US 12,067,487 B2
Method and apparatus employing distributed sensing and deep learning for dynamic spectrum access and spectrum sharing
Stanley Vitebsky, Millburn, NJ (US); and Marc J. Beacken, Randolph, NJ (US)
Assigned to CACI, Inc.—Federal, Reston, VA (US)
Filed by CACI, Inc.—Federal, Arlington, VA (US)
Filed on Apr. 21, 2020, as Appl. No. 16/854,289.
Prior Publication US 2021/0326695 A1, Oct. 21, 2021
Int. Cl. G06N 3/08 (2023.01); H04L 43/0888 (2022.01); H04W 16/00 (2009.01); H04W 16/02 (2009.01); H04W 24/06 (2009.01); H04W 24/08 (2009.01); H04W 28/22 (2009.01)
CPC G06N 3/08 (2013.01) [H04L 43/0888 (2013.01); H04W 16/00 (2013.01); H04W 16/02 (2013.01); H04W 24/06 (2013.01); H04W 24/08 (2013.01); H04W 28/22 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method of training a neural network via deep reinforcement learning (DRL) comprising the steps of:
receiving, via the neural network, a policy from a third party;
receiving, via the neural network, features of plural telecommunication groups located in an RF network;
observing, via the neural network, a graphical representation of the received features of the plural telecommunication groups in the RF network;
assigning, based on the observation, one of the plural telecommunication groups to one of plural channels in the RF network;
determining, via the neural network, a change in throughput of the RF network based on the assignment; and
adjusting, based on the determined change in throughput, the policy received from the third party via the DRL,
wherein the features include pixel intensity based upon one or more of transmit power, signal to noise ratio levels and measured interference.