Echo-Based Location Privacy Attacks in Video Conferencing

This conference case examines privacy risks in video conferencing due to audio channel vulnerabilities. It introduces novel echo-based attacks that exploit residual distortions for location inference. Combining acoustic sensing challenges under AEC with cutting-edge transformer-based algorithms, the approach identifies and demonstrates success rates for in-channel and off-channel attacks. The methodology includes preprocessing echo signals, extracting location embeddings, and analyzing...