View article

The asymptotic–in the number of antennas–theoretic capacity of a MIMO (Multiple Input Multiple Output) system is derived when considering Rice distribution entries. Assuming perfect knowledge of the channel at the receiver, analytical expressions of the capacity are derived in the case of perfect and partial (based on the mean and limiting eigenvalue distribution of the mean) knowledge at the transmitter. Remarkably, the capacity depends only on a few meaningful parameters, namely, the limiting eigenvalue distribution of the mean matrix, the signal to noise ratio (SNR), the Ricean factor, and the system load. These results show in particular that, for a given SNR and in contrast to the SISO (Single Input Single Output) case, the MIMO Rice channel does not always outperform the MIMO Rayleigh channel in terms of capacity. Moreover, the results are also useful to quantify the effect of feedback on general MIMO systems.