Detection of Anomalous Integrated Circuits Using a Cavity Resonator System With Machine Learning

This paper proposes a novel method to detect counterfeit and anomalous integrated circuits (ICs). The method involves comparing IC signatures that are obtained by placing the IC inside the cavity resonator and measuring the scattering parameters (S-parameters) with a Vector Network Analyzer (VNA). The S-parameters are collected between 10 GHz and 19 GHz. The data from the signatures are then preprocessed and split into a training and test dataset. The training dataset is used to train a support vector machine (SVM) machine learning algorithm implemented in the Python programming language. The test dataset is used to test the algorithm’s accuracy. Measurements and tests are done on ICs of several different packaging types, technologies, and authenticities. The results indicate that the method distinguishes counterfeit ICs from authentic ICs in all cases tested; however, is not as accurate in distinguishing similar authentic ICs. It is likely that some of the false positives amongst the authentic ICs are ICs that contain manufacturing defects.