Downlink multi-user mimo systems

In this paper, precoding methods used in downlink multi-user, multi-input multioutput systems (downlink MU-MIMO) for Rayleigh channels with perfect and imperfect channel knowledge are investigated. Precoding methods provide inter-user interference management and are designed to maximize the achievable sum rate. Downlink MU-MIMO systems can be underloaded, critically loaded or overloaded depending on the number of antennas at the base station and users and the number of users. This study provides a comparison between the generalized power iteration precoding (GPIP) (which is newly proposed in the literature) with the minimum mean square error precoding (MMSE) and rate splitting (RS). It is observed that for channels with perfect knowledge, the GPIP method results in similar or worse sum rate performance than the MMSE method in all load cases. Moreover, the MMSE method has lower complexity. For channels with imperfect knowledge, GPIP has better performance than MMSE method, but worse performance than RS method and performs faster than the MMSE in all different load cases. Keywords: multi-user multiple input multiple output systems (MU-MIMO), generalized power iteration precoding (GPIP), minimum mean square error precoding (MMSE), semi-orthogonal user selection zero forcing precoding (SUSZF), rate splitting(RS)