Polar Coding for Information Regular Processes

Polar codes use an explicit channel model, so the precise nature of the communication channel determines if and when polarization occurs. In practice, communication channels typically have memory which affects the probability of error in nearby symbols. In this paper, we extend existing results to show polarization for the more general family of information regular processes and define a growth rate for promptness. We then show that information regular processes with linear promptness polarize and provide an example with practical applications for which the rate of polarization is very slow.