LoRD uses heuristics to detect hardware Trojans in synthesized netlists

Researchers at UW-Madison and Marist University unveiled LoRD, a lightweight heuristic method for detecting and localizing RTL Trojans after synthesis. It scored an average of 2.957 out of 3 on contest testcases, beating a transformer-based ML baseline and the top five teams without training overhead.

• LoRD exploits stable structural and signal-flow patterns of RTL Trojans post-synthesis
• Average score of 2.957 out of 3 on Trojan-implanted contest testcases
• Outperformed a transformer-based ML baseline and top five teams
• No data or tuning overhead required