Buildings Occupancy Estimation: Preliminary Results Using Bluetooth Signals and Artificial Neural Networks

The energy consumption in the European Union continues to grow above the expected values, and buildings are one of the largest consumers in front of industry and transportation sectors. As buildings have different roles with different requirements and characteristics, new approaches are required to increase the efficiency of new and old buildings to reduce consumption. Locating people inside buildings can be done using cameras, sensors, or radio signal strengths, where their intrusion may vary. In this paper, we present our approach to locating building occupants using Bluetooth Low Energy (BLE) Scanners without previously requiring the fingerprint of the area where the system is deployed. To do it, we created a Machine Learning pipeline to locate the devices. Using our approach, we obtained an average error of 5.68m. We also demonstrate that rotating the positions of the Scanners while maintaining their distances does not have an impact on the location accuracy.