Terahertz cyber security testing using artificial intelligence (FA-002)

The focus of this proposal is FA-002 - artificial intelligence (AI), which will be used for the hardware cyber security. AI will link the THz and sub-THz responses at the pins to the defects and deviations from design of the integrated circuits under test. An increasing complexity of digital and mixed-signal systems makes establishing the authenticity of a chip to be a key problem. New rapid, inexpensive, and non-destructive technology is needed to address this problem. We will develop a new terahertz AI testing technique for non-destructive identification of genuine integrated circuits, even in-situ and even under bias, by measuring their response to terahertz and/or sub-terahertz radiation at the circuit pins and using AI algorithms for the response classification and interpretation. A big advantage of this new technique compared to other radiation enhanced testing is that this new approach does not affect the device operation. In contrast to the conventional THz imaging, this technique will use the intensity, polarization, frequency, and bias dependences of the THz response at the VLSI or MMIC pins. Our recent results demonstrated that this technique allowed for the unambiguous identification of the MMIC defective transistors.