Ensemble 알고리즘을 이용한 Machine Learning 하드웨어 Backdoor 탐지방법

Abstract

As integrated circuit (IC) technology is applied to various fields every year, the importance of security and reliability of integrated circuits is also emphasized in various fields. Recently, security attacks have been carried out in various ways, and hardware Backdoor is a representative issue. Hardware Backdoors are circuits that have been modified for malicious purposes in electronic circuits. Depending on the purpose of the Backdoor designer, it may trigger in a specific environment and cause serious security problems in the entire system. In this paper, the international standard encryption AES (Golden model) and the AES_T700 and AES_T200 Backdoor models provided by the Trust-Hub benchmark were used. In this paper, we propose a method for detecting Backdoors through a machine learning model using an ensemble algorithm by extracting power consumption data in the time domain of each model.