Statistical Test Methodology for Evaluating Electromagnetic Information Leakage From Mobile Touchscreen Devices

This paper presents a statistical-test-based methodology for evaluating the electromagnetic (EM) emanations from mobile touchscreen devices. These EM emanations potentially contain information that can be exploited to reconstruct screen images of the device remotely. The proposed method evaluates the threat of such EM-based information leakage by performing a statistical test of measured EM waves and environmental noise without performing ad-hoc screen restoration or image processing. In particular, we employ the Kolmogorov-Smirnov test, a nonparametric statistical test method, to compare two distributions of radiated EM waves: one that potentially contains screen information and another with environmental noise. In addition, this paper demonstrates the validity of our methodology via experiments involving authentic tablet devices in real-world settings and discusses a possible method to determine the confidence interval that can be adopted as a criterion for security evaluation depending on the environment.