Performance estimation for maximum-likelihood 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. A maximum-likelihood algorithm was considered for detection of signals in such noise. In this work, the performance of the detector, based on this algorithm, is compared to the traditional Viterbi algorithm (VA) and a modified Viterbi algorithm (MVA) by computer simulations. Results show an improvement of up to 5 dB In signal-to-noise-ratio (SNR) under typical conditions with a reasonable complexity.