A cross-layer protocol for cooperative content delivery over mobile ad-hoc networks
International Journal of Communication Networks and Distributed Systems
| Hi-index | 0.00 |
Peer-to-peer video streaming will enable large scale television distribution, in real time, over the Internet. In this type of system, viewers contribute their resources to the network to act as relays for the media streams and no dedicated infrastructure is required. As packets are exchanged between unreliable peers, it is particularly challenging to achieve high video quality and low end-to-end delay. The contribution of this work is to present adaptive video coding and streaming techniques which improve the performance of conventional client-server systems and extend them to enhance peer-to-peer multicast. In throughput-limited environments, network congestion precludes low-latency applications, such as interactive video streaming. We analyze this effect through a rate-distortion model which captures the impact of self-generated congestion on the decoded video quality of a stream. In addition, we introduce the concept of congestion-distortion optimized (CoDiO) packet scheduling, which determines which packets should be sent, and at which time, to achieve the highest decoded video quality while avoiding the creation of unnecessary congestion. We analyze the error-resilience properties of H.264 SP and SI frames, which provide the ability of creating flexible pre-encoded representations where the robustness can be adapted to varying network conditions. We derive the coding efficiency of these pictures and compare it to that of I and P pictures. We present an adaptive SI insertion algorithm and identify scenarios where it is beneficial compared to streaming with periodic I frames. Distortion-optimized packet scheduling is extended to the case of peer-to-peer multicast, where viewers are used as real-time relays for a media stream. We present a prioritization scheme which transmits in priority the most important video packets to the receivers which will serve, in turn, the largest set of peers. This algorithm is combined with a distortion-optimized retransmission scheduler which operates at the receivers and adapts to the video content to request the most important missing packets first. Experiments carried out over simulated networks with hundreds of peers show the advantage of adaptive scheduling for low-latency streaming.