Efficient hybrid Neumann series based MMSE assisted detection for 5G and beyond massive MIMO systems. Issue 22 (26th February 2021)

• Record Type: Journal Article
• Authors: Khurshid, Kiran; Imran, Muhammad; Khan, Adnan Ahmed; Rashid, Imran; Siddiqui, Haroon
• Is Part Of: IET communications. Volume 14:Issue 22(2020)
• Page Start: 4142
• Page End: 4151
• Publication Date: 2021-02-26
• Subjects: computational complexity -- least mean squares methods -- error statistics -- message passing -- MIMO communication -- Rayleigh channels -- 5G mobile communication -- signal detection; efficient hybrid Neumann series -- MMSE assisted detection -- massive MIMO systems -- cutting edge technologies -- spectral efficiency -- M‐MIMO systems -- individual signals -- composite signal -- optimal detectors -- approximate message passing -- variant detectors -- substantial importance -- decreased complexity -- AMP algorithm -- nonlinear detectors -- vertical Bell‐Labs layered space‐iime -- bit error rate -- linear minimum mean square error detector -- BER performance -- Neumann series based MMSE detectors -- MMSE computational complexity -- hybrid Neumann series based MMSE detector -- detected signal -- neighbourhood selection algorithm -- optimised neighbourhood set selection -- reducing computational complexity; Telecommunication systems -- Periodicals; Speech processing systems -- Periodicals; 621.38205
• DOI: 10.1049/iet-com.2020.0670
• Languages: English
• ISSNs: 1751-8628

Abstract

Abstract : 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 computational complexity. Simulation results confirm that the proposed scheme outperforms aforementioned algorithms in terms of BER performance and computational complexity in a continuously changing Rayleigh channel.