Abstract
In this paper, a family of cost-efficient hard-output detection algorithms for large multiple-input multiple-output (MIMO) systems is proposed. The schemes employ punctured QR decomposition (QRD) instead of regular QRD to reduce complexity. The bit error rate performance is studied analytically, where it is shown that channel matrix puncturing does not affect the diversity gain of the detectors. Through empirical simulations, the proposed schemes are shown to achieve significant reductions in computational complexity with graceful performance degradation. In particular, at an SNR cost of 4 dB, 77% of complex multiplications in nulling and cancellation are saved in 16x16 MIMO, while 30% of multiplications are saved at a 2 dB cost in 4x4 MIMO. The savings can reach 94% in 64x64 MIMO.