Application of Machine Learning Techniques in Phased Array Antenna Synthesis: A Comprehensive Mini Review

Abstract

With the rapid development of modern communication systems, phased array antennas (PAAs) are widely used in many applications such as radars and 5G networks. In a PAA composed of multiple elements (antennas), beamforming or beam steering can be achieved by adjusting the phase difference in the excitation signals that feed each element of the array, eliminating the need for mechanical antenna movement. The performance quality of the communication systems heavily relies on the precise synthesis of the PAAs. PAA synthesis entails determining the geometric or physical shape of an antenna based on knowledge of its electrical parameters. Conventional methods for PAA synthesis use conventional electromagnetic models embedded in antenna design software’s. However, these models often pose challenges due to resource-intensive computations, lengthy simulation times, and potential high error rates. Machine learning (ML) techniques can be employed to optimize solutions in various telecommunication systems, including PAAs synthesis. In this article, we review and investigate the application of ML techniques in the synthesis of PAAs. The results of this study show that utilizing ML techniques can expedite the design process by threefold, while simultaneously reducing errors and increasing accuracy up to 99%.