Optimization And Implementation Of The Nonlinear MIMO Detector In Ultra High-Speed Wireless LAN

MIMO technology can significantly increase the channel capacity andthe spectral efficiency of wireless communication system withoutsacrificing additional bandwidth, which makes it the key technology toenhance the next-generation wireless system’s throughput towards severalhundreds of Mbps or even Gbps. The signal detection at the receiver isdeterminant of the performance and the practicality of MIMO system. TheMaximum Likelihood detection algorithm has the best performance, but itscomplexity grows exponentially with the antenna and constellation size.The linear detection has low complexity, but its performance is not ideal.So how to design a high-performance low-complexity MIMO detector is achallenging task now.Considering the detection performance data throughput and areacomprehensively and based on the conventional K-Best detector, thisthesis designs and implements a nonlinear soft-output MIMO detectorwhich supports1×12×23×3and4×4antenna array èBPSK QPSK16-QAM and64-QAM modulation. The proposed detector is integrated tothe Ultra High-Speed Wireless LAN. MMSE-SQRD channel processingand utilization of discarded paths are adopted to improve performance.Interleaved parallel sorting is used to enhance throughput. On-demandexpansion scheme and abandoning discarded paths in lower levels candecrease the complexity.Simulation result shows that the proposed K-Best detector of K=10inparallel sorting scheme has the same performance with the one of K=8inserial sorting scheme. With the same channel processing, the proposeddetector has1dB performance improvement in BER=10-4compared with the conventional K-Best detector of K=8. After synthesis of SMIC0.13Pmlibrary, the proposed detector can achieve the throughput of1.56Gbps,consuming796K gates in area. Compared with other soft-output K-Bestdetector, this work has an advantage of better performance and higher datathroughput.