EMFORCED: EM-based Fingerprinting Framework for Counterfeit Detection with Demonstration on Remarked and Cloned ICs

Today’s globalized electronics supply chain is prone to counterfeit chip proliferation. Existing techniques to detect counterfeit integrated circuits (ICs) are limited by relatively high cost, lengthy inspection time, destructive nature, and restriction to a pre-packaging environment. We propose a novel method of counterfeit IC detection which takes advantage of design-specific electromagnetic (EM) fingerprints generated by simulating on-chip clock distribution networks. Through exploitation of the chip’s physical characteristics, our technique can help detect foundry of origin. We validate our approach on 8051 microcontrollers from three different vendors and utilize principal component analysis to distinguish the acquisitions by vendor. Our results show that near-field EM measurements combined with unsupervised machine learning provide ≈ 99% accuracy in counterfeit detection through design-specific fingerprint classification.

当今的全球化电子供应链容易伪造芯片扩散，以检测伪造的综合电路（ICS）受到相对较高的成本，冗长的检查时间，破坏性的性质和限制，我们提出了一种新颖的方法。通过模拟片上时钟分配网络生成的伪造IC检测，它利用了特定于设计的电子（EM）指纹。通过剥削芯片的物理特征，我们的技术可以帮助检测原产地，我们对8051个微控制器的方法与三个不同的供应商进行了验证，并利用主成分分析来区分我们的结果。通过无监督的机器学习提供了≈99％的精度，可通过设计特定的指纹伪造检测分类。