On adaptive MIMO space-time processing

Abstract

In this thesis, we develop a novel adaptive MIMO space-time processing method which adapts to changes in the wireless environment for the maximization of wireless link performance. -- First, we develop a fully simulated MIMO wireless environment in Matlab® to facilitate algorithm development and performance benchmarking. Second, we develop the theory supporting an adaptive MIMO layered space-time processing method called adaptive BLAST (A-BLAS1). Thirdly, known non-adaptive diversity and spatial multiplexing layered space-time processing techniques are implemented and benchmarked under simulated MIMO channels using two element, four element, and eight element antenna arrays respectively. Fourth, we implement and benchmark an A-BLAST approach under equivalent simulated MIMO channels. Finally, simulation results are provided which demonstrate a performance improvement using A-BLAST over the non-adaptive benchmarks. -- The main contributions of this thesis include the creation of a highly parameterized MIMO simulation environment in Matlab®, as well as, the development of a novel A-BLAST layered space-time processing algorithm complemented with adaptation based on reference BER data obtained under MIMO channels of varying spatial rank using the developed simulation environment. The ABLAST code structure developed provides additional space-time codeword mappings not previously defined through traditional non-adaptive BLAST methods. These additional codeword mappings are shown to provide more granular control over the relative weighting of spatial multiplexing gain and diversity order and are better suited to a broader range of MIMO channel environments. Using estimates of the MIMO channel spatial rank and receive signal-to-noise ratio, combined with a residual bit error rate threshold, the developed adaptation algorithm is shown to be able to automatically select an A-BLAST codeword mapping from the available A-BLAST codeword set, improving MIMO link performance when compared with non-adaptive BLAST techniques which are optimized for the spatial multiplexing and diversity encoding respectively.