Blind Adaptive Beamforming of Narrowband Signals using an Uncalibrated Antenna-Array by Machine Learning

Multiple wireless communication systems compete within the same frequency range. The spatial domain adds an extra degree of freedom to the time and frequency domain. In this paper we introduce a machine learning model in the spatial domain designed to enhance the signal of interest and mitigate interfering signals. The method described does not require a calibrated antenna-array-RF-chain. It is based on a neural network that learns the structure of the signal of interest and separates it from other signals in the time, frequency and spatial domains. It does so without the need of any additional information and can very easily be used as a ballast to increase the Signal to Interference and Noise Ratio (SINR).