A Method for Extracting Plausible Images From EM Leakage Measured at Low Sampling Rates

It has been assumed that attackers with advanced expertise who can use a large setup for precise measurements perform the TEMPEST, which entails eavesdropping on the screen information by measuring and analyzing electromagnetic (EM) waves that leak from video device units (VDUs) [1]. Meanwhile, recently, the possibility that a non-expert attacker might use software to take control of low-cost, small, portable instruments and execute attacks has surfaced [2]. However, the sampling rate of low-cost instrumentation is lower than the pixel clock frequency of the VDU; the pixel information obtainable through leaked EM waves is limited. Therefore, the screen images reconstructed from measured data have pixel information deficits and degraded image quality. Based on the above, we propose a method to reconstruct plausible images from EM leakage, even in measurement environments with low sampling rates and fluctuating clocks. The proposed method focuses on clock jitter, which causes degradation of the quality of the images reconstructed using conventional methods, and measure pixel information over multiple frames using the fluctuation of the sampled pixels caused by clock jitter (Fig. 1). We then reconstruct high-quality images using an optimization problem formulated based on the observed pixel information and timing.