[youtube https://www.youtube.com/watch?v=FDZ39h-kCS8]
The idea of using Wi-Fi signals to identify a person through a wall sounds like something out of a James Bond or Mission: Impossible movie. In fact, it’s a new piece of research coming out of the University of California, Santa Barbara. The work was conducted by a lab belonging to UC Santa Barbara professor Yasamin Mostofi. It demonstrates how a pair of Wi-Fi transceivers can be effectively used to reveal whether a person hidden behind a wall is the same individual who appears in a piece of video footage.
“Our key enabling idea is to translate the video content to the wireless domain and properly extract key gait features from both the real Wi-Fi signal and the translated video-to-Wi-Fi-one to establish if they belong to the same person,” Mostofi, a professor in the Department of Electrical and Computer Engineering, told Digital Trends. “Our proposed pipeline consists of a 3D mesh recovery algorithm that extracts the human mesh from the video. [This is] followed by an electromagnetic wave generation step that uses the extracted mesh to simulate the RF signal that would have been measured if the person in the video was walking in the Wi-Fi area. Next, we deploy our time-frequency processing pipeline to extract the spectrogram of both signals, which implicitly contains their corresponding gait information.”
By extracting several relevant features from both spectrograms and properly comparing them, the researchers demonstrated that it is possible to establish if the person in a video is also the person behind the wall in the Wi-Fi area. The technology does not require any prior video or Wi-Fi data of the person to be identified, or any knowledge of the operation area.
As neat as this is as a stand-alone tech demo, however, Mostofi suggests that the approach could also have practical applications. It could be a useful security or surveillance tool, with police able to identify criminals by using crime scene video footage and a pair of Wi-Fi transceivers outside. In public places, the existing Wi-Fi infrastructure could be used to help detect the presence of the suspect, by crossreferencing the crime-scene video footage.
The technology could also be used to help improve smart environments. For example, it may be useful for keyless car entry, where a pair of Wi-Fi transceivers on your car could detect your presence and unlock your car accordingly. It could additionally help identify any person walking into a part of their house and customize settings such as lighting, temperature, and music selection to suit them.
“We certainly would like to see this technology be of help to the society,” Mostofi said. “Towards that goal, we are currently working with our tech transfer office towards its commercialization.”