Towards Rate-decoder Complexity Optimisation in Turbo-coder based Distributed Video Coding

Conventional hybrid video coding such as H.264 is compared to turbo-coder based Distributed Video Coding (DVC) from a complexity point of view. It is shown here that the overall workload in DVC can exceed that of H.264 by a substantial amount. Hence, DVC has the advantage of low-complexity encoding but at the price of high-complexity decoding, exceeding the encoder complexity of H.264. Given a turbo-coder based DVC setup, this work introduces a method to vary and possibly optimise decoder complexity while keeping encoder burden fixed at a low level. Using operational curves of the channel coding tools used in the DVC, a relation between bitrate and decoder complexity is given. It is demonstrated that in certain regions of these operational curves, large reductions in computations can be traded against relatively small increases in bitrate. Moreover, a variation of channel coder memory has influence on decoder complexity. Together with expected prediction error-rates, this permits to select an optimal operation point for given overall constraints.