Research On Coding And Decoding Algorithms Based On LDPC Codes For Synchronization Channels

Channels with synchronization errors due to a mismatch of the transmitter andreceiver clocks are of great interest in practical systems. The coding technique undersynchronization channels is one of the most attractive research fields in channel coding.LDPC(Low-Density Parity-Check) codes was been widely used in synchronizationchannels, as it has strong error-correction capability and low complexity of decoding.Based on the existing research on LDPC codes, this paper mainly focus on theerror-correction of synchronization channels. Some results are obtained and summarizedas follow:1. LDPC codes and Tanner graph model are introduced. The message passingdecoding algorithm on BEC(Binary Erasure Channel) and AWGN(Additive WhiteGaussian Noise) channels are given. The current research of LDPC codes andsynchronization channels are briefly summarized.2. The characteristics of synchronization channels are analyzed. An effectiveconcatenated coding scheme is given. The resynchronization of symbols is obtained bycalculating the maximum-a-posteriori using the forward-backward algorithm. Asymbol-level synchronization algorithm and decoding scheme are given. The simulationresults show that the symbol-level synchronization algorithm has a lower error rate thanbit-level synchronization algorithm.3. Numerical issues in floating-point simulations of LDPC decoders are examined.The reasons on these numerical issues are analyzed. A non-saturating sum-productalgorithm is given by the forward-backward algorithm. Simulation results show that thenon-saturating sum-product algorithm has a better performance than the originalsum-product algorithm.