Machine Learning-Based Beamforming in K-User MISO Interference Channels

In this paper, we consider a K-user multiple-input and single-output interference channel propose machine learning-based beamforming design. To circumvent the difficulties of design in channels, scheme that combines two well-known schemes, which are maximum ratio transmission zero-forcing, with one finite combining factors. However, problem is still NP-hard, requires brute-force search for optimal beamforming, so adopt learning to find vectors. Our exploits deep neural network structure, whose input nodes take vectors transmit power, while output return factors transmitters' beamforming. numerical results show our proposed achieves sum rate more than 99% best numerically found search.