CPC H04B 7/0695 (2013.01) [G06N 20/00 (2019.01); H04B 7/0452 (2013.01)] | 18 Claims |
1. A method, comprising:
selecting multi-user multiple input multiple output candidate beams using one or more deep neural networks;
selecting, by a single-stage deep neural network beam selector, the candidate beams simultaneously through a single pass of the deep neural network; and
selecting paired users based on the selected beams,
wherein the one or more deep neural networks are trained to maximize a multi-user priority metric or a heuristic of the multi-user priority metric.
|