US 11,875,235 B2
Machine learning voltage fingerprinting for ground truth and controlled message error for message and ECU mapping
Shabbir Ahmed, Beaverton, OR (US); Christopher Gutierrez, Hillsboro, OR (US); Marcio Juliato, Portland, OR (US); Qian Wang, Portland, OR (US); Vuk Lesi, Cornelius, OR (US); and Manoj Sastry, Portland, OR (US)
Assigned to Intel Corporation, Santa Clara, CA (US)
Filed by Intel Corporation, Santa Clara, CA (US)
Filed on Sep. 17, 2020, as Appl. No. 17/024,232.
Prior Publication US 2021/0004725 A1, Jan. 7, 2021
Int. Cl. G06N 20/00 (2019.01); G06F 21/55 (2013.01); G06N 5/04 (2023.01)
CPC G06N 20/00 (2019.01) [G06F 21/554 (2013.01); G06N 5/04 (2013.01); G06F 2221/034 (2013.01)] 25 Claims
OG exemplary drawing
 
1. A computing apparatus comprising:
a processor; and
memory storing a machine learning (ML) model and instructions, the instructions when executed by the processor, configure the apparatus to:
train the ML model to infer one of a plurality of electronic control unit (ECU) labels from voltage signatures associated with a plurality of message identifications (MIDs) of messages transmitted on a bus;
determine whether a first one of the plurality of MIDs overlap at least a second one of the plurality of MIDs based on an overlap threshold, the overlap threshold to comprise a percentage value that represents an amount of time the first one of the plurality of MIDs or the at least second one of the plurality of MIDs is misclassified by the ML model;
collapse the first one of the plurality of MIDs and the at least the second one of the plurality of MIDs into a one of the plurality of ECU labels to generate an updated mapping between MIDs and ECU labels responsive to a determination that first one of the plurality of MIDs overlaps at least the second one of the plurality of MIDs; and
train the ML model on the updated mapping between the plurality of MIDs and the plurality of ECU labels.