Context-Aware Outdoor LoRa Localization and Tracking using a Single Base Station in Urban Settings

Abstract

Accurate outdoor localization at long-range remains an important open problem in wireless systems. A leading low-power wide-area networking technology LoRa is touted to accelerate the deployment of outdoor sensors across application domains. Prior approaches for localizing LoRa clients either require specialized hardware at ground stations to measure phase, or remain woefully inaccurate even at a moderate range of communication. This paper attempts to bridge this gap by developing a single base station localization and tracking solution that requires only RSSI measurements. This paper presents RayTrack, a ray-tracing and RSSI-based localization and tracking system for LoRa clients. First, we adapt the ray-tracing based RF-propagation estimation to better mimic the real-world by providing antenna and surface characteristics. Next, we collect RSSI information across frequencies and leverage the ray-traced RSSI map along with user context to present a new algorithm for user-aware localization. Finally, we present a single-frequency RSSI-based tracking system that finds the path between the two checkpoints at low-latency. Our evaluation on a campus-scale testbed of 300,000 m2 showcases a localization error of 24.22 m (user context-unaware : 73.5 m) and a tracking accuracy of 2.57 m at a 36 m checkpoint spacing. (Code and data: https://github.com/qiancheng-li/RayTrack)