Explore a groundbreaking approach to wireless sensing in this 17-minute conference talk from USENIX NSDI '23. Delve into SLNet, a novel deep wireless sensing architecture that combines learning-based spectrogram generation with spectrogram learning. Discover how this innovative method overcomes the time-frequency uncertainty limitation and utilizes a polarized convolutional network to learn both local and global features from spectrograms. Examine the application of SLNet in four real-world scenarios: gesture recognition, human identification, fall detection, and breathing estimation. Learn how this new technique outperforms state-of-the-art models in accuracy while maintaining a smaller model size and lower computational requirements. Gain insights into the potential widespread applications of SLNet's techniques beyond WiFi sensing, opening new possibilities in the field of wireless technologies.
Overview
Syllabus
NSDI '23 - SLNet: A Spectrogram Learning Neural Network for Deep Wireless Sensing
Taught by
USENIX