A new joint source and channel coding scheme for packet-based scalable multimedia streams

In this paper, we develop a new technique for partitioning of source and channel coding bits for packet-based scalable streams and a general unequal error protection channel coding scheme. Unlike the previous algorithms, we regard coded source streams as a collection of different subsets with different importance and adaptively select some of them to transmit with optimal channel coding subject to channel status. Both packets extraction strategy and forward error correction coding(FEC) assignment are considered in the joint source channel coding problem. We propose a double hill climbing method and two simplified algorithms to solve the mixed optimization problem. To validate the algorithms, we exploit the scalability of H.264 scalable video coding (SVC) and derive a simple yet accurate loss distortion model for simulation. The results show that the new algorithms with extraction strategy result in less expected distortion than others with all source packets transmitted. It is also shown that the simplified algorithms with very low complexity are also appealing and can achieve near-optimal performance.