Performance Analysis of Hybrid MMSE/Max-log-MAP MIMO Detector

Ever increasing demand of higher data rates is pushing Multiple-Input Multiple-Output (MIMO) systems into the ongoing standardization of future wireless systems, where this added spatial dimension will be predominantly employed for spatially multiplexed transmission. In order to make these systems practically realizable, low complexity efficient detection algorithms are eagerly sought, both by the industry and the academia. In this paper, we propose a two-stage hybrid detector based on the classical performance-complexity trade off of the MIMO systems. This detector is characterized by the lower complexity of linear detection and superior performance of maximum likelihood (ML) detection. The first stage is a linear filter that adapts the system dimension in accordance with the processing capability of the second stage, which is a non-linear low complexity ML detector. We consider the pragmatic ranges of MIMO systems (2x2, 3x3, 4x4 and 5x5) and look at the effect of the distribution of streams amongst the two stages. Adaptability of the system in the form of stream allocation to the two stages allows having a trade off between the complexity and the performance of detection. This is dictated by the alteration in the diversity order of the system as we vary number of streams between the two stages. Performance of the proposed hybrid detector improves as the number of streams to be suppressed in the first stage decreases but it is coupled with the increase in the cost of complexity. The observations made through this analysis can facilitate the design of the future systems in accordance with the limitations imposed by the hardware constraints.