Performance analysis of a distributed video coding system — Application to broadcasting over an error-prone channel

In this paper, we present a novel approach for determining the theoretical compression bound for Wyner-Ziv coded video sequences transmitted over error-prone channels. This estimation takes into account the amount of motion inside each particular frame, as well as the transmission channel conditions. We compare the analytical results with those of a practical Wyner-Ziv coding system where a turbo-decoder acts as a joint decoder for recovering the compressed information and correcting errors due to transmission channel impairments. Furthermore, we show how the entropy limits can be used for estimating the optimal amount of information to be transmitted per frame, in case of a broadcasting application where a feedback channel is not available. Simulation results show a small degradation in the system rate-distortion performance, compared to the case of source-channel coding based on feedback information from the receiver.