Implement of a secure selective ultrasonic microphone jammer

Eavesdropping via microphones has been a serious threat to security and privacy. Recent advances in utilizing non-linearity property of microphone amplifiers enable ultrasonic transducers to act as jammers. Due to the advantages of low-cost, inaudibility, and high jamming performance, those ultrasonic microphone jammers are promising in resisting covert eavesdropping. However, there are some barriers to their effective implementation. On one hand, existing approaches do not support authorized devices to record clean audios, which severely limits the usage of ultrasonic jammers. On the other hand, the unauthorized adversary can utilize noise reduction methods to recover the original audios, while current ultrasonic jammers cannot combat such attacks. In this paper, we propose a secure and selective microphone jamming system, which can prevent unauthorized devices from eavesdropping and ensure authorized recording devices operate normally. We utilize ultrasounds to jam unauthorized recording devices. Meanwhile, jamming noise is delivered through multiple wireless channels to authorized devices, which can use adaptive noise filter to remove the noise. Moreover, we specifically utilize multiple broadband jamming signals to improve the security of our microphone jamming system and defend against several adversary audio recovery methods. Experimental results show that less than 1% of words in unauthorized recordings can be recognized while in authorized recordings 92% of words can be recognized. Furthermore, even using noise reduction methods, 95.9% of words still cannot be recognized in unauthorized recordings.