AdaTest: Reinforcement Learning and Adaptive Sampling for On-chip Hardware Trojan Detection

This paper proposes AdaTest, a novel adaptive test pattern generation framework for efficient and reliable Hardware Trojan (HT) detection. HT is backdoor attack that tampers with the design of victim integrated circuits (ICs) . AdaTest improves existing detection techniques in terms scalability accuracy detecting smaller Trojans presence noise variations. To achieve high trigger coverage, leverages Reinforcement Learning (RL) to produce diverse set inputs. Particularly, we progressively generate vectors ‘reward’ values an iterative manner. In each iteration, evaluated adaptively expanded as needed. Furthermore, integrates sampling prioritize samples provide more information detection, thus reducing number while improving samples’ quality faster exploration. We develop Software/Hardware co-design principle optimized on-chip architecture solution. AdaTest’s minimizes hardware overhead two ways: (i) Deploying circuit emulation on programmable accelerate reward evaluation input; (ii) Pipelining computation stage by automatically constructing auxiliary input generation, evaluation, sampling. evaluate performance various benchmarks compare it prior works use logic testing Experimental results show engenders up orders speedup size reduction compared achieving same level or higher rate.