PRML performance estimation considering medium noise down track correlations

Application of partial-response maximum-likelihood (PRML) sequence detection in magnetic recording has been demonstrated with substantial increase in the linear density. This method achieves optimal performance when the noise is white and Gaussian. In a recording channel dominated by medium noise, the noise is correlated due to inter-granular interactions. The characteristics of the noise strongly affect the PRML performance. The signal dependent medium noise is characterized approximately by the cross track average magnetization in the presence of transitions. Error rates for detecting a single bit, a dibit, and a tribit with medium noise correlations included are calculated. It is found that errors in detecting a dibit or a tribit are mainly induced by the magnetic noise addition of adjacent transitions