EasiTrack: Decimeter-Level Indoor Tracking With Graph-Based Particle Filtering
Chenshu Wu, Feng Zhang, Beibei Wang, K. J. Ray Liu
Abstract
Despite decades of efforts, existing indoor location systems do not easily scale with low cost while maintaining high accuracy. We present EasiTrack, an indoor tracking system that achieves decimeter accuracy using a single commodity WiFi access point (AP) under non-line-of-sight (NLOS) conditions and can deploy at scale with almost zero cost. EasiTrack makes two key technical contributions. First, it incorporates RF-based inertial measurement algorithms that can accurately infer a target's moving distance purely using the RF signals received by itself. Second, EasiTrack devises a map-augmented tracking algorithm that outputs fine-grained locations by jointly leveraging the distance estimates and an indoor map that is ubiquitously available nowadays. We build a fully functional real-time system centering around a satellite-like architecture, which enables EasiTrack to support an unlimited number of clients. We have deployed EasiTrack in seven different scenarios (including offices, hotels, museums, and manufacturing facilities) to track both humans and machines. The results reveal that EasiTrack achieves a median 0.25 m and 90%tile 0.69-m accuracy in distance measurement, a median 0.58 m and 90%tile 1.33-m location accuracy for tracking objects, and a median 0.70 m and 90%tile 1.97-m accuracy for tracking humans in both line-of-sight and NLOS scenarios and supports a broad coverage of 50 m × 60 m using a single AP. It is also verified that EasiTrack can be easily deployed in massive buildings with little cost, promising a practical solution for ubiquitous indoor tracking.