QoE-driven dissemination of real-time videos over vehicular networks

Live video dissemination over Vehicular Ad Hoc Networks (VANETs) is fundamental for several services, e.g., roadside video emergency, advertisement's broadcast, and smart video surveillance. All these applications face many challenges due to stringent video quality level requirements, dynamic topologies, and broadcast environments. To cope with these challenges, as well as reduce network routing overhead, geographic Statistical Routing Protocols (SRPs) have been proposed as a suitable solution for the distribution of video flows in VANETs, usually by using positioning and Quality of Service (QoS) parameters. However, rather than only these parameters, a satisfactory video dissemination from the user's perspective also requires video and human-awareness issues. In real situations, due to different requirements and hierarchical structures of multimedia applications, network level and position parameters alone are not enough to select the best relay nodes and build up reliable backbones to multi-hop video dissemination with satisfactory reachability and Quality of Experience (QoE) levels. This article focuses on improving the disseminated quality of on-road live videos in VANETs. Thus, we propose the cross-layer Qoe-driven REceiver-based (QORE) mechanism, which is modularly coupled to SRPs to offer QoE-aware and video-related parameters for the relay node selection and backbone maintenance. Thus, nodes decide for themselves to retransmit further the video sequences, enhancing the capacity of the system in delivering videos with better QoE assurance. On top of this, an application-level Error-Control (EC) scheme, namely Interleaving, allows mitigating the effects of frame loss by spreading out the bursty losses. We added QORE to a straightforward SRP built using the Distance method, named DQORE protocol. Results show the gains of DQORE compared to SRPs, achieving video dissemination with QoE support, less routing overhead, and satisfactory reachability, in V2V topologies. (C) 2016 Elsevier B.V. All rights reserved.