With rapid advance of wireless communications technology, high data rates and high quality services have been continually demanded. To satisfy these requirements, a better use of available limited resources (e.g., frequency bands and transmit power) has to be promoted. One of the most promising technologies is multiple-input multiple-output (MIMO) wireless systems that use multiple antennas at both wireless links. The great potential of MIMO systems can only be realized with multiple radio-frequency chains as well as prohibitively complex baseband signal processing algorithms as compared to single antenna systems. In particular, the complexity issue at the receiver side is one of the most challenging tasks of MIMO systems due to the size and power limitations of the mobile units. Here, the transmitted data has to be detected with as small probability of error as possible.
We only focus on high rates, spatial multiplexing MIMO systems mainly viewed from complexity perspective. At the receiver, an optimum detection algorithm offers best performance, but its complexity becomes prohibitive as the increased number of transmit antennas together with higher-order modulation. To avoid huge complexity, various suboptimal algorithms with low complexity have been suggested. However, most of them suffer from a significant performance degradation compared to optimal performance. To provide better performance-complexity tradeoff, in recent years, considerable research works have been conducted to investigate and develop a near- or exact-ML performance algorithm with reduced complexity especially for spatial multiplexing MIMO systems. For example, sphere detection and its variants were introduced toward this end in many publications. However, its complexity may be still high for certain applications with limitations in size and power. Thus, if possible, it is favorable to further reduce the receiver complexity. In this dissertation, we propose and investigate an unified tr...