Maximum a Posteriori Decoding System Using an ARMA Process for a Partial Response Channel with Burst Errors

Inhalt:
One of new idea of maximum a posteriori (MAP) decoding with an autoregressive and moving-average (ARMA) process is introduced into the iterative (recursive) decoding processes between MAP decoding and turbo or sum-product decoding in a perpendicular magnetic recording channel. In this paper, MAP decoding with linear prediction is called MAP-ARMA decoding. The estimation techniques based on multivariate time series analysis, stochastic models and processes are effective for burst errors or continuous erasures. It is shown that this estimation problem for missing data values (observations) is solved by maximum likelihood estimation for an ARMA process with missing values by using multivariate statistical techniques. For the given (4608,4097) low-density parity check (LDPC) coding system, the error rate performance using proposed channel predicting and MAP decoding systems in the channel with both a burst of length 70 or less and random errors attains the almost same performance of conventional MAP decoding in the channel with only random errors.