Machine Learning based Digital Beamforming for Line-of-Sight optimization in Satcom on the Move Technology

Abstract

With the evolving communication systems, the need for beamforming to improve the gain of the transmitting or receiving antenna has also increased. Beamforming allows to direct the radiated energy with the intended choice of direction efficiently. The main focus of this work is to develop an effective method for beamforming at the receiver side antennas for deploying Line-of-Sight (LOS) communication in Satellite Communication (Satcom) by using machine learning algorithms to detect signals as accurately as possible and to reduce the time taken to steer the beam as well as complexity of operations if a standard beamforming algorithm was used. To implement this, the antenna array weights are pre-calculated for a number of beam directions and kept as a database which are given to a linear regression machine learning model. The signal weights that are calculated for each array element by using their progressive measured phase difference is due to the arriving signal, that are given as input to a linear regression model and the direction of arrival (DOA) of the signal is predicted. The curve fitted linear regression model can be implemented in real-time geostationary satellite communication systems to accurately intercept the signal of interest.