Neural-based model of spiral antenna radiation patterns for detection of angle of arrival

Neural networks have been continually growing in popularity and have been utilized in many new fields and applications. The neural network technique can replace the traditional 'look-up table' in electronic support (ES) receivers [1]. In this paper, neural networks were implemented to model the desired spiral antenna radiation patterns. These spiral antennas are used in ES payload for ultra wideband (UWB) applications. It is not possible or practical to measure spiral antenna radiation patterns at all the radiation angles due to the limitation of time. Thus, in this paper we present a neural network that can model the measured antenna patterns of two adjacent ultra-wide bandwidth spiral antennas. Thus the amplitude and phase of the spiral antenna radiation field at a given radiation direction can be predicted based on the knowledge of polarization, frequency, elevation angle (theta), and azimuth angle (phi). Using polarization, frequency, and the ratio of adjacent antenna received powers as inputs, a neural network was built to predict the angle of arrival (AOA) of an incoming wave.