MIMO Multi-User Detection In Time-Varying Channels

There has been considerable recent interest in Multi-Input Multi-Output (MIMO) system, due to its potential for significant gain in channel capacity. The Vertical Bell Laboratories Layered Space-Time (V-BLAST) system, as an important branch of MIMO space-time coding techniques, can provide high capacity over rich scattering channels. The spectral efficiencies of the V-BLAST can reach 20 to 40 bits/s/Hz at average SNRs ranging from 24 to 34 dB.However, the conventional detection algorithms for V-BLAST are not suitable for time-varying channels due to the high complexity. In this thesis, the classical V-BLAST detection algorithms, such as Zero Forcing (ZF) algorithm and Minimum Mean Square Error (MMSE) algorithm, are discussed in detail. Based on the MMSE detection, we propose an adaptive detection algorithm, which combines Maximum Likelihood (ML) channel estimator and Least Mean Squares (LMS) algorithm together. The proposed algorithm is suitable for flat slow time-varying channels. The computational complexity of the receiver is lower and it is easy to implement.More recently, attention has been shifted from MIMO single user (MIMO-SU) systems to MIMO multi-user (MIMO-MU) systems. In MIMO-MU systems, the uplink can be considered as a V-BLAST system, while the downlink detection is much more difficult. In this thesis, we discuss the downlink transmit preprocessing technique in detail, especially the linear Block Diagonalization (BD) scheme which decomposes the MIMO-MU channel into parallel MIMO-SU channels to eliminate the inter-user interference for Spatial Division Multiple Access (SMDA). The proposed adaptive structure is used in the base station of the MIMO-MU system. The V-BLAST adaptive receiver and the transmit preprocessing technique are jointly designed to realize a base station transceiver architecture for MIMO-MU detection over time-varying channels.