Research On Distributed Source Coding Based On LDPC Codes

Distributed Source Coding (DSC) is an information compression coding technique. It can be used to compress the interrelated sources that are related but can not communicate to each other. This information compressing method has great potential applicative value in the fields like sensor networks whose encoder is energy-limited, as it is simple in encoding, and consumes little energy.This thesis, supported by National Natural Science Foundation of China"The Research of Key TTechnology in Noncoherent MIMO Communication SystemT", is focused on the Distributed Source Coding. The majority of existing Distributed Source Coding algorithms use the knowledge of channel coding. It is proved that the high compressing performance can be obtained when using the channel coding, the performance of which is good, to design the source coding algorithm. According to the existing algorithms and the theory of coset partition, this thesis proposes the Distributed Source Coding encoding algorithm based on two structured Low-Density Parity-Check (LDPC) codes. The two kinds of LDPC codes are Irregular Repeat Accumulate LDPC codes based on Q matrix and Quasi-Cyclic LDPC codes from IEEE802.16e standard. Besides, the bit-flipping iterative decoding algorithm used in Distributed Source Coding is proposed. The existed Tbelief-propagation iterative decoding algorithm which can be used in the Distributed Source Coding is of high complexity. The bit-flipping decoding algorithm has low computing and realizing complexities with simple checksum not complex probability calculations. On account of the decoding algorithm above, the thesis also gives the simulation results in the two models of different source relation. The relation in the two models is represented by TBinary Symmetric Channel Tand Additive White Gaussian Noise Channel. Finally, the thesis analyzes the algorithm of combining the Distributed Source Coding and the Network Coding.