US 12,438,905 B1
Anomaly detection in copper networks
Victor C. Le, Vancouver, WA (US)
Assigned to FRONTIER COMMUNICATIONS HOLDINGS, LLC, Dallas, TX (US)
Filed by FRONTIER COMMUNICATIONS HOLDINGS, LLC, Dallas, TX (US)
Filed on Dec. 30, 2024, as Appl. No. 19/005,816.
Int. Cl. H04L 9/40 (2022.01); H04L 12/28 (2006.01); H04L 41/16 (2022.01)
CPC H04L 63/1433 (2013.01) [H04L 12/2878 (2013.01); H04L 41/16 (2013.01)] 27 Claims
OG exemplary drawing
 
1. A computer-implemented method for detecting anomalies occurring in a copper network, the method comprising:
obtaining, by one or more processors, historical data associated with a copper network, the historical data including data indicative of a multiplicity of historical copper network equipments, a multiplicity of historical operating behaviors of the copper network which have occurred, and a multiplicity of historical events corresponding to the copper network which have occurred;
extracting, by the one or more processors and from the historical data, a plurality of copper network equipment features, each copper network equipment feature indicative of a respective behavior of a respective copper network equipment;
transforming, by the one or more processors, at least one extracted copper network equipment feature, thereby generating at least one additional copper network equipment feature;
training, by the one or more processors, a machine learning (ML) model on the extracted plurality of copper network equipment features and the at least one additional copper network equipment feature to discover one or more historical behaviors of the copper network, the one or more historical behaviors being indicative of the copper network operating within a target operating range, and the training including:
validating the ML model, the validating including determining a set of excess mass curves of the ML model; and
tuning one or more hyperparameters corresponding to the validating of the ML model, the tuning including adjusting the one or more hyperparameters based on the set of excess mass curves;
detecting, by the one or more processors based on current data of the copper network and by utilizing the trained ML model, one or more anomalies occurring in the copper network; and
initiating, by the one or more processors and responsive to the detecting, a mitigating action for the detected one or more anomalies occurring in the copper network.