US 11,902,817 B2
Systems and methods for utilizing machine learning and neural network models to identify issues in wireless networks
Christian Winter, Highland Village, TX (US); Brian A. Ward, Fort Worth, TX (US); Richard S. Delk, Irmo, SC (US); and Xia Li, Sunnyvale, CA (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 Oct. 6, 2022, as Appl. No. 17/938,434.
Application 17/938,434 is a continuation of application No. 16/691,400, filed on Nov. 21, 2019, granted, now 11,477,678.
Prior Publication US 2023/0088342 A1, Mar. 23, 2023
This patent is subject to a terminal disclaimer.
Int. Cl. H04W 24/10 (2009.01); H04B 17/345 (2015.01); G06N 3/04 (2023.01); G06N 3/08 (2023.01)
CPC H04W 24/10 (2013.01) [G06N 3/04 (2013.01); G06N 3/08 (2013.01); H04B 17/345 (2015.01)] 20 Claims
OG exemplary drawing
 
1. A method, comprising:
extracting, by a device, data from input data to generate extracted data identifying a quantity of radio frequency branches and data identifying a bandwidth,
wherein the data includes time and date data, branch identification data, and interference per physical resource block number data associated with a wireless network;
creating, by the device, physical resource block images based on the extracted data;
processing, by the device, the physical resource block images to associate labels with each of the physical resource block images and to identify potential issues associated with the physical resource block images;
processing, by the device, data identifying the potential issues, with a machine learning model, to determine probability scores associated with the potential issues; and
performing, by the device, one or more actions based on a potential issue, of the potential issues, that has a greatest probability score, of the probability scores.