Parametric counterfeit IC detection via Support Vector Machines

Abstract

We present a method to detect a common type of counterfeit Integrated Circuits (ICs), namely used ones, from their brand new counterparts using Support Vector Machines (SVMs). In particular, we demonstrate that we can train a one-class SVM classifier using only a distribution of process variation-affected brand new devices, but without prior information regarding the impact of transistor aging on the IC behavior, to accurately distinguish between these two classes based on simple parametric measurements. We demonstrate effectiveness of the proposed method using a set of actual fabricated devices which have been subjected to burn-in test, in order to mimic the impact of aging degradation over time, and we discuss the limitations and the potential extensions of this approach.