Wifi-based robust indoor localization for daily activity monitoring

Abstract

Achieving indoor localization enables several intelligent home applications, such as monitoring overall activities of daily living (ADL) and triggering location-specific IoT devices. In addition, ADL information further facilitates physical and mental health monitoring and extracting valuable activity insights. While many approaches are proposed to attack this problem, WiFi-based solutions are widely celebrated due to their ubiquity and privacy protection. However, current WiFi-based localization approaches either focus on fine-grained target localization demanding high calibration efforts or cannot localize multiple people at the coarser level, making them unfit for robust ADL applications. In this work, we propose a robust WiFi-based room/zone-level localization solution that is calibration-free, device-free(passive), and built with commercial WiFi chipsets. We extract features indicative of the motion and breathing patterns, thus detecting and localizing a person even when there is only subtle physical movement. Furthermore, we used the correlation between the movement patterns to break ambiguous location scenarios. As a result, we achieved an average detection rate of 96.13%, including different activity levels, and localization accuracy of 98.5% in experiments performed across different environments.