US 12,189,742 B2
Machine learning fingerprinting of wireless signals and related systems, methods, and computer-readable media
Benjamin C. Gookin, Denver, CO (US); David Domico, Aurora, CO (US); Justin Kopacz, Aurora, CO (US); Matthew D. Popovich, Denver, CO (US); Michelle Jin, Santa Monica, CA (US); and Nathanael M. Harmon, Colorado Springs, CO (US)
Assigned to Northrup Grumman Systems Corporation, Falls Church, VA (US)
Filed by Northrop Grumman Systems Corporation, Falls Church, VA (US)
Filed on Feb. 7, 2022, as Appl. No. 17/666,353.
Prior Publication US 2023/0252118 A1, Aug. 10, 2023
Int. Cl. G06F 21/32 (2013.01); G06N 3/084 (2023.01); G06V 10/44 (2022.01); G06V 40/12 (2022.01); H04W 12/062 (2021.01); H04W 12/10 (2021.01); H04W 12/69 (2021.01)
CPC G06F 21/32 (2013.01) [G06N 3/084 (2013.01); G06V 10/454 (2022.01); G06V 40/12 (2022.01); H04W 12/062 (2021.01); H04W 12/10 (2013.01); H04W 12/69 (2021.01)] 19 Claims
OG exemplary drawing
 
1. A device fingerprinting system, comprising:
one or more servers configured to store, in a database, fingerprint data and known identification codes associated with a plurality of identified transmitting devices;
a receiver configured to receive, from an unidentified transmitting device, wireless communications; and
a machine learning computing device configured to:
provide a preamble portion of each of the wireless communications as an input to a previously trained convolutional neural network (CNN) configured to perform few-shot learning techniques to determine a device fingerprint based, at least in part, on the preamble portion;
capture features in the preamble portion of each of the wireless communications via the CNN; and
average the captured features;
determine a device fingerprint responsive to the wireless communications using the CNN; and
compare the determined device fingerprint to the stored fingerprint data in the database.