Quantization Error Correction Schemes for Lattice-Reduction Aided MIMO Detectors

Abstract

Lattice reduction aided (LRA) linear detectors have been known to achieve near optimal performance at lowcomplexity. However, one weakness of LRA detector is that the quantization step in LRA detector is not optimal. Basedon simulation results, we show that most of detection errors in LRA linear detectors are due to quantization errors. Wethen propose two methods to correct the quantization errors. In the first method, sphere detectors are introduced to correctquantization errors at low additional complexity. As a second approach, we propose a list quantization scheme which cangenerate a list of candidate symbols from the original LRA estimated symbols. From these listed symbols, decisions aremade according to the minimum Euclidean distance between the received and estimated points. It is shown by simulationsthat both methods provide significant BER performance improvements with only a small additional complexity.