Enabling and Advancing Human and Probabilistic Context Awareness for Smart Facilities and Elder Care
Abstract:
We are at the end of a fouryear effort that has dramatically improved the capability of the use of RF sensors, particularly those that measure received signal strength (RSS) to sense the locations and context of people in buildings and homes. We have investigated both systems which use RFID tags to identify a person or object, and those which use a static deployed network of transceivers for devicefree localization, to locate people moving in the environment who do not carry any tag or device. Locating people who don’t wear a device is critical for building security and automatic safety systems, and is important in elder care applications in which we do not want to assume that a person being monitored will remember to wear a sensor or device.
In summary, our efforts have led to orderofmagnitude reductions in localization error per system size (area or sensor numbers). We have developed technologies which make RFbased devicefree localization a very accurate and robust indoor localization method and capable to be used to determine a person’s activity. Our methods are real time and capable of being deployed for long periods of time in realworld dynamic environments, and still tracking multiple people. A spinoff company, Xandem, has succeeded in selling products that commercialize RF sensing methods. Our team used our system in the 2012 Evaluating AAL Systems through Competitive Benchmarking (EvAAL) international competition and had significantly better performance than the next best localization system. We have shown that devicefree localization has higher accuracy than radio device localization, which would not have been believed when we started this work. We have shown that it can be used to help reliably determine the activity which a person is doing in their home.
We have found that a still person’s breathing rate can be estimated within 0.5 breaths per minute, as accurate as medical sensors. We have found that a still, but breathing person can be located in a home using RSS measurements based only on the person’s inhalation and exhalation. The results have application in home health care as well as in finding people buried in the rubble of a collapsed building.
We have also found that our existing wireless networks, by transmitting, expose information about the building occupant’s positions and movements, information which would otherwise be presumed private. This radio window attack is a significant privacy challenge, one that requires new wireless system design guidelines in order to prevent in future networks.
We have developed and taught a new graduate class, and have spun off technologies to products now being sold. Our results have appeared in the popular press, for example, in Engadget, Popular Science, Fast Company, and BBC Radio