Convexity of error probability in binary scalar detection

Abstract

This paper obtains results on the convexity of error probability in the detection of binary-valued scalar signals corrupted by additive noise. It is shown that the error probability of the maximum likelihood receiver is a convex function of the signal-to-noise ratio (SNR) when the noise has a unimodal, differentiable probability density function. This result has some interesting implications on the optimum strategies of the transmitter and the jammer, as well as the optimal use of multiple channels.<>