A Defense Method Based on a Novel Replay Attack
Juan Guo, Liang Chen, Haoran Sun, Aidong Xu, Zeguang Li, Yinwei Zhao, Yixin Jiang, Tengyue Zhang, Yunan Zhang · 2021 IEEE International Conference on Power Electronics, Computer Applications (ICPECA) · 2021
Replay attack is a common attack model, and we propose a novel voice replay attack against intelligent voice assistants. This is a speaker array-based attack method that modulates the attack commands onto a high-frequency carrier and uses the nonlinear self-demodulation of the speaker array and air to produce speech commands audible to the human ear that which confirm can be executed by a voice smart assistance. In this paper, a novel voice replay attack against intelligent voice assistants is proposed by characterizing the speech attack and non-attack signals generated by the loudspeaker arrays and analyze their amplitude and frequency characteristics. We also propose and validate a software defense method based on machine learning and neural networks, and the results show that the method is effective.