On the information-theoretic capacity of magnetic recording systems in the presence of medium noise

The compound behavior of the magnetic recording channel is modeled by combining the Lorentzian read-back pulse, the microtrack channel model, and additive white Gaussian noise (AWGN). By noting that at the output of this model the read-back signal is cyclostationary, the average autocorrelation function and corresponding power spectral density over one period are computed. The average power spectral density is then used to characterize the capacity of the magnetic recording channel for various linear density and medium noise scenarios by using the conjectured Shamai-Laroia lower bound. It is shown that from a capacity point of view, medium noise is better in certain cases than AWGN.