Researchers Train a Neural Network to Detect Human Occupancy by Sniffing Ambient Wi-Fi Signals
By monitoring the Wi-Fi signals in a room, and passing the data through a convolutional neural network, human presence can be detected.
Researchers from Syracuse University have developed a new way to detect whether a room is occupied by humans or not: by analyzing the effect of their movement on ambient radio signals using a convolutional neural network.
"Using Wi-Fi signal[s] as an example, we demonstrate that the channel state information (CSI) obtained at the receiver contains rich information about the propagation environment. Through judicious pre-processing of the estimated CSI followed by deep learning, reliable presence detection can be achieved," the researchers claim in the abstract to their paper. "Several challenges in passive RF sensing are addressed.
"With presence detection, how to collect training data with human presence can have a significant impact on the performance. This is in contrast to activity detection when a specific motion pattern is of interest. A second challenge is that RF signals are complex-valued. Handling complex-valued input in deep learning requires careful data representation and network architecture design. Finally, human presence affects CSI variation along multiple dimensions; such variation, however, is often masked by system impediments such as timing or frequency offset.
"Addressing these challenges," the researchers continue, "the proposed learning system uses pre-processing to preserve human motion induced channel variation while insulating against other impairments. A convolutional neural network (CNN) properly trained with both magnitude and phase information is then designed to achieve reliable presence detection. Extensive experiments are conducted. Using off-the-shelf Wi-Fi devices, the proposed deep learning based RF sensing achieves near perfect presence detection during multiple extended periods of test and exhibits superior performance compared with leading edge passive infrared sensors. The learning based passive RF sensing thus provides a viable and promising alternative for presence or occupancy detection."
"Exploiting the ubiquity of ambient RF signals such as Wi-Fi, Bluetooth or cellular signals for situational awareness information provides added value to existing RF infrastructure," researcher Biao Chen explains in an interview with TechXplore on the subject. "Occupancy detection, for example, is an application where RF sensing can be a low-cost and infrastructure-free alternative or complement to existing approaches."
Chen and colleagues aren't the first to have the same core idea, however: Late last year researchers from the Universities of Chicago and California at Santa Barbara updated a 2018 paper to detail how an off-the-shelf smartphone sniffing Wi-Fi signals can be used for both occupancy detection and positioning of human-sized subjects through a solid wall; a team at the Massachusetts Institute of Technology (MIT), meanwhile, combined radio-based monitoring with computer vision algorithms to track occupants' movement with enough accuracy to recognise gestures as fine-grained as shaking hands and playing a smartphone game.
The full paper, Harvesting Ambient RF for Presence Detection Through Deep Learning, is now available under open access terms on arXiv.org.