Chakrabarty Wins 2021 IEEE Transactions on VLSI Systems Prize
New machine learning-based approach to detect Trojans in microprocessor cores is proven highly accurate
More and more frequently, the phases of semiconductor design and manufacturing are outsourced to companies around the globe. While this strategy can make the process cheaper, it also makes it riskier; outsourcing increases opportunities for bad actors to add malicious logic–referred to as "hardware Trojans”– to the original designs. These Trojans can cause denial-of-service attacks, degrade the system's performance and change the functionality of a microprocessor core.
The complexity of modern microprocessors makes it difficult to detect hardware Trojans in the early stages of semiconductor production, but ECE professor Krishnendu Chakrabarty and collaborators Rana Elnaggar ECE PhD ’20 and Karlsruhe Institute of Technology’s Mehdi Tahoori have introduced a machine learning-based run-time hardware to detect Trojans in microprocessor cores. The team’s method detects these Trojans more than 99% of the time, with a false positive rate of 0%.
Their paper, "Hardware Trojan Detection Using Changepoint-Based Anomaly Detection Techniques,” earned the team the 2021 IEEE Transactions on Very Large Scale Integration (VLSI) Systems Prize Paper Award.