11-09-2017, 12:43 PM
Silence suppression, an essential feature of voice communications over the Internet, saves bandwidth by deactivating voice packet transmissions when silence is detected. However, mute suppression allows an adversary to retrieve conversation patterns from packet synchronization. In this article, we investigate the privacy leak through the mute suppression function. More specifically, we propose a new class of traffic analysis attacks to encrypted voice communications with the aim of detecting speakers of encrypted voice communications. These attacks are based solely on packet time information and attacks can detect voice communications loudspeakers made with different codecs. We evaluated the proposed attacks with extensive experiments on different types of networks, including commercial anonymity networks and campus networks. Experiments show that proposed traffic analysis attacks can detect high-accuracy encrypted voice communications based on 15-minute traces on average.