Real-Time Indoor Localisation using Passively Captured Wi-Fi Probe Requests
Abstract
Digital transformation of buildings and public spaces increasingly depends on knowing how those spaces are used. Yet, the dominant sensing options create a poor trade-off: cameras are comprehensive but intrusive, while discrete heat sensors preserve privacy at the cost of coverage. This paper presents an implementation of a localisation system that occupies the middle ground. It passively monitors the IEEE 802.11 probe requests actively transmitted by Wi-Fi-enabled clients. Using low-cost ESP32s, distributed in the monitored space, we record the associated received signal strength indicators (RSSIs) for these frames and estimate the device positions on an edge server. The system fits a parametric propagation model that accounts for known room geometry and per-wall attenuation, requires only lightweight calibration, and runs in real time. Evaluated across two deployments in both a 470 m2 and 200 m2 area, the system achieves a median localisation error of roughly ± 2 and ± 3 m, respectively. Devices are classified in the correct room with roughly 87% accuracy for the entire 470 m2 space. It matches the performance of other model-based publications, while adding real-time operation, minimal-calibration deployment and attempts to be scalable. The result is a privacy-conscious, distributed localisation system for occupancy and anomaly sensing as part of a smart infrastructure.