Enhancing Supply Chain Security: ML Assisted Manufacturer Identification of SRAM PUFs
Harshdeep Singh, Fathi Amsaad · 2024
The widespread manufacturing of memory chips at untrustworthy facilities has led to the entry of fake memory chips into the supply chain. Identification of these counterfeit chips is crucial because it poses substantial concerns about the security of electronic systems. Current approaches generally focus on identifying specific kinds of counterfeiting, perhaps neglecting underlying flaws that develop throughout system operation. This paper uses process variations of the manufacturer, allowing machine learning models to identify differences in manufactur-ers without complex procedures. The technique's usefulness is demonstrated through validation utilizing 345 SRAM chips from five reputable manufacturers. It results in F1 ratings of 0.98 for accurately identifying the device manufacturer, respectively.