| Video transmission over error-prone networks has to deal with the transmission errors on lossy channels. This thesis proposes a new error resilient video coding technique to address this challenge. In addition, the other two problems, transport scheme design and joint source-channel coding, which are associated with the new video coding technique, are also studied.; A new multiple description video coding system is proposed. This system employs a second-order motion predictor for motion compensation. Furthermore, the mismatch between the prediction frames at the encoder and that at the decoder is explicitly coded. Combining the second-order predictor and the mismatch coding can achieve a wide range of trade-off between coding efficiency and error resilience.; We further investigate the problem of coding control in the proposed video encoder. Rate-distortion models of this coder are developed and an optimal control scheme are established, which can select the coding parameters adaptively according to the loss rates and the signal characteristics.; We propose to combine multistream source coding with multi-path transport for video transmission over ad-hoc networks. We propose three schemes and study them via extensive simulations. The results show that significant improvement in video quality can be achieved over the standard scheme.; Finally, we study the joint source channel coding problem for video communications over Multiple-Input Multiple-Output (MIMO) wireless systems by comparing the performances of several typical source-channel coding combinations through both theoretical and simulation studies. We show that no single combination is the best for the entire range of channel conditions, rather different combinations may be best for different sub-ranges. |
