LDPC DECODING ALGORITHM WITH ESTIMATION OF NOISE VARIANCE

As the code length approaches infinite, LDPC codes have the potential of approaching Shannon limits. The original message passing (sum-product) algorithms for AWGN channel require the knowledge of noise variance. If the assumed noise variance deviates from the true value, the performance tumbled. In this paper, an iterative noise variance estimation algorithm using the decoded data is presented. The true noise variance is substituted by the estimated one. The computational load is analyzed. Simulation results demonstrate the effectiveness of this method