(Note: This is a fictionalized account of how a journalist may have covered the paper, “Multi-Person Localization via RF Body Reflections” by Fadel Adib, Zachary Kabelac and Dina Katabi. None of the quotations below are true and must be interpreted as fiction only.)
Massachusetts Institute of Technology
Imagine a smart home that keeps track of your and your family’s movements throughout the house without the need for multiple sensors placed in every room. Recent research at MIT is bringing this vision closer to reality.
Their solution is dubbed WiTrack2.0, and it represents the second version of the system the team developed in 2013. “The first version of our system focused on tracking a single user with RF signals as they moved across rooms,” says the first author of the study. “The key assumption that allowed us to detect and track a user in a home environment was user motion. However, our system could only localize a single user at a time, and also couldn’t account for stationary users. With this work, we have tried to address these limitations by building a system that can not only localize multiple users at once, but also detect the presence of stationary users.”
Their system uses multiple antennas that transmit and receive specialized radio signals termed FMCW, or frequency modulated carrier wave signals. Similar to radar signals, these signals have a particular structure that allows the researchers to easily measure the time delay between the transmitted and received signals even with a narrow system bandwidth.
These signals bounce off the environment before being received back at the receive antennas. A key insight that allows the system to localize a moving user is that the reflections from static objects remain constant over time. Therefore, a moving user can be localized by subtracting the received signals obtained from signal transmissions separated from one another by a few milliseconds. By using multiple transmit and receive antennas, the researchers are able to localize up to 4 moving users.
However, this subtraction technique also subtracts away the responses from static users. “The trick to localize stationary users,” says the lead author, “is to realize that by breathing, they too are subtly moving their bodies, albeit at a much lower pace. Therefore, instead of subtracting closely-spaced received signals as we do for localizing moving users, we subtract the signal responses corresponding to signal transmissions a few seconds apart. This allows us to localize up to 5 stationary users.”
“What’s exciting about our work is that it allows us to localize multiple users even when they are not directly in the view of the system,” says the senior scientist associated with the study. “Current systems based on optical and infrared cameras, like Kinect, fail to address problem and need users to be moving in direct sight of the system. Also, our work departs from previous works that required users to hold devices in order to be localized.
“This opens up a world of possibilities where our technology could be applied – hands-free virtual reality gaming and smart home automation are just some of the many places where device-free multi-user localization would play a major role.”
###
