一种新颖的多用户MIMO系统下行链路的信道分解方法

块对角化(BD, Block-Diagonalization)算法将多用户MIMO系统下行链路信道分解为若干个独立并行的单用户MIMO信道。该算法并没有考虑各用户接收噪声功率的差异，而是通过各等效单用户MIMO信道为不同的用户提供相同的空间发射分集，因此具有最大噪声功率的“最差”用户将会严重的恶化整个多用户MIMO系统的性能。我们结合下行链路信道块三角化（BT, Block-Triangularization)算法以及Tomlinson-Harashima 非线性预编码 (THP)提出了一种下行链路的改进BD分解算法，使分解得到的各单用户信道具有优于BD算法的空间发射分集。并且通过优先对“最差”用户进行预编码使得该用户获得“最优”的等效信道，提升了整个系统的性能。

A Novel Decompose Method for the Downlink of Multi-User MIMO Systems

The BD(Block-Diagnalization) method decomposes the Multi-User(MU) MIMO downlink channel into multiple parallel independent single-user MIMO channels. Each equivalent single-user (SU) MIMO channel tends to provide the corresponding user a same spatial transmit diversity without considering the differences between the noise powers received by different users. Consequently, the“ worst”user with the largest noise power would degrade the performance of the whole system severely. As a combination of the Tomlinson-Harashima Precoding (THP) and the Block-Triangularization(BT) algorithm, We introduce a novel algorithm to get the equivalent channels with larger spatial transmit diversity than BD algorithm. Additionally, by designating the “worst” user to be first precoded, we allocate the “best” channel to the “worst” user and boost the performance of the system.