Building an automated measurement platform for EM-Leakage analysis

Abstract

Our laboratory has published preliminary results on the detection of internal behaviors by neural network (NN) through IC electromagnetic side-channel leakage(EM-leak). In this paper, we use a customized program, an oscilloscope, and a robotic arm to design a dedicated measurement platform of EM-leak information data collection for NN training. In order to ensure the accuracy of the training results, the training data size must be large enough and the measurement platform should be designed to acquire data automatically. A computer is connected to a three-axis robotic arm and controls the arm moving accurately and repetitively among different measurement points. The computer also controls the oscilloscope to collect data. Different parameters can be adjusted in real-time. Experimental results show the proposed platform can be used as a NN model testing set data acquisition for EM-leak attack analysis.