Performance evaluation of low-density parity-check codes on partial-response channels using density evolution

A method to evaluate the performance of a low-density parity-check (LDPC) code on partial-response (PR) channels in terms of the noise threshold and decoding error is presented. Given a particular codeword or assuming an independent and uniformly distributed (i.u.d.) codeword for transmission, the density-evolution algorithm is used to compute the probability density function of messages passing in the decoding process, from which the decoding error is extracted. This estimated i.u.d. decoding error is used to approximate the decoding error of an ensemble of LDPC codes on arbitrary PR channels. Comparison with simulation results shows that it is a very good approximation for the simulated codes, provided their length is large enough.