Analysis and results on optimized MIMO detection with threshold-based adaptive cancellation

We present an analysis of, and improved results for, a recently presented algorithm for detection of the received signal of an encoded wireless MIMO system. The method is based on a BLAST algorithm with successive interference cancellation, which is now applied in an adaptive manner based on some threshold decision. Therefore, reliability information on bits is extended to symbol reliability. The principle idea of this method is not restricted to specific receiver designs, but can be applied to any receiver with symbol interference cancellation. We use the threshold criterion to improve cancellation and present three methods to obtain the optimum threshold. One is an exact analytical approach; the second is based on an EXIT chart technique; the third is a rule of thumb stemming from some simulation observations. To have a fair comparison, we also introduce a so called genie-threshold receiver, analogous to the known genie detector. As a result, we obtain an SNR gain of up to 1 dB. Furthermore, by applying this scheme to a very low complexity iterative structure, we achieve a turbo cliff for a full rate 6×6 16-QAM system at 8.3 dB, which is, to our knowledge, unreached by MIMO systems with comparable low complexity.