INFORMATION CAPACITY OF CHANNELS WITH PARTIALLY UNKNOWN NOISE .1. FINITE-DIMENSIONAL CHANNELS

Information capacity is considered for a finite-dimensional communicat ion channel in which the noise is the sum of a known Gaussian noise an d an independent component with unknown statistical properties, The pr oblem is modeled as a zero-sum two-person game between the coder and a jammer, with mutual information as the payoff function. The jammer do es not control the ambient Gaussian noise. The constraints employed ar e appropriate for extension to infinite-dimensional channels and for c onsideration of coding capacity. The original problem is reformulated to prove existence of both a saddle value and a saddle point, The sadd le value and a saddle point are determined, along with the jammer's mi nimax strategy.