Summary

International Conference on Emerging Technologies for Communications

2020

Session Number:SD3

Session:

Number:SD3-2

Low-complexity Large MIMO detection via Gaussian Belief Propagation

Takumi Takahashi,  Shinsuke Ibi,  Seiichi Sampei,  

pp.-

Publication Date:2020/12/2

Online ISSN:2188-5079

DOI:10.34385/proc.63.SD3-2

PDF download

PayPerView

Summary:
In the uplink of large multi-user multi-input multi-output (MIMO) systems, a Bayesian iterative detection scheme via Gaussian belief propagation (GaBP) has attached great attention as a promising signal detector with low-complexity and high-accuracy. However, the BP-based detectors successfully converge to the Bayes-optimal solution in the large-system limit only when the measurement matrices are Gaussian with zero mean. In other words, its convergence property is severely degraded in practical MIMO configurations assuming spatially correlated fading channels and an insufficient number of channel dimensions. This paper introduces some methodologies to compensate for this BP-specific drawback, such as belief damping, belief scaling, and node selection, and clarify the effectiveness and scope in terms of bit error rate (BER) performances through computer simulations. Finally, as a part of the latest trends, the development of BP-based detectors with the aid of machine learning is briefly presented.