Rates and empirical properties of good codes for partial response channels

The objective of this work is to study the properties of good codes for the partial response channel. It is known that to achieve the channel capacity C, a non-linear error correcting code is required. In Kavcic, A (2001), the channel input process was modeled as a Markov random process, and an efficient algorithm was developed to optimize the Markov input state transition probabilities. This optimized Markov input distribution induces a channel capacity lower bound [Arnold, D et al., 2001 and Pfister, HD et al., 2001], which was shown in S. Yang et al. (2005) to be very tight. We here address the desired code rates and power spectral density of the optimized channel input Markov distribution.