Abstract:

In the block diagonalization （BD） algorithm, the difference in received noise power between users is ignored. In order to solve this problem, a novel channel-decomposing algorithm for downlink is proposed by combining the block triangularization （BT） of downlink channel and the nonlinear Tomlinson-Harashima precoding （THP）. By using the proposed algorithm, the downlink channel can be decomposed into several single equivalent channels with different spatial transmission diversities, and the ＂worst＂ user can obtain the ＂best＂ equivalent channel via precoding. Simulated results show that the proposed algorithm has better performance than the BD algorithm.

Key words: multi-user system, multiple-input multiple-output system, downlink, channel decomposition