Towards Accurate and Real-time Information Extrapolation on Charging Pile Network

The Internet of Things era of transportation brings many new challenges. One of the most important issues in cyberspace is to build an information infrastructure that provides services for unmanned electric vehicles. For example, nearby available charging stations. This paper presents the first systematic study on the information extrapolation problem in the charging pile network (CPN). The challenge is how to deal with incomplete data caused by device failure, communication anomalies, crowdsourcing and commercial competition. This paper provides a novel solution to infer the real-time charging pile occupancy status. We analyzed the spatiotemporal pattern and user charging behavior of charging events, and proposed a data extrapolation algorithm based on Monte Carlo maximum likelihood estimation and Gibbs sampling technique. The experimental results show that our method achieves a 94% accuracy, which is 14% and 11% higher than the two benchmark methods, respectively, and it can increase user satisfaction by up to 44% on the end-to-end service.