On the achievable information rates of finite state ISI channels

Abstract
In this paper, we present two simple Monte Carlo methods for estimating the achievable information rates of general finite state channels. Both methods require only the ability to sim- ulate the channel with an a posteriori probability (APP) detector matched to the channel. The first method estimates the mutual in- formation rate between the input random process and the output random process, provided that both processes are stationary and ergodic. When the inputs are i.i.d. equiprobable, this rate is known as the Symmetric Information Rate (SIR). The second method esti- mates the achievable information rate of an explicit coding system which interleaves m independent codes onto the channel and em- ploys multistage decoding. For practical values of m, numerical results show that this system nearly achieves the SIR. Both meth- ods are applied to the class of partial response channels commonly used in magnetic recording.

This publication has 8 references indexed in Scilit: