Design of low complexity detection algorithms for uplink massive MIMO systems

Abstract

Massive multiple-input multiple-output (mMIMO) wireless communication systems play a crucial role in realizing the demand for higher data rates and improved service quality for 5G and beyond communications. The potential benefits of mMIMO are further enhanced by incorporating media-based modulation (MBM). Consequently, reliable detection of transmitted information bits from all the users is one of the challenging tasks for the practical implementation of mMIMO systems. The conventional linear detectors such as zero forcing (ZF) and minimum mean square error (MMSE) achieve nearoptimal bit error rate (BER) performance for high system loading factors. However, ZF and MMSE require large dimensional matrix inversion, which induces high computational complexity for symbol detection in such systems. Furthermore, due to the constellation diversity and the consequent sparse nature of symbol vectors in mMIMO with MBM (MBM-mMIMO), those linear detectors’ performance drastically degrades. It motivates for devising alternate low-complexity near-optimal detection algorithms for uplink mMIMO systems. In this thesis, with this motivation, different low complexity detection algorithms are studied, and promising solutions are proposed for symbol detection in uplink mMIMO systems. In the first part of the thesis, state-of-the-art detection algorithms for large MIMO systems are investigated. Due to implementation tractability and the requirement of less computational load, evolutionary algorithms are observed to outperform several conventional detection algorithms for large MIMO systems. After investigating the potential drawbacks of existing evolutionary algorithms, a stochastic evolutionary algorithm is proposed for uplink symbol detection in large MIMO systems. However, when the number of antennas scales up in the system, state-of-the-art conventional and evolutionary algorithms are incapable of achieving near-optimal performance in mMIMO systems. Moreover, detection algorithms suitable for large MIMO systems are incapable of benefitting from the hardening nature of mMIMO channel. Hence, in the second part of the thesis, iterative algorithms are investigated for mMIMO systems. As the existing algorithms yield near-optimal performance for a mMIMO system with only high system loading factors, their performance degrades with increased users’ numbers. Hence, to improve the drawbacks of existing iterative algorithms, two algorithms based on nonstationary and pseudo stationary iterations are proposed in this thesis. Two mechanisms called quality ordering and reliability feedback are also introduced to improve existing detectors’ performance for mMIMO systems. A deep unfolded sparse refinement model is also proposed for low complexity symbol detection in uplink mMIMO systems. Finally, considering the benefits of MBM, MBM-mMIMO systems are considered. A graph-theoretical approach and minimum support recovery criteria based detection algorithms are proposed for low complexity symbol detection in MBM-mMIMO systems. Simulation results show that the proposed algorithms significantly outperform recently reported large MIMO, mMIMO and MBM-mMIMO detection techniques in terms of BER performance. Convergences of the proposed algorithms are also theoretically analyzed. Computational complexities of the proposed algorithms are substantially lower as compared against existing state-of-the-art algorithms for achieving the same BER performance. It indicates that the proposed algorithms exhibit a desirable trade-off between the complexity and the performance for mMIMO systems and are viable candidates for uplink symbol detection in 5G and beyond wireless communications.