LDPC encoder identification in time-varying flat-fading channels

This paper tackles the low-density parity-check (LDPC) encoder identification problem encountered in the time-varying flat-fading channels which are modeled as finite-state Markov chains. The in-phase and quadrature-phase components of the channel coefficients are both represented by a number of states. To greatly simplify the computation of the channel observation probabilities, each channel-state region is further quantized to an interior point. Based on our proposed finite-state Markov model, the Viterbi algorithm is thus invoked to blindly estimate the unknown channel-state sequence from each received signal segment. To mitigate the phase ambiguity which is inherent in the channel-state estimation process, the pilot-aided channel estimation method is also proposed here. The LDPC encoder is finally identified in the framework of the log-likelihood ratio of syndrome a posteriori probability. The performance of our proposed LDPC identification scheme is investigated for different normalized Doppler rates and different mechanisms to reconstruct the channel-state information. Monte Carlo simulation results suggest that pilot symbols are necessary for leading to a satisfactory identification performance for time-varying flat-fading channels.