Real-Time Anomaly Traffic Data Identification Method for Connected Vehicles in V2X Communication

The state data of connected vehicles (CVs) can be shared in real time through vehicle-to-everything (V2X) communication to improve traffic safety and efficiency. In order to improve the reliability of traffic data, a method based on improved random forest and adaptive Kalman filter (IRF-AKF) is proposed in this paper to identify anomaly traffic data. Firstly, the data cells of CVs consisting of position, velocity, acceleration are input into random forest (RF). RF is applied to identify the presence of anomaly data cells, and whale optimization algorithm (WOA) is adopted to optimize the parameters of RF, including the number and maximum depth of decision trees. Then, the data cells with presence of anomalies are input to adaptive Kalman filter (AKF). In AKF, the noise is dynamically adjusted and each item of a data cell is identified separately. Finally, an anomaly traffic data identification experiment system based on edge computing is designed. The final experimental results have shown that the IRF-AKF method can achieve a recall of 98.05% in real-time anomaly traffic data identification, which will make an important contribution to the safe driving of CVs.