An Internet-of-Things-Enabled System for Road Icing Detection and Prediction
Zhuo Chen, Gengang Xiong, Yao Sun, Yun Li, Yan Li
Abstract
Road icing has become one of the most critical factors threatening traffic safety. This article proposes an Internet of Things (IoT)-enabled road icing detection and prediction system. In the proposed system, we first design a low-power icing sensor equipped with IoT function to periodically collect current road status and transmit the sampled data to IoT gateway through Long Range Radio (LoRa). Then, we design a simple but effective algorithm deployed on IoT gateway to identify road icing in time. The algorithm is proposed based on the change trend of the sampled data of the road state, and can be adapted to the icing recognition on the road covered with various impurities. Furthermore, we put forward a newly designed deep neural network model called Trans-CGAN to achieve accurate road icing prediction even the positive and negative samples are imbalanced. Through a real system deployment and experiments, the results show that our proposed system can detect the formation of road icing effectively and timely, and shows better prediction performance of road icing than several representative models.