urba sceario: The secod subsectio is the urba sceario, where a large umber of vehicles are employed ad simulated. I this sceario, the settigs of the topology are based o the map of Johor Bahru (Jala Abu Bakar). The coectio employed is based o VV commuicatio patter, hece oly ad-hoc routig is eabled. The simulatio has bee carried out by cosiderig differet umbers of vehicular odes, so as to test the performace of IMSLND for differet vehicular ode desities. The results depicted i this subsectio, are the average results of all trasmitted ad received video packet at the receiver ed. It is believed that this will provides the actual performace of IMSLND o differet vehicle desities. The results of the (b)

16 Page of video quality obtaied are preseted i Figs. (a)-(d). The Figs. (a) ad (b), demostrate that IMSLND protocol has the highest mea PSNR ad mea SSIM idex whe compared to MSLND ad FEC. Based o the simulatio results, it is observed that the quality of the video icreases as the umber of vehicles icreases from 0 to 00. Almost a stable video quality is experieced whe the umber of vehicles is betwee 00 to 00. The icreased video quality achieved is because there is a substatial umber of vehicles, which serves as a ext forwardig vehicle for the video trasmissio. I additio, it is due to the ode selectio criteria cosiderig iterferece at each ode. However, the video quality starts to degrade as the desity of vehicles is icreased from 00 to 00. This is i coectio with the icrease i the umber of video streamig request at the source vehicle, which is due to the icreased desity of vehicle i the etwork. Additioally, as stated by Xie, et al. [] that large vehicle desity causes lik saturatio due to the broadcastig of routig packets. The video quality of FEC decreases faster compared to the MSLND. However, the IMSLND attais a higher mea video quality compared to that of MSLND ad FEC i the simulatio. The receivig data rate result for the three differet protocol is preseted i Fig. (c), which is the average umber of successfully received video packet at the destiatio vehicle. The DRR is used to test ad measure the performace of IMSLND. The simulatio result demostrates that the IMSLND has the highest mea received data rate compared to MSLND ad FEC. Oe of the factors that determie video quality is the data receivig rate. The delay observed i the simulatio results for urba sceario slightly differ due to the icrease i vehicle desity, as show i Fig. (d). The delay of IMSLND protocol is still high, but slightly less tha that of the MSLND protocol. Nevertheless, the average delay obtaied does ot exceed the allowed limit of 0. secods. The delay ecoutered i the simulatio could be attributed to the itermittet discoectio of vehicles whe the vehicles are fewer ad the ature of TCP trasmissio. However, if RSUs are deployed to aid coectio ad the TCP trasmissio delay is hadled, the the delay issue i IMSLND protocol ca be improved. (a) (b) (c) (d) Fig.. The result of simple lae sceario. (a) ad (b) represets the result of PSNR ad SSIM idex compared with differet vehicle desity respectively, (c) ad (d) represets the result of receivig data rate ad delay compared with differet vehicle desity respectively. V. CONCLUSION I this paper, a iterferece-aware multipath video streamig solutio cosiderig ode disjoit ad lik disjoit (IMSLND) protocol is proposed ad simulated. The purpose of this paper is to miimize iterferece i a multipath video trasmissio i order to achieve high-quality video streamig i VANETs. The proposed protocol employs selectio of dispersed vehicles with zero or miimal route couplig i multipath trasmissio. The lik ad ode disjoit are also utilized to further ehace the dispersed vehicle selectio to achieve miimal iterferece. Further, the lik quality metrics icludig the lik sigal power ad badwidth capability of the multipath lik. I additio, mathematical formulatios are derived for dispersed vehicle selectio ad the lik quality estimatio, which is based o badwidth capacity, packet error, SNR ad received sigal power. The proposed iterferece miimizatio protocol is useful for multipath video streamig by improvig quality of video streamig. However, to further exted this paper, the future research work would focus o video streamig optimizatio cosiderig delay parameters i order to improve video quality i VANETs commuicatio. ACKNOWLEDGEMENT The research is supported by Miistry of Educatio Malaysia (MOE) ad coducted i collaboratio with Research Maagemet Ceter (RMC) at Uiversity Tekologi Malaysia (UTM) uder VOT NUMBER: RJ0000..F0. CONFLICT OF INTEREST The authors declare o coflict of iterest. REFERENCES [] A. Aliyu et al., "Cloud Computig i VANETs: Architecture, Taxoomy, ad Challeges," IETE Techical Review, pp. -, 0. [] F. Soldo, C. Casetti, C.-F. Chiasserii, ad P. Chaparro, "Streamig media distributio i VANETs," i Global Telecommuicatios Coferece, 00. IEEE GLOBECOM 00. IEEE, 00, pp. -: IEEE. [] F. Xie, K. A. Hua, W. Wag, ad Y. H. Ho, "Performace study of live video streamig over highway vehicular ad hoc etworks," i Vehicular Techology Coferece, 00. VTC-00 Fall. 00 IEEE th, 00, pp. -: IEEE.

17 Page of [] J.-S. Park, U. Lee, S. Y. Oh, M. Gerla, ad D. S. Lu, "Emergecy related video streamig i VANET usig etwork codig," i Proceedigs of the rd iteratioal workshop o Vehicular ad hoc etworks, 00, pp. 0-0: ACM. [] C. Rezede, A. Boukerche, M. Almulla, ad A. A. Loureiro, "The selective use of redudacy for video streamig over Vehicular Ad Hoc Networks," Computer Networks, vol., pp. -, 0. [] M. Xig, J. He, ad L. Cai, "Maximum-utility schedulig for multimedia trasmissio i drive-thru Iteret," IEEE Trasactios o Vehicular Techology, vol., o., pp. -, 0. [] H. Xie, A. Boukerche, ad A. A. Loureiro, "A multipath video streamig solutio for vehicular etworks with lik disjoit ad ode-disjoit," IEEE Trasactios o Parallel ad Distributed Systems, vol., o., pp. -, 0. [] J.-M. Vella ad S. Zammit, "A survey of multicastig over wireless access etworks," IEEE Commuicatios Surveys & Tutorials, vol., o., pp. -, 0. [] M.-F. Tsai, C.-K. Shieh, T.-C. Huag, ad D.-J. Deg, "Forward-lookig forward error correctio mechaism for video streamig over wireless etworks," IEEE Systems Joural, vol., o., pp. 0-, 0. [0] N. Qadri, M. Altaf, M. Fleury, ad M. Ghabari, "Robust video commuicatio over a urba VANET," Mobile Iformatio Systems, vol., o., pp. -0, 00. [] P. Bucciol, J. L. Zechielli-Martii, ad G. Vargas-Solar, "Optimized trasmissio of loss tolerat iformatio streams for real-time vehicle-tovehicle commuicatios," i Computer Sciece (ENC), 00 Mexica Iteratioal Coferece o, 00, pp. -: IEEE. [] C.-H. Li, C.-K. Shieh, ad W.-S. Hwag, "A access poit-based FEC mechaism for video trasmissio over wireless LANs," IEEE Trasactios o Multimedia, vol., o., pp. -0, 0. [] S. Zaidi, S. Bitam, ad A. Mellouk, "Ehaced user datagram protocol for video streamig i VANET," i Commuicatios (ICC), 0 IEEE Iteratioal Coferece o, 0, pp. -: IEEE. [] C. Rezede, A. Boukerche, H. S. Ramos, ad A. A. Loureiro, "A reactive ad scalable uicast solutio for video streamig over VANETs," IEEE Trasactios o Computers, vol., o., pp. -, 0. [] T. Volkert, M. Osdoba, A. Mitschele-Thiel, ad M. Becke, "Multipath video streamig based o hierarchical routig maagemet," i Advaced Iformatio Networkig ad Applicatios Workshops (WAINA), 0 th Iteratioal Coferece o, 0, pp. 0-: IEEE. [] R. Wag, M. Almulla, C. Rezede, ad A. Boukerche, "Video streamig over vehicular etworks by a multiple path solutio with error correctio," i Commuicatios (ICC), 0 IEEE Iteratioal Coferece o, 0, pp. 0-: IEEE. [] C. Rezede, A. Mammeri, A. Boukerche, ad A. A. Loureiro, "A receiverbased video dissemiatio solutio for vehicular etworks with cotet trasmissios decoupled from relay ode selectio," Ad Hoc Networks, vol., pp. -, 0. [] M. De Felice, E. Cerqueira, A. Melo, M. Gerla, F. Cuomo, ad A. Baiocchi, "A distributed beacoless routig protocol for real-time video dissemiatio i multimedia VANETs," Computer commuicatios, vol., pp. 0-, 0. [] M. Kserawi, S. Jug, D. Lee, J. Sug, ad J.-K. K. Rhee, "Multipath video real-time streamig by field-based aycast routig," IEEE Trasactios o Multimedia, vol., o., pp. -0, 0. [0] A. Aliyu, A. H. Abdullah, O. Kaiwartya, F. Ullah, U. M. Joda, ad A. N. Hassa, "Multi-Path video streamig i vehicular commuicatio: Approaches ad challeges," i Studet Project Coferece (ICT-ISPC), 0 th ICT Iteratioal, 0, pp. -: IEEE. [] O. Kaiwartya, S. Kumar, D. Lobiyal, P. K. Tiwari, A. H. Abdullah, ad A. N. Hassa, "Multiobjective dyamic vehicle routig problem ad time seed based solutio usig particle swarm optimizatio," Joural of Sesors, vol. 0, 0. [] A. Rao, S. Ratasamy, C. Papadimitriou, S. Sheker, ad I. Stoica, "Geographic routig without locatio iformatio," i Proceedigs of the th aual iteratioal coferece o Mobile computig ad etworkig, 00, pp. -0: ACM. [] O. Kaiwartya ad S. Kumar, "Guarateed geocast routig protocol for vehicular adhoc etworks i highway traffic eviromet," Wireless Persoal Commuicatios, vol., o., pp. -, 0. [] J. C. Navas ad T. Imieliski, "GeoCast geographic addressig ad routig," i Proceedigs of the rd aual ACM/IEEE iteratioal coferece o Mobile computig ad etworkig,, pp. -: ACM. [] I. Stojmeovic, A. P. Ruhil, ad D. Lobiyal, "Vorooi diagram ad covex hull based geocastig ad routig i wireless etworks," Wireless commuicatios ad mobile computig, vol., o., pp. -, 00. [] O. Kaiwartya, S. Kumar, D. Lobiyal, A. H. Abdullah, ad A. N. Hassa, "Performace improvemet i geographic routig for vehicular Ad Hoc etworks," Sesors, vol., o., pp. -, 0. [] R. S. Raw ad S. Das, "Performace aalysis of P-GEDIR protocol for vehicular ad hoc etwork i urba traffic eviromets," Wireless persoal commuicatios, pp. -, 0. [] A. N. Hassa, A. H. Abdullah, O. Kaiwartya, D. K. Sheet, ad A. Aliyu, "Geographic forwardig techiques: Limitatios ad future challeges i IVC," i Studet Project Coferece (ICT-ISPC), 0 th ICT Iteratioal, 0, pp. -: IEEE. [] N. Brahmi, M. Boussedjra, J. Mouza, ad M. Bayart, "Adaptative movemet aware routig for vehicular ad hoc etworks," i Proceedigs of the 00 Iteratioal Coferece o Wireless Commuicatios ad Mobile Computig: coectig the world wirelessly, 00, pp. 0- : ACM. [0] J. Gog, C.-Z. Xu, ad J. Holle, "Predictive directioal greedy routig i vehicular ad hoc etworks," i Distributed Computig Systems Workshops, 00. ICDCSW'0. th Iteratioal Coferece o, 00, pp. -: IEEE. [] H. Tu, L. Peg, H. Li, ad F. Liu, "GSPR-MV: A routig protocol based o motio vector for VANET," i Sigal Processig (ICSP), 0 th Iteratioal Coferece o, 0, pp. -: IEEE. [] D. Wu, S. Ci, ad H. Wag, "Cross-layer optimizatio for video summary trasmissio over wireless etworks," IEEE Joural o Selected Areas i Commuicatios, vol., o., 00. [] C. Zhu, Streamig Media Architectures, Techiques, ad Applicatios: Recet Advaces: Recet Advaces. IGI Global, 00. [] M. Asefi, J. W. Mark, ad X. S. She, "A mobility-aware ad qualitydrive retrasmissio limit adaptatio scheme for video streamig over VANETs," IEEE Trasactios o Wireless Commuicatios, vol., o., pp. -, 0. [] M. Amadeo, C. Campolo, ad A. Moliaro, "Ehacig IEEE 0. p/wave to provide ifotaimet applicatios i VANETs," Ad Hoc Networks, vol. 0, o., pp. -, 0. [] A. Bradai ad T. Ahmed, "ReViV: Selective rebroadcast mechaism for video streamig over VANET," i Vehicular Techology Coferece (VTC Sprig), 0 IEEE th, 0, pp. -: IEEE. [] H. Xie, A. Boukerche, ad A. A. Loureiro, "MERVS: A Novel Multichael Error Recovery Video Streamig Protocol for Vehicle Ad Hoc Networks," IEEE Trasactios o Vehicular Techology, vol., o., pp. -, 0. [] Y.-L. Hsieh ad K. Wag, "Dyamic overlay multicast for live multimedia streamig i urba VANETs," Computer Networks, vol., o., pp. 0-, 0. [] N. Kumar ad J. Kim, "Probabilistic trust aware data replica placemet strategy for olie video streamig applicatios i vehicular delay tolerat etworks," Mathematical ad Computer Modellig, vol., o., pp. -, 0. [0] J. Joshi, K. Jai, Y. Agarwal, M. J. Deka, ad P. Tuteja, "COMIP: Cluster based overlay ad fast hadoff mobile IP system for video streamig i VANET's," i Smart Istrumetatio, Measuremet ad Applicatios (ICSIMA), 0 IEEE rd Iteratioal Coferece o, 0, pp. -: IEEE. [] C.-M. Huag, C.-C. Yag, ad Y.-C. Li, "A adaptive video streamig system over a cooperative fleet of vehicles usig the mobile badwidth aggregatio approach," IEEE Systems Joural, vol. 0, o., pp. -, 0. [] M. Asefi, S. Céspedes, X. She, ad J. W. Mark, "A seamless qualitydrive multi-hop data delivery scheme for video streamig i urba VANET scearios," i Commuicatios (ICC), 0 IEEE Iteratioal Coferece o, 0, pp. -: IEEE. [] C. Xu, F. Zhao, J. Gua, H. Zhag, ad G.-M. Mutea, "QoE-drive user-cetric VoD services i urba multihomed PP-based vehicular etworks," IEEE Trasactios o Vehicular Techology, vol., o., pp. -, 0. [] T. A. Q. Pham, K. Piamrat, ad C. Viho, "Qoe-aware routig for video streamig over vaets," i Vehicular Techology Coferece (VTC Fall), 0 IEEE 0th, 0, pp. -: IEEE. [] C. Quadros, E. Cerqueira, A. Satos, J. Lim, ad M. Gerla, "Beaco-less video streamig maagemet for VANETs based o QoE ad likquality," i Itegrated Network Maagemet (IM), 0 IFIP/IEEE Iteratioal Symposium o, 0, pp. -: IEEE.

18 Page of [] R. Wag, C. Rezede, H. S. Ramos, R. W. Pazzi, A. Boukerche, ad A. A. Loureiro, "LIAITHON: A locatio-aware multipath video streamig scheme for urba vehicular etworks," i Computers ad Commuicatios (ISCC), 0 IEEE Symposium o, 0, pp : IEEE. [] C. Li, J. Zou, H. Xiog, ad C. W. Che, "Joit codig/routig optimizatio for distributed video sources i wireless visual sesor etworks," IEEE Trasactios o Circuits ad Systems for Video Techology, vol., o., pp. -, 0. [] J. Zou, H. Xiog, C. Li, L. Sog, Z. He, ad T. Che, "Prioritized flow optimizatio with multi-path ad etwork codig based routig for scalable multirate multicastig," IEEE Trasactios o Circuits ad Systems for Video Techology, vol., o., pp. -, 0. [] W. Sog ad W. Zhuag, "Performace aalysis of probabilistic multipath trasmissio of video streamig traffic over multi-radio wireless devices," IEEE Trasactios o Wireless Commuicatios, vol., o., pp. -, 0. [0] Z. Zhu, S. Li, ad X. Che, "Desig QoS-aware multi-path provisioig strategies for efficiet cloud-assisted SVC video streamig to heterogeeous cliets," IEEE Trasactios o Multimedia, vol., o., pp. -, 0. [] M.-F. Tsai, N. Chilamkurti, J. H. Park, ad C.-K. Shieh, "Multi-path trasmissio cotrol scheme combiig badwidth aggregatio ad packet schedulig for real-time streamig i multi-path eviromet," IET commuicatios, vol., o., pp. -, 00. [] T. Wag, L. Sog, ad Z. Ha, "Coalitioal graph games for popular cotet distributio i cogitive radio VANETs," IEEE Trasactios o Vehicular Techology, vol., o., pp. 00-0, 0. [] R. K. Schmidt, T. Köllmer, T. Leimüller, B. Böddeker, ad G. Schäfer, "Degradatio of Trasmissio Rage i VANETS caused by Iterferece," PIK-Praxis der Iformatiosverarbeitug ud Kommuikatio, vol., o., pp. -, 00. [] A. Goldsmith, Wireless commuicatios. Cambridge uiversity press, 00. [] V. Erceg et al., "A empirically based path loss model for wireless chaels i suburba eviromets," IEEE Joural o selected areas i commuicatios, vol., o., pp. 0-,. [] L. Breslau et al., "Advaces i etwork simulatio," Computer, vol., o., pp. -, 000. [] J. Klaue, B. Rathke, ad A. Wolisz, "Evalvid-a framework for video trasmissio ad quality evaluatio," Computer Performace Evaluatio/Tools, vol., pp. -, 00. [] M. Behrisch, L. Bieker, J. Erdma, ad D. Krajzewicz, "SUMO simulatio of urba mobility: a overview," i Proceedigs of SIMUL 0, The Third Iteratioal Coferece o Advaces i System Simulatio, 0: ThikMid. [] R. Dosselma ad X. D. Yag, A formal assessmet of the structural similarity idex. Uiversity of Regia. Departmet of Computer Sciece, 00. [0] A. Aliyu et al., "Towards video streamig i IoT Eviromets: Vehicular commuicatio perspective," Computer Commuicatios, 0.

19 Page of Ahmed Aliyu, is a PhD cadidate i Faculty of Computig at Uiversiti of Techologi Malaysia. He received Master s i Computer Sciece ad Techology degree i July 0 from LUT, Chia. His mai research areas are VANETs, ad Cloud Computig. Abdul Haa Abdullah (M ) received his Ph.D. degree from Asto Uiversity i Birmigham, Uited Kigdom i. He is curretly a Professor ad head of Pervasive Computig Research Group at Faculty of Computig, Uiversiti Tekologi Malaysia, Johor Bahru, Malaysia. His research iterest icludes Wireless Sesor Networks, Vehicular Adhoc Networks, ad Next Geeratio Networks. Nauma Aslam (M 0) received his Ph.D. i egieerig mathematics from Dalhousie Uiversity, Caada, i 00. He is curretly a Reader (Associate Professor) i the Departmet of Computer ad Iformatio Scieces, Northumbria Uiversity, UK. He was a assistat professor at Dalhousie Uiversity from 00 to 0. He has extesive research experiece i wireless ad-hoc ad sesor etworks. Raja Zahilah Raja Mohd. Radzi received her Dr. Eg i Electrical & Iformatio System, from Osaka Prefecture Uiversity, Japa o 0. Curretly, she is a Seior Lecturer at Departmet of Computer Sciece, Faculty of Computig, Uiversiti Tekologi Malaysia. Her research iterests iclude IP over WDM Network, Iteret of Thigs, Software Defie Networkig, Wireless Network ad Embedded System. Ayma Altameem received his PhD i Iformatio Techology from Uiversity of Bradford, UK, ad M.Sc i Iformatio Systems from Lodo South Bak Uiversity, UK. He is curretly workig as vice dea i college of applied studies ad commuity services Kig Saud Uiversity Riyadh KSA. His research iterests focus o E-commerce, Iteret of thigs, iformatio security ad artificial itelligece. Rupak Kharel (M ) received his Ph.D. degree i Computer Sciece from the Departmet of Computer ad Iformatio Scieces, Northumbria Uiversity, UK i 0. He is curretly a Seior Lecturer at the School of Egieerig, Machester Metropolita Uiversity, UK. His research iterests focus o Cyber Security orieted techical advacemets for Iteret of Thigs, Wireless Sesor Networks, ad Idustry.0. Mohammed Joda Usma, is a PhD cadidate i Faculty of Computig at Uiversiti of Techologi Malaysia. He received Master s i Computer Sciece ad Techology degree i July 0 from LUT, Chia. His mai research areas are Cloud Computig, Wireless Sesor Network, Artificial Itelligece.