Research On Distributed Source Coding Algorithms And Applications

Distributed source coding (DSC) refers to the compression of the outputs of two or more physically separated sources. The catch is that the sources do not communicate with each other (hence distributed coding). These sources send their compressed outputs to a central point for joint decoding. This encoding method greatly reduces the workload of encoding nodes, and the decoder take the redundancy information between nodes into account, improve the coding efficiency. And the complexity is partly transferred to the decoder from the encoder. So DSC is able to adapt to the encoding capacity-constrained scenarios, such as wireless sensor network. This thesis focuses on the researches on distributed source coding algorithms and applications.This paper discusses the distributed source coding using LDPC. Using AWGN channel to imitate the correlation between the sources, we compress the source with the syndrome encoding method, then decode it by the way of bit reversion with side information. We analyze the performance of Distributed Source Coding using LDPC with the emulation results. Then LDPC is introduced into the PRISM (Power-efficient, Robust, high-compression, Syndrome-based Multimedia coding) scheme in this paper. Contrast to the hash module in PRISM, this paper proposes a hash-based motion search scheme to get the most possible side-information before syndrome decoding. And this proposal scheme not only lowers the bits of hash in encoder, but also greatly lowers the computation complexity and delay in decoder.