A Hardware Trojan Detection Method Based on Recognizing the Waveform Similarity of Oscillating Circuit

To solve the problem that the hardware Trojan occupies a small number of resources, and the triggering condition is difficult to detect, this paper proposes a Hardware Trojan detection method based on recognizing the waveform similarity of oscillation circuit. First, a model for detecting the hardware Trojan is built. Then, four hardware Trojans aimed at the AES encryption chips in the Trust-Hub library are detected respectively by using a ring oscillation circuit or a multivibrator circuit with external connection. The square wave signals at the feature points are extracted by principal component analysis and the results were obtained by correlation analysis, i.e. recognizing the waveform similarity by calculating the coefficient of association. Experimental results show that the four hardware Trojans could be identified, and one of the four Hardware Trojans only occupies 0.26% of the total chip resources. In addition, compared to the traditional detection methods based on detecting the power consumption, this method has higher accuracy and efficiency for detecting the Hardware Trojans at the inactive state.