Benchmarking RSS-based localization algorithms with LoRaWAN

• DOI: 10.1016/j.iot.2020.100235
• Publication Date: 2020-05-29T15:41:55Z
• Authors: Thomas Janssen; Rafael Berkvens; Maarten Weyn
• Keywords: 0202 electrical engineering, electronic engineering, information engineering; 02 engineering and technology; Engineering sciences. Technology

Abstract Initially, Low Power Wide Area Networks (LPWAN) were designed for communicating small sensor readings in a metropolitan area. With the rapidly increasing number of Internet of Things (IoT) devices, however, there is a growing need to localize these devices with LPWAN. Given the extreme battery lifetime requirements of IoT-enabled devices, we want to avoid satellite-based solutions. We use a publicly available outdoor LoRaWAN data set to evaluate Received Signal Strength fingerprint-based and range-based location estimation algorithms in terms of accuracy and computational performance. On the one hand, a fingerprint-based approach leads to 50% more accurate results when compared to a range-based approach, resulting in mean location estimation errors of 340m and 700m, respectively. On the other hand, in a range-based approach, no training database is required and the system can be deployed instantly in any environment with network coverage. Thus, when choosing a localization algorithm for any IoT application, a trade-off between accuracy, deployment cost and computational performance should be taken into account.