Decoding acceleration of low-density parity-check codes for mobile wirelessMAN

Abstract

The low-density parity-check (LDPC) codes have a relatively simple algorithm for error correction. The best performance of such codes is demonstrated for large frame sizes and a substantial number of iterations. The decoding delay depends on the number of iterations being performed with the received noisy data in order to obtain the best estimate to the transmitted data. A modification of the standard LDPC decoding algorithm is proposed, which is based on a statistical comparison of true and False convergence of an attempted decoding acceleration with the use of the Epsilon algorithm for convergence acceleration of the series of decoding iterations for the individual bits in each transmitted frame. Using the LDPC codec for Mobile WirelessMAN as an illustration, it is shown that the True convergence dominates over the False convergence which follows to a better bit error rate (BER) performance as compared to the standard (plain) scheme for a given number of iterations.