Development of an IoT based Real-Time Traffic Monitoring System Using Magnetic Sensors

The Internet of Things (IoT) is playing an increasingly vital role in Intelligent Transportation Systems (ITS), enabling real-time sensing and communication. In ITS, data on vehicle types, traffic volume, and speeds are essential for effective road traffic management. However, existing on-road traffic monitoring methods often fall short of delivering cost-efficient solutions to meet the growing demands of modern traffic systems. In this paper, an innovative framework is developed for on-road traffic surveillance using small and easy-to-install magnetic sensors. The magnetic sensor system is wireless-connected, cost-effective, and environmentally friendly which consists of an accelerometer and a magnetometer, and it is located in the pavement. A novel feature extraction and vehicle classification framework is proposed which has the potential to improve the recognition rate of magnetic sensors. Firstly, the features of magnetic waveforms for vehicle identification and speed estimation are extracted from local magnetic field perturbations caused by moving vehicles. Then, the order of the extracted features is reduced using an optimization algorithm to find the most informative and relevant characteristics of the sensor signals that contribute significantly to vehicle classification by removing irrelevant and redundant features. Finally, the reduced set of sensor-derived characteristics selected to maximize classification accuracy are used as inputs into classification models. The proposed framework, which includes two modes of offline training and real-time classification, is validated using real-field experimental data collected from a pilot site located in Rudná street in Kuncičky, Czech Republic. Field test results show that the proposed framework can detect traffic with a correct vehicle category detection rate of over 93% in real-time classification mode. Furthermore, the performance of the proposed estimation technique is compared with other vehicle-type estimation algorithms, which shows the superior accuracy of the proposed technique. These show that the proposed system can be considered for widespread implementation in smart city infrastructure due to its low installation cost, energy-efficient operation and scalability.