Artificial neural network model training method, frequency uniform multi-beam generation method, and computer readable storage medium storing instructions to perform artificial neural network model training method and frequency uniform multi-beam generation method

Abstract

An artificial neural network model training method is proposed. The method may include preparing training data including digital intermediate frequency data based on a signal received from an antenna array of a multiple input multiple output (MIMO) system as input data, and at least one of a target beam pattern and a target weight vector as output data. The method may also include training an artificial neural network model for wideband frequency uniform multi-beam generation using the training data. A weight vector of the artificial neural network model may be recursively trained using at least one of a target beam pattern-based loss function and a target weight vector-based loss function necessary to generate desired multiple beams.