A Lower Bound on the Information Rate of Intersymbol Interference Channels Based on the Polyphase Decomposition

We consider the discrete-time intersymbol interference (ISI) channel model with additive white Gaussian noise and input symbols independently drawn from discrete alphabets. We provide a provable lower bound for the achievable rate. In contrast to available lower bounds, the proposed one does not require the input signal to be stationary. Therefore, it can be employed to analyze the information rate of ISI channels with independent non-identically distributed symbols. This analysis reveals that when the channel input is restricted to discrete input alphabets, the throughput can be improved by using non-identically distributed input symbols as opposed to Gaussian input case. This presents a new opportunity for the transmission adaptation for communication through ISI channels.