RAPTOR Turns a Beady Eye Onto Counterfeit or Maliciously Modified Chips

Researchers from Purdue University and Oak Ridge National Laboratory have come up with a system that, they say, could help fight counterfeit electronic chips — turning a deep learning system called RAPTOR onto the problem.

"The global chip industry is grappling with dual challenges: a profound shortage of new chips and a surge of counterfeit chips valued at $75 billion, introducing substantial risks of malfunction and unwanted surveillance," the researchers claim. "To counteract this, we propose an optical anti-counterfeiting detection method for semiconductor devices that is robust under adversarial tampering features, such as malicious package abrasions, compromised thermal treatment, and adversarial tearing."

The team's answer to this problem is a deep-learning system dubbed RAPTOR, or Residual, Attention-based Processing of Tampered Optical Response. This, the team says, can deliver high-accuracy detection of counterfeit chips in milliseconds, by looking for a visible implementation of a physical unclonable function (PUF) — in this case, a cluster of gold nanoparticles including in the chip's manufacture.

Trained on an in-house dataset of 10,000 images of randomly-distributed gold nanoparticles, RAPTOR measures the distance between particles — and can, its creators say, distinguish between chips that have been subject to normal wear-and-tear and those which can been tampered with, either maliciously to include a hardware Trojan or simply to disguise a relabeled or entirely counterfeit part.

"Using semantic segmentation and labeled clustering," the team explains, "we efficiently extract the positions and radii of the gold nanoparticles in the random patterns from 1000 dark-field images in just 27 ms and verify the authenticity of each pattern using RAPTOR in 80ms with 97.6% accuracy under difficult adversarial tampering conditions."

The team's work has been published in the journal Advanced Photonics under open-access terms.