Massive multiple-input multiple-output (M-MIMO) is one of the cutting edge technologies that provides significant improvement in throughput, coverage and spectral efficiency. The challenge with M-MIMO systems is to ex...
详细信息
Massive multiple-input multiple-output (M-MIMO) is one of the cutting edge technologies that provides significant improvement in throughput, coverage and spectral efficiency. The challenge with M-MIMO systems is to extract individual signals from the composite signal, thus making optimal detectors prohibitively complex. Recently, approximate message passing (AMP) and its variant detectors have gained substantial importance due to their decreased complexity and improved performance. However, AMP algorithm does not always converge. Non-linear detectors like vertical Bell-Labs layered space-iime improve the bit error rate (BER) but with high complexity, whereas, linear minimum mean square error (mmse) detector offers low complexity while compromising on BER performance. Neumann series based mmse detectors further reduce mmse computational complexity, however, the BER performance remains the same. In this work, the authors propose a hybrid Neumann series based mmse detector which decomposes the detected signal into its constituent components and apply a neighbourhood selection algorithm on the obtained components thus improving the overall performance. Another contribution of this work is derivation of an off-set value for optimised neighbourhood set selection that enables more accurate detection while further reducing computationalcomplexity. Simulation results confirm that the proposed scheme outperforms aforementioned algorithms in terms of BER performance and computationalcomplexity in a continuously changing Rayleigh channel.
暂无评论