Error-resilient video streaming over lossy channels

This thesis provides solutions for error-resilient video transmission over unreliable networks. The proposed solutions are categorized into three groups: (1) those introduced at the source and channel encoder to make the bitstream more resilient to potential errors; (2) those invoked at the decoder upon detection of errors to conceal the effect of errors; and (3) those which require interactions between the codec and the network so that the encoder can adapt its operation based on time-varying network conditions.; To make a compressed bitstream resilient to transmission errors, we propose a simple bit-plane-wise unequal error protection algorithm that assigns an unequal amount of forward error correction (FEC) to each bit-plane. It requires considerably less computation than previously published algorithms by using a dynamic programming approach, which is important if the coder is to adapt to rapid changes in the channel. To get additional error resilience at high packet loss rates, we propose two multiple description coding methods for wavelet-based coders that generate two independent substreams. Each substream is protected by our proposed FEC-based unequal error protection algorithm and transmitted over different paths. Any information that is lost during transmission for any substream is estimated at the receiver by using our proposed error concealment algorithm. The proposed error concealment algorithm estimates the lost information for one substream by using the correlation between the substreams and the smoothness of the video signal.; To minimize the overall distortion and exploit the benefits of path diversity, we propose a distributed video streaming framework using unbalanced multiple description coding and unequal error protection. In the proposed system, two senders simultaneously stream complementary substreams to a single receiver over different paths. Our proposed rate allocation algorithms adjust the optimal source coding rates and the amount of channel coding for the two senders in a coordinated fashion. The proposed unbalanced MDC method generates two unbalanced substreams according to the available source coding bit rates for the two senders. Finally, we compare the error-resilience capabilities of layered coding and multiple description coding techniques through extensive experimentation, and explore the environments where each performs best.