Detection for channels with transition noise

Transition noise is known to be a major cause of errors for high density magnetic recording. This noise is signal dependent and can be modeled as multiplicative noise in a linear channel model. The maximum-likelihood method was not considered for detection of signals in such noise in the past. In this study, a detector model for an asymptotic maximum-likelihood (AML) detection is developed for systems with such noise. Based on a linear partial response channel model, a recursive procedure is obtained as a tree search algorithm, leading to the maximum likelihood detection asymptotically, as the tree-search depth is increased. Performance estimation will be discussed in a separate paper.