Signal Detection Algorithms For MIMO Communication Systems

Signal detection for multiple-input multiple-output (MIMO) communication system is an important technique in this system. This thesis studies V-BLAST (Vertical Bell laboratories LAyered Space-Time) algorithm and sphere decoding algorithm for MIMO signal detection, and equalization algorithm for MIMO channel time-variation suppression. The main contributions of the thesis are as follows:1. Soft-decision V-BLAST algorithm. This algorithm introduces soft decision into the each layer's detection of the traditional V-BLAST algorithm, and reserves all results with sufficiently large posterior probabilities in this detection, and consequently greatly improves the error rate performance of the system. The theoretical analysis indicates that by selecting proper posterior probability threshold, this algorithm can achieve an error rate performance near to sphere decoding.2. A reduced-complexity V-BLAST detection algorithm. In each stage of the V-BLAST detection, this algorithm selects all layers with sufficiently large SNRs/SINRs in addition to the layer with the largest SNR/SINR, and therefore significantly reduces the preprocessing computational complexity of the V-BLAST detection.3. An MMSE-solution-based method for initial radius selection of sphere decoding. This method not only makes the selected radius small enough, but also guarantees at least one lattice point inside this sphere, and thus significantly reduces the computational complexity of the sphere decoding at low or intermediate SNR systems.4. A simplified V-BLAST algorithm for MIMO OFDM system. This algorithm utilizes the obtained parameters in the V-BLAST detection of some sub-channels to compute those parameters of their neighboring sub-channels, and thus greatly decreases the computational complexity for the V-BLAST detection in MIMO OFDM system.5. A decorrelating bootstrap equalizer for MIMO channel time-variation suppression. This equalizer makes decorrelating equalizations for the output signals of the MIMO ordinary equalizer based on the statistical properties of the source signals, and just uses its own inputs to update its parameters. As a result, it effectively suppresses the co-channel interferences produced by the time-variation of the MIMO channel at a very low computational cost.