Linear Precoding Based on Maximum SJNR in Multi-user MIMO Downlink

What's the capacity of the multi-user MIMO downlink and how to get the capacity are the hot research topics in the latest years.The BD (Block Diagonalization) algorithm,based on the constraint which the interferences between different users are zero,have two drawbacks.One is the capacity loss because the signal spaces of different users are overlapped,the other is that it is required that the number of the access point's antennas must larger than the sum of the hum of users' antennas.This paper presents two linear precoding algorithms.One is the optimal algorithm based on the maximization of SJNR,the other is a modified suboptimal algorithm which can improve the downlink capacity.By capacity analysis and simulation,the performances of the two al- gorithms both are better than BD algorithm.