Demo Abstract: Privacy-Preserving Room Occupancy Estimation Using Federated Analytics of BLE Packets

We present a privacy-preserving room occupancy estimation method using federated analytics of Bluetooth Low Energy (BLE) packets. By processing data locally and reporting only aggregated device counts, our approach preserves user privacy while achieving 95% accuracy in occupancy prediction. This method provides a cost-effective alternative to traditional sensing technologies like PIR and mmWave, balancing privacy, accuracy, and ease of deployment. Future work will expand testing to multi-room setups and enhance privacy measures. Code URL: https://github.com/Johnnybyzhang/BLE-capture

Duke Scholars

Author Bing Luo DKU Faculty