US 11,792,664 B2
Virtualized architecture for system parameter identification and network component configuration with reinforcement learning
Said Soulhi, Boston, MA (US); and Bryan Christopher Larish, Westfield, NJ (US)
Assigned to Verizon Patent and Licensing Inc., Basking Ridge, NJ (US)
Filed by Verizon Patent and Licensing Inc., Basking Ridge, NJ (US)
Filed on Jun. 16, 2021, as Appl. No. 17/349,455.
Prior Publication US 2022/0408284 A1, Dec. 22, 2022
Int. Cl. H04W 24/06 (2009.01); H04W 24/02 (2009.01); H04B 17/391 (2015.01)
CPC H04W 24/06 (2013.01) [H04B 17/3912 (2015.01); H04W 24/02 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
for each system parameter combination corresponding to system parameters of network components of a communication network:
generating a state comprising the system parameter combination, a traffic model, and a channel assignment;
executing a network traffic scenario through a virtualized testbed using the state, wherein the virtualized testbed emulates communication over a virtualized communication network between emulated communication devices and emulated network components based upon the system parameter combination, the traffic model, and the channel assignment; and
generating a reward for the system parameter combination based upon key performance indicators output by the network traffic scenario; and
configuring a network component of the communication network based upon a select system parameter combination selected based upon a reward policy and rewards of the system parameter combinations.