Improving Effectiveness of Hardware Trojan Detection using Fault Injection
Abstract
Outsourcing the ICs for manufacturing introduces potential security threats such as hardware Trojans. Trojan inputs usually come from nets with rare activity. Therefore, increasing the probability of rare event activities in the circuit can increase the detection probability. We propose a method to increase the Trojan detection probability using fault injection. The proposed method has no hardware overhead, and has the potential for detection of several types of Trojans that are triggered by rare events. It increases the probability of Trojan activation by up to three times, and achieves Trojan detection rate above 95% with false alarm rate below 3%.
Keywords
Controllability, Fault Injection, Rare Events, Transition Probability, Trojan Detection