Hardware Trojan horse detection method based on Adaboost algorithm

Abstract

The invention discloses a hardware Trojan horse detection method based on Adaboost. The method comprises steps of defining the gate-level structure features of hardware Trojan horses, calibrating thepositions of all hardware Trojan horses in a netlist and the Trojan horse feature vectors of gate-level elements, building a training and testing data set, and building and training an Adaboost hardware Trojan horse detection model according to the data set. When the error rate of the model is lower than a set threshold value, the model can be used for detecting hardware Trojan horses in the circuit netlist to be detected.

Description

technical field [0001] The application belongs to the technical field of integrated circuit hardware Trojan horse detection, and specifically designs a hardware Trojan horse detection method based on the Adaboost algorithm. Background technique [0002] With the continuous expansion of the scale of integrated circuits, the complexity of SoC (System on chip) systems is getting higher and higher. Circuit designers usually need to use third-party IP (Intellectual Property) cores in the system to achieve corresponding functions, and IP cores from untrusted third-party designers may be artificially implanted into hardware Trojan horses during the design process, so , in order to ensure the security of the SoC system, it is necessary to effectively detect the hardware Trojan horse circuit that may be introduced in the circuit design process. At the same time, with the increase of circuit complexity, the static information and dynamic information contained in the SoC system are be...

AI Technical Summary

Problems solved by technology

[0004] The invention provides a hardware Trojan detection method based on the Adaboost algorithm. The invention is mainly used for detecting a hardware Trojan implanted in a circuit netlist in the circuit design process. Compared with traditional logic testing and side channel analysis methods, it may be due to Problems such as low test coverage and noise interference affect the accuracy of detecting hardware Trojans. The hardware Trojan detection method proposed in this patent, by defining the structural characteristics of hardware Trojans, analyzing the structural characteristics of circuit netlists based on the Adaboost algorithm, completes the screening of hardware Trojans, and then Improve the detection efficiency of hardware Trojans

Images