RadSense: Radio Sensing for Crime Monitoring in Restricted-Access Areas

Abstract

Wireless technology has become increasingly crucial and contributes to both short-range and long-range communication applications, such as remote health monitors and smart homes. Given the expectation of high-dense wireless communications in sixth-generation networks (6G), radio sensing technologies have recently gotten much attention. This work demonstrates a novel technique to monitor criminals in a restricted-access building via radio sensing. The sensing is performed on the latest commercial WiFi device. Accordingly, the data are discretely collected at several common positions to build baseline signal patterns. Next, we propose a few-shot learning model to train on a limited dataset. The testing results show that our system can recognize the mobility of a criminal behind the door, even with sparse training data. In some hazardous circumstances, this application can let law enforcement officials figure out criminal activities without physical intrusion, thus mitigating potential risks. Further, this model can also be applied to other commercial wireless networks.