Feedback information rate and entropy in a discrete system†

Abstract

A multi-state deterministic system controlled via a noisy Binary Symmetric Channel (BSC) is modelled by an equivalent Mar-kov chain. The system performs a random walk whose mean deviation from zero is numerically equal to the expected number of steps between zero crossings, and whose entropy is equal to the average amount of transmitted control information between zero crossings. Simple expressions are found for the received control information and for the mutual information via the BSC.