Non-Coherent Rayleigh Fading Mimo Channels: Capacity And Optimal Input

Information transfer over a discrete time uncorrelated Rayleigh fading multiple input multiple output (MIMO) channel is considered, where neither the transmitter nor the receiver has the knowledge of the channel state information (CSI) except the fading statistics. We derive a capacity supremum with the receive antenna number at any signal to noise ratio (SNR) using Lagrange optimisation. We show that the asymptotic capacity is double logarithmic when the input power is large. We prove that to achieve the capacity, the amplitude of the multiple input needs to have a discrete distribution with a finite number of mass points, one of them necessarily located at the origin. We show how to compute the capacity numerically in multi-antenna configuration at any SNR with the discrete input using the Kuhn-Tucker condition for optimality. Furthermore, we show that the capacity with two mass points is optimal at low SNR signifying on-off keying. As the number of receive antennas increases, the maximum SNR at which two mass points are optimal decreases.