Performance analysis of maximum likelihood detector for non-orthogonal multiuser signals

The maximum likelihood (ML) detector is an optimum detector for non-orthogonal multiuser signals since its probability of error is the minimum. Due to its complexity, performance analysis of the ML detector has relied heavily on approximation bounds obtained by the technique of error sequence decomposition. In this paper, exact formulas for the error probabilities of the ML detector are developed. The exact solutions are in the form of multi-dimensional integrals, and it is possible to numerically compute them. Simulation examples are used to verify the theoretical results.