Contactless Activity Identification Using Commodity WiFi

Activity monitoring in home environments has become increasingly crucial for elder care, well-being management, and latchkey child safety. However, traditional approaches require expensivewearable sensors or specialized hardware installations, which can be intrusive and uncomfortable.To address this issue, this chapter presents a low-cost system for device-free and location-oriented activity identification at home using existing WiFi access points and devices. The system leverages the complex web of WiFi links and fine-grained channel state information that can be extracted from them to identify both in-place activities and walking movements by comparing them against signal profiles. The construction of signal profiles can be semi-supervised and adaptively updated to account for the movement of mobile devices and signal calibration. Experimental evaluation in two apartments of different sizes demonstrates that our approach achieves a high average true positive rate for distinguishing a set of in-place and walking activities with only a single WiFi access point. Furthermore, the prototype also indicates that the system can work with a wider signal band (802.11ac) with even higher accuracy, making it a promising alternative to traditional approaches for activity monitoring in home environments.