If you used an indoor antenna back in the days of analog TV you know that moving around in a room can affect the RF environment. The reflections were visible as “ghosts” on the analog TV. The indoor antenna picking up analog TV is gone today but many houses now have Wi-Fi signals in them using MIMO technology designed to take advantage of different propagation paths to increase bandwidth. Considering how moving around affected the VHF and UHF analog TV signals, it's reasonable to expect humans would have the same impact on MIMO Wi-Fi signals.
Computer scientists at the University of Washington have been able to analyze how humans affect the RF environment and have even developed gesture recognition technology that allows people in a house to control other devices, even if they aren't in the same room! The UW researchers built a smart receiver that can receive all the wireless signals coming from devices throughout a home, including smartphones, laptops and tablets. A modified MIMO Wi-Fi router with multiple antennas could be adapted to function as a receiver. The multiple antennas allow the receiver to identify individual gestures from a group of people in the same house.
The new technology, called “WiSee”, has been submitted to the 19th Annual International Conference on Mobile Computing and Networking.
Lead research Shyam Gollakota, a UW assistant professor of computer science and engineering, said, “This is re-purposing wireless signals that already exist in new ways. You can actually use wireless for gesture recognition without needing to deploy more sensors.”
When a person moves in an RF field, the signal reflected from the person changes frequency slightly – several Hertz. Moving a hand or a foot causes the receiver to detect this Doppler shift. According to the news release, the technology can identify nine different whole-body gestures, ranging from pushing, pulling and punching to full body bowling. Researchers tested these gestures with five users in a two bedroom apartment and an office environment. Out of 900 gestures performed, WiSee accurately classified 94 percent of them.