Efficient SOR Based Massive MIMO Detection Using Chebyshev Acceleration

Abstract

Massive multiple-input multiple-output (MIMO) is considered as one of the key techniques in today's 5G wireless communication systems. Though massive MIMO can achieve higher data rate and spectral efficiency compared with small-scale MIMO, its high complexity becomes a problem when hundreds of antennas are equipped. For uplink massive MIMO detection, conventional schemes like zero forcing (ZF) and minimum mean square error (MMSE) are prohibitive due to the unaffordable complexity. To this end, successive over relaxation (SOR) detection is proposed, which iteratively approaches the performance of MMSE with much lower complexity. However, the performance of SOR is often unsatisfactory enough especially in some ill channel conditions. For a better compromise between performance and complexity, Chebyshev-SOR detection is proposed in this paper. Using Chebyshev acceleration, the proposed method can achieve faster convergence and better performance than conventional SOR, especially in ill channel conditions. Numerical results with different channel conditions are given in detail, which show that Chebyshev-SOR method achieves 5dB gain with little complexity overhead. Computational complexity comparison is also given in this paper.