Joint source-channel coding approach to transport of digital video on lossy networks

The use of forward error-control (FEC) coding, possibly in conjuction with automatic repeat request (ARCS) techniques, has emerged as a promising approach for real-time video transport over ATM networks for cell-loss recovery. The use of an FEC-based transport protocol over networks with bandwidth constraints, however, is not obvious since FEC coding introduces overhead which can possibly cause additional cell losses which, furthermore, tend to occur in bursts.; An information-theoretic performance evaluation methodology to investigate the efficacy of FEC coding in bandwidth constraint networks is developed which is useful in predicting ultimate performance bounds and as a guide in designing practical systems. This is achieved by solving separately the channel-coding problem for the modeled transmission channel and the queueing problem for the network. The transport channel is modeled as a block interference channel (BIC) and the multiplexer as single server, deterministic service, finite buffer supporting N users. Based upon an information-theoretic characterization of the BIC and queueing theoretic results on the buffer overflow probability, the described methodology provides theoretically achievable upper limits on the number of sources multiplexed.; A joint source-channel coding (JSCC) approach for an end-to-end rate-distortion analysis of video transport over ATM-based networks is developed. This JSCC approach is developed for a single-layer encoder and later extended for a multi-layer encoder. The performance of an eminently practical class of JSCC schemes employing an MPEG-2 source encoder and RS channel codes is also presented. The important role played by passive error recovery (PER) techniques is also demonstrated when used either with uncoded systems or together with FEC-based JSCC approaches.; The information-theoretic predictions show that significant gains in the performance can be achieved with FEC coding by judiciously choosing the channel coding rate. The performance of specific coding techniques is shown to approach the information-theoretic predictions with increasing levels of complexity.