A Near-Optimal and Efficiently Parallelizable Detector for Multiple-Input Multiple-Output Wireless Systems

Abstract

Broadband Wireless Communications and Multiple-Input Multiple-Output (MIMO) systems have been the focus of much research over the past decade. A variety of MIMO detection algorithms have been proposed for detecting the data signals from the multiple received and demodulated baseband signals. Among the detectors, sphere decoding algorithms are known to be near-optimal but they are relatively complicated and have variable detection latencies, and are therefore inconvenient to implement. Also, the variable latency of most sphere decoder algorithms makes them difficult to implement efficiently on parallel hardware. This thesis work evaluates several alternative MIMO detectors and proposes a near-optimal and efficiently parallelizable detector. The new MIMO detector has much lower computational complexity than a sphere decoder, and has a convenient parallel structure comprising multiple instances of the Vertical Bell Laboratories Layered Space Time (V-BLAST) MIMO detection scheme.