Massive Multiple-Input Multiple-Output Antenna Architecture for Multiband 5G Adaptive Beamforming Applications

Enahoro, S, Ekpo, S ORCID: https://orcid.org/0000-0001-9219-3759 and Gibson, A ORCID: https://orcid.org/0000-0003-2874-5816 (2022) Massive Multiple-Input Multiple-Output Antenna Architecture for Multiband 5G Adaptive Beamforming Applications. In: 2022 IEEE 22nd Annual Wireless and Microwave Technology Conference (WAMICON), 27 April 2022 - 28 April 2022, Clearwater, Florida, USA.

Preview

Accepted Version Download (1MB) | Preview

Abstract

One major challenge to full 5G systems deployment especially in mm-Wave band is the poor signal propagation. One approach to mitigate this effect is the use of new 5G technologies such as massive MIMO, adaptive beamforming, reconfigurable antennas etc. which can enhance the performance of the system. Adaptive beamforming algorithm uses advance digital signal processing techniques to generate main beams in the direction of interest while placing nulls in interfering signals direction to reduce interference. The beams are formed in the receiver rather in free space. It is therefore very crucial to develop an algorithm that can optimize the system to improve performance by generating signals at a faster convergence rate.In this paper, the performance analysis of various adaptive beamforming systems for 5G applications are presented using various LMS algorithms including a novel sign-leaky LMS algorithm. A uniform linear array antenna of varying element configurations, inter-element spacing, varying step-size, direction of arrival angles of the desired signals are analysed using various algorithms to determine the optimum performance of the systems. Simulation result shows that the convergence rate is highly enhanced, with the proposed algorithm converging with at least 5 iterations less than conventional LMS algorithm, while reducing interference effects by placing deeper nulls in interfering signal direction of arrivals using the proposed beamforming algorithm. There is also at least-2dB drop in normalized power of the sidelobe level compared to the LMS algorithm.