Capacity of a Burst-Noise Channel

Abstract

A model of a burst-noise binary channel uses a Markov chain with two states G and B. In state G, transmission is error-free. In state B, the channel has only probability h of transmitting a digit correctly. For suitably small values of the probabilities, p, P of the B $rA G and G $rA B transitions, the model simulates burst-noise channels. Probability formulas relate the parameters p, P, h to easily measured statistics and provide run distributions for comparison with experimental measurements. The capacity C of the model channel exceeds the capacity C(sym. bin.) of a memoryless symmetric binary channel with the same error probability. However, the difference is slight for some values of h, p, P; then, time-division encoding schemes may be fairly efficient.