Study On Iterative Decoding

Channel coding is a necessary technique for reliable communication over noisy channels, while the decoding algorithm decides its application in practical communication systems. A new coding method is always followed by a new decoding algorithm or a renewal of the old one. In the recent ten years, Turbo code has proved itself a great success for good error-correcting performance. The iterative decoding, on the same time, evokes the rediscovery of LDPC codes. Actually, turbo iteration not only works for Turbo, LDPC codes and other concatenated systems, but also can be applied as a principle to many fields like channel equalization, modulation, multi-user detection and so on.This thesis investigates some aspects of Turbo iterative decoding, with emphasis on its application in LDPC decoding and two specific serially concatenated systems. The main contents are as follows:The turbo iterative decoding algorithm is introduced, and the extrinsic information discussed. A common soft-input soft-output flow chart is presented, as well as the BER performance of a serially concatenated system using turbo decoding.The Turbo iterative decoding algorithm of LDPC codes, TDMP algorithm, is introduced, and the relationship between the the interleaver and the parity-check matrix is discussed. To generate extrinsic information, the constitutional decoder of TDMP algorithm may adopt several algorithms, including the BCJR algorithm, SPA algorithm and their simplified versions, based on which a new decoding algorithm, TDMP Normalized MSA algorithm is proposed. The new algorithm is testified on computer for the LDPC coded in DMB-T standard.An LDPC code and an RS code are concatenated according to the STiMi channel coding scheme. Different ways for extrinsic information transfer are considered between the two decodes, and simulation results are presented specifically.Further research work includes:1. For the LDPC code and RS code concatenation system, whether it will work better if the RS decoder adopts the soft-input soft-output decoding algorithm.2. The turbo iterative decoding sacrifices computational convenience for better performance. Thus, complexity reduced algorithms with good performance deserve further research.