Maximum a Posteriori Estimation Using the EM Algorithm for Linear Codes With Low-Density Generator Matrix in Magnetic Recording Systems
Hidetoshi Saito, Masayuki Hayashi, Ryuji Kohno · IEEE Transactions on Magnetics · 2007
In high-density digital magnetic recording channels, magnetic recording systems are influenced by media noise. It is known that jitter noise is one of the principal sources of the media noise and depends on recording signal patterns. For such a jitter-noise dominant recording channel with continuous erasures, maximum a posteriori (MAP) decoding is hard to minimize the error rate performance of signal processing systems. To improve MAP decoding performance, sequence estimation using the expectation maximization (EM) algorithm is introduced to identify each distribution of the jitter noise in the normal mixture model and estimates the missing sequences which are caused by continuous erasures