Evaluation of Impact of Differential Transfer Efficiency of EM Leakage on Screen Reconstruction against High-Resolution Displays

In TEMPEST attacks targeting high-resolution displays that divide the screen into multiple areas, the reconstruction of screen information is difficult owing to multiple emanation sources. This is because each pixel is transmitted on a different path, which causes overlap of pixel information on reconstructed images. To solve this, we have proposed a reconstruction method of images on display with multiple areas based on the differences in transfer efficiencies between emanation sources on each screen area and the measurement point and consequently separating each screen information. However, we confirmed that variation in reconstruction accuracies of information is dependent on measurement parameters used to cause differences in transfer efficiencies. The selection of measurement parameters that renders highly accurate reconstruction has yet to be investigated. In this study, we investigated the effects of measurement parameters on the reconstruction accuracy of a reconstructed image, focusing on measurement positions and receiving frequencies as measurement parameters to vary transfer efficiencies. Specifically, the evaluation of the effects of measurement positions and receiving frequencies on transfer efficiencies was rendered possible by controlling radiation intensity based on operations of the screen image and measuring their spectrum. In addition, the reconstruction accuracy of reconstructed images was evaluated while altering measurement parameters using convolutional neural networks. The above evaluations indicate that the reconstruction rate was approximately 38% in cases involving the alteration of measurement positions, whereas it was approximately 62% in cases altering receiving frequencies. Therefore, it was confirmed that the alteration of receiving frequencies can improve the separation accuracy of screen information, resulting in high information reconstruction rate.