IoT-Assisted Estimation of Vehicular Traffic Status over Wide-Area

Monitoring of traffic status over a large area is an important component of a smart-city system. The existing approaches primarily utilize camera-based solutions for vehicle detection, where live images of the roads are transmitted over an active Internet connection and analyzed with the aid of cloud services for traffic estimation. Such solutions involve high costs for deployment and heavily depend on an Internet connection with sufficient bandwidth available ubiquitously. This not only makes the wide-area deployment of such solutions difficult, but also keeps a significant fraction of the Internet bandwidth occupied. Several studies so far have found that vehicular traffic has an impact on low-power wireless communication. In this work, we leverage this phenomenon to develop an alternative, low-cost, and lightweight solution for sensing real-time traffic status over a wide area. In particular, the targeted area is considered to be equipped with a specialized low-power IoT system comprising low-power devices installed at the designated locations. Subsequently, the existing lightweight data collection strategies are exploited to collect data at a sink node. The solution thus enables the data collection process to interact with itself. The characteristics or performance of the process are next analyzed using standard ML models to infer the traffic status at different locations. A simulation-based study of the proposed low-cost solution shows an average accuracy upto 90% in even identifying congested locations over the city.