An Artificial Neural Network Approach to Doa Estimation and Switched Beamforming in Rectangular Array Based Smart Antennas

Abstract

Switched beamforming using electronic phase shifters is commonplace. Digital switched beamformers offer a premise of better performance than electronic phase shift switched beamformers. It is also worth noting that current unknown signal Direction of Arrival (DoA) estimation methods (commonly MUltiple SIgnal Classification (MUSIC) and Estimation of Signal Parameters via Rotational Invariance Techniques (ESPRIT)) are generally computationally intensive. In this paper, signal DoA estimation and digital switched beamforming using aptly designed Artificial Neural Network (ANN) classifiers are looked into. Initially, signals detected at a rectangular receiving array are mapped onto a DoA through an ANN classifier. A second ANN classifier maps the selected DoA onto an optimal set of beamforming weights leading to an optimal switched beamforming reception pattern. The ANN classifiers' performance in DoA estimation and beamforming is tested over a variety of trials, yielding good results. The designed ANN beamformer premises to yield high-speed and accurate switched beamforming performance, most notably in large array systems. The ANN DoA estimator/beamformer can be easily adapted to nonuniform arrays wherein closed form DoA estimation/beamforming solutions are impractical. MATLAB software environment has been used as the main analysis tool.

An efficient direction‐of‐arrival estimation of multipath signals with impulsive noise using satin bowerbird optimization‐based deep learning neural network

“…Robert Macharia et al (Macharia et al, 2019) presented signal DOA estimation along with digital switched beamforming (BF) grounded on artificial neural network (ANN) classifiers. First, via an ANN, the signals identified at a rectangular receiving array were mapped on a DOA.…”