Reconstruction of Sound Information Leakage Signals Obtained from Multiple Demodulation Methods

Speakerphones used in remote work environments have been reported to leak audio information through electromagnetic (EM) emanations. A method to improve information reconstruction quality has been proposed by simultaneous measurement of multiple leakage channels. However, acquiring sufficient reconstruction quality is difficult when the number of observable channels is limited due to limitations of measurement resources. In this study, we propose a method to increase the number of leakage channels virtually in order to reconstruct audio information with sufficient quality, even if observable channels are limited. We assume that leaked EM waves are modulated by multiple schemes and simultaneously extract audio information from a single leakage channel by amplitude and frequency demodulation methods. Furthermore, we synthesize audio signals obtained from different demodulation methods by a phase-aligned-spectrum-averaging method and indicate that the improvement in the reconstruction quality is comparable to measuring multiple leakage channels.