Unsupervised Recycled FPGA Detection Based on Direct Density Ratio Estimation

With the expansion of the semiconductor supply chain, recycled field-programmable gate arrays (FPGAs) have become a serious concern. Several methods for detecting recycled FPGAs by analyzing the ring oscillator (RO) frequencies have been proposed; however, most assume the presence of known fresh FPGAs (KFFs) as the training data used for machine-learning-based classification, which is an impractical assumption. In this study, we propose a novel KFF-free recycled FPGA detection method based on an unsupervised anomaly detection scheme. As the RO frequencies in the neighboring logic blocks on an FPGA are similar because of systematic process variation, our method compares the RO frequencies and does not require KFFs. The proposed method efficiently identifies recycled FPGAs through outlier detection using direct density ratio estimation. Experiments using Xilinx Artix-7 FPGAs demonstrate that the proposed method successfully distinguishes two recycled FPGAs from 10 fresh FPGAs. In contrast, a conventional KFF-free recycled FPGA detection method results in certain misclassification.