Cross-layer design for robust video delivery over unreliable networks.

摘要

Video communications over unreliable networks such as wireless networks has remained a challenging problem due to the complex structure of bit streams generated by today's advanced video coders, and potential limitations on bandwidth and the time-varying nature of wireless channels. Video coding has advanced significantly in the recent years, achieving incredibly high compression efficiencies. Nevertheless, high compression ratio is not sufficient to guarantee good video quality in such environments as bit streams become more vulnerable to errors propagations. This dissertation addresses the problem of efficient video transmission over various types of packet lossy networks.;Efficient transport of video over unreliable links can be achieved by intelligent exploitation of the available resources as well as source and network error control techniques. These source and network parameters are jointly considered in a cross-layer optimization framework in order to maximize the end users' experienced video quality. First, we consider the transport of a single-layer video stream over wireless channels by utilizing intra-refreshment, multi-reference motion compensated prediction, channel feedback, and forward error correction (FEC). Then, we investigate transmission of scalable video streams for various different applications. An accurate distortion model for the scalable extension of the H.264/AVC standard (SVC) is presented. This model allows for fast evaluation of the impact of various parts of the bit stream on the quality of the video signal. Such a model plays an essential role in any distortion-aware cross-layer optimization framework. Utilizing this model, based on the specific available resources, for each application, we propose a content-aware optimization framework to unequally allocate resources from multiple layers in order to enhance the end-quality of the video sequence. More specifically, the following scenarios are considered: (1) Source only bit extraction at intermediate bit rates, also referred to as rate adaptation; (2) Joint source and channel rate adaptation where packets to be transmitted are selected jointly with their optimal error protection rate; (3) Content-aware packet scheduling and resource allocation for multi-user downlink streaming over wireless 3G/4G networks; (4) Content-aware, foresighted resource reciprocation strategies for media streaming over peer-to-peer (P2P) networks. The envisioned P2P network consists of autonomous and self-interested peers trying to maximizes their individual utilities. The resource reciprocation among such peers is modeled as a stochastic game and peers determine the optimal strategies for resource reciprocation using a Markov Decision Process (MDP) framework. Unlike existing solutions, this framework takes the content and characteristics of the video signal into account by introducing an artificial currency in order to maximize the video quality in the entire network.