Performance of low density parity check (LDPC) codes with bootstrap decoding algorithm on a fast fading channel

Abstract

Low density parity check (LDPC) codes can be decoded in various ways, namely the bit-flipping (BF) algorithm, the weighted BF (WBF) algorithm and the log-likelihood ratio belief propagation (LLR-BP) algorithm, and so on. These algorithms provide a wide range of tradeoffs among decoding complexity, decoding speed, and error rate performance. Recently, the bootstrapped WBF (BWBF) algorithm has been proposed to improve the error rate performance and the decoding complexity of the WBF algorithm. In this paper, we study the effects of the predetermined threshold /spl alpha/ of the bootstrap step on BER for various LDPC codes. Furthermore, we apply the bootstrap step to the LLR-BP algorithm. We evaluate the BWBF and the bootstrapped LLR-BP (B-LLR-BP) algorithms on a fast fading channel. We show that the optimal threshold depends on the row and column weights of the LDPC codes. We also show that the BWBF algorithm provides a large improvements in both the error rate performance and the decoding complexity on a fast fading channel. Moreover, we show that the LLR-BP algorithm hardly receives the effect of the bootstrap step.