Implementation of a direction-of-arrival finding antenna using neural networks

Summary form only given. The purpose of this paper is to present the results of applications of neural networks for direction-finding systems. Direction-finding systems for radio signals are mostly used in mobile communications and avionics applications for antenna tracking or navigation. Such systems require, generally, accurate calibration may be sensitive to external noise and interference and operate only for a limited range of azimuth angles. We present the performance of a system consisting of a 4-element antenna-array, a simple RF signal beamformer and a neural network. The direction of arrival problem is approached as a mapping which can be modeled using a suitable neural network trained with input output pairs. We investigate the response of the system for a full range of both azimuth and elevation angles, as well as the error caused by training limitations or other factors.