Analysis of Magnetic Signatures for Vehicle Detection Using Dual-Axis Magneto-Impedance Sensors

The use of magnetic sensors for vehicle detection is proposed as a solution for monitoring traffic owing to its advantages, including ease of setup, low cost, strong resistance to interference, and privacy preservation. In this study, we proposed a vehicle detection system using dual-axis magneto-impedance (MI) sensors, which are placed on the roadside. The bar magnet and magnetic dipole models were proposed for simulating the movement of driving vehicles, and the applicability of both models was discussed. The experiment was conducted on a straight road in Aichi, Japan, and six common vehicle types on urban roads were used. A detection algorithm based on the real-time geomagnetic field was used to automatically detect moving vehicles and extract valid signals. The algorithm achieved a detection accuracy of 91.6% for vehicles driven under normal conditions. The impact of various experimental parameters, including the sensor placement height, measurement distance, and actual driving conditions, on the measurement results was evaluated. The best detection results were achieved when the sensor was placed at a height of 66 cm and was approximately 0.9 m away from the vehicle. Five magnetic features in the time and frequency domains were used to analyze the driving state of vehicles. In the frequency domain, the bandwidth and dominant frequency were confirmed to have a strong correlation with the driving speed, whereas their relationship with the vehicle type and other experimental conditions was relatively weak, making them suitable for speed estimation.