Exploring effectiveness of a predictive light control mechanism for wireless sensor networks: poster abstract

Light management for buildings or streets is important for safety, security and comfort. However, it is not economical to keep the lights on in places without people. Although there is a lighting system with power-saving feature that lights up automatically when it detects moving objects, it has a problem that it does not light up unless the objects approach the immediate vicinity of the lights. In this work, we aim to realize the system that collects motion detection data of objects using a wireless sensor network, predicts the approach of a person from a position farther than before, and appropriately controls lighting on / off. In this system, data is exchanged between neighboring nodes, and each node autonomously judges lighting control. We discussed a predictive control method and made experiments for an adequacy confirmation. First, we constructed a system to verify the predictive control method and made a data acquisition experiment using PIR motion sensors. Next, we examined methods for a movement prediction and lighting control based on the acquired data, and confirmed that control results provide sufficient utility.