CONFERENCE (INTERNATIONAL) AI-BPO: Adaptive incremental BLE beacon placement optimization for crowd density monitoring applications
Yang Zhen (Tokyo Institute of Technology), Masato Sugasaki (Tokyo Institute of Technology), Yoshihiro Kawahara (The University of Tokyo), Kota Tsubouchi, Matthew Ishige (The University of Tokyo), Masamichi Shimosaka (Tokyo Institute of Technology)
29th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (ACM SIGSPATIAL 2021)
November 19, 2021
With the pandemic of COVID-19, indoor crowd density monitoring has become one of the most critical responsibilities of public space managers. Since the performance of crowd density monitoring highly depends on how BLE beacons are allocated, BLE beacon placement optimization has been tackled as fundamental research work in the Wireless Sensor Network and Robotics communities. However, the previous research focuses on batch optimization based on the simulation of beacon signal propagation which cannot provide the optimal solution for the real-world environment. Some other research proposing beacon selection from densely placed beacons ignores the actual workload to obtain the optimal placement result. In this research, we propose a novel beacon placement optimization approach to incrementally place the beacon on the updated detection status adaptively in favor of Bayesian optimization, which can help to provide the optimal beacon placement. Our proposed method can optimize the beacon placement effectively to improve the signal coverage quality in the given environment and minimize human workload. We conduct the experiment in a laboratory setting and confirm that our method places the similar to best sensor placement in all candidates of sensor placement. In addition, we conduct a feasibility study in a wild environment, and the result shows that our proposed method can achieve 19% higher detection coverage than the beacon-placement inexperienced person's solution and 1.24% over the expert's solution in the real-world experiment. The result also shows that our proposed method can reduce 72.9% of the walking distance and 59.4% of the optimization time compared with optimization by dense data gathering.