A novel parallel detection architecture for modular MIMO receivers

Recent advances in information theory show that, given the right propagation environment, an arbitrarily large channel capacity can be obtained if multielement antenna arrays are used at both the transmitter and the receiver. In such schema, parallel detection is advantageous, since it is faster than sequential detection, it does not require power ordering at the receiver and it reduces the system latency. Previous work has proposed a parallel detector (PD) based on successive symbols approximation. Parallel detectors found in the literature suffer from high computational complexity. We first propose a modified parallel detection (MPD) algorithm that reduces the system computational complexity considerably. The algorithm can be realized using a parallel modular architecture, which consists of a set of elementary sub-detectors. The data flow between the sub-detectors is significantly reduced. The MPD also allows performance improvements when it is used as a refinement to sequential vertical Bell Labs layered space-time algorithm (V-BLAST). However, both PD and MPD algorithms do not generally perform as well as sequential V-BLAST. Thus, secondly, we suggest another parallel detection architecture. This is based on a minimum mean square error (MMSE) criterion in its first stage and on weighted parallel interference cancellation in its second stage. Simulation results show that our parallel detector performs better than sequential V-BLAST for the same order of complexity.