Novel Linear and Non-linear Multi-user Mimo Downlink Precoding with Improved Diversity and Capacity

Space division multiple access (SDMA) promises high gains in the system throughput of wireless multiple antenna systems. If SDMA is used on the downlink of a multi-user multiple-input multiple-output (MIMO) system, either long-term or shortterm channel state information has to be available at the base station (BS) to facilitate the joint precoding of the signals intended for the different users. Precoding is used to efficiently eliminate or suppress multi-user interference (MUI). It also allows us to perform most of the complex processing at the BS which leads to a simplification of the mobile terminals. In this paper we introduce two novel precoding techniques (a linear and a non-linear one) that improve the system capacity and the diversity order. The new non-linear precoding technique combines successive minimum-mean-squared error (SMMSE) and Tomlinson-Harashima precoding (THP) to provide a higher capacity and diversity. But it is very sensitive to channel estimation errors. The new linear precoding technique is called regularized block diagonalization (RBD). Although it has a smaller capacity than SMMES THP it achieves a higher diversity order than SMMSE THP. At high data rates it can extract the maximum diversity order of the channel. RBD is less sensitive to channel estimation errors than non-linear techniques and can be also used for precoding with long-term channel state information. Index Terms MIMO precoding, Tomlinson-Harashima precoding, MMSE precoding.