Statistical Learning Over Time-Reversal Space for Indoor Monitoring System

As embedded in wireless signals, information of an indoor environment is captured during radio propagation, motivating the development of emerging wireless sensing technologies. In this paper, we propose a smart radio system that leverages the informative wireless radios to enable intelligent environment and extend human senses to perceive the world. In particular, owing to the time-reversal (TR) technique that captures changes in multipath profiles, the proposed TR indoor monitoring system (TRIMS) is capable of monitoring indoor events and detecting motion through walls in real time. A statistic model of intraclass TR resonance strength is developed and treated as the feature for TRIMS. Moreover, a prototype of TRIMS is implemented using commercial WiFi devices with three antennas. We investigate the performance of TRIMS in different single family houses with normal resident activities. In general, TRIMS can have a perfect detection rate with almost zero false alarm rates for seven target events, whereas during a two-week experiment TRIMS achieves a detection rate of 95.45% in the indoor multievent monitoring. The proposed TRIMS illustrates the potential of smart radio applications in smart homes, thanks to the ubiquitous WiFi.