Design of Optimal Unitary Constellation with Noncoherent ML Receiver

In this paper, we are interested in a noncoherent MIMO system with a single transmitter having two antennas communicating with a receiver having at least two antennas over a Rayleigh fading channel. It is assumed that the channel coefficients keep constant during four successive time slots, after which they become new independent value. For such a system, we propose a novel parameterized coding method to systematically and efficiently design an optimal unitary constellation by optimizing both noncoherent diversity and coding gains with the maximum likelihood (ML) receiver. We first characterize the optimal structure for each parameter when the number of bits allocated for each parameter space is fixed and then, find an optimal bit allocation for each parameter space subject to a total transmission bits constraint by further maximizing the coding gain. Computer simulations demonstrate that our proposed coding scheme significantly outperforms the traditional methods in literature for the same noncoherent system.