Performance analysis of video on demand in an IEEE 802.11p-based vehicular network

Abstract

We consider a VoD (Video on-Demand) platform designed for vehicles traveling on a highway or other major roadway. Typically, cars or buses would subscribe to this delivery service so that their passengers get access to a catalog of movies and series stored on a back-end server. The network infrastructure comprises IEEE 802.11p RSUs (Road Side Units) that are deployed along the highway and deliver video content to traveling vehicles. In this paper, we propose a simple analytical and yet accurate solution to estimate two key performance parameters for a VoD platform: (i) the average download data rate experienced by vehicles over their journey and (ii) the average "interruption time", which corresponds to the fraction of time the video playback of a given vehicle is interrupted because of an empty buffer. Through multiple examples, we investigate the influence of several parameters (e.g., the video bit rate, the number of vehicles, the distance between RSUs, the vehicle velocity) on these two performance parameters whose outcome may help the sizing of an IEEE 802.11p-based VoD platform.

Delivering Video-on-Demand services with IEEE 802.11p to major non-urban roads: A stochastic performance analysis

“…Overall, the works that come closest to ours are the ones presented in [19,20]. In [20], the considered video streaming infrastructure is almost the same, except that the RSU coverage areas overlap whereas, in our work, they do not.…”

“…It follows that the performance model can handle virtually any number of vehicles. Furthermore, it can take into account much more vehicles than in [19]. Moreover, in this paper, we validate our proposed model on a more realistic simulator than in [19].…”

“…The evaluation is focused on the packet losses due to handovers, whereas we consider different performance evaluation parameters (namely the download rate and the interruption time). In [19], the video streaming infrastructure under study is the same as the one we consider in our work. However, the modeling approach as the validation part are very different.…”

“…However, the modeling approach as the validation part are very different. In [19], each vehicle is considered individually in the modeling making the approach hardly scalable when the number of vehicles grows. On the other hand, in the current paper, the proposed model is based on the vehicle density (possibly different for each vehicle type) naturally lending itself to a stochastic modeling.…”

“…Furthermore, it can take into account much more vehicles than in [19]. Moreover, in this paper, we validate our proposed model on a more realistic simulator than in [19]. Indeed, we use the SUMO simulator (Simulation of Urban MObility) [21] to simulate realistic vehicular trajectories that are then injected into the ns-3 simulator, whereas in [19], the proposed solution is only validated with ns-3.…”