Optimizing Traffic Safety: A Dual Multi-Scale Attention Network for Anomaly Detection in VANETs

Vehicular Ad Hoc Networks (VANETs) improve road safety and traffic flow, but face security threats due to their decentralized nature. Conventional anomaly detection mechanisms are ineffective, requiring higher scalable accuracy in computationally feasible solutions. This research introduces the real-time monitoring of anomalies through comprehensive introduction to using a Dual Multi-Scale Attention Network (DMSA) that was optimized via Planet Optimization for anomaly detection in systems related to the monitoring of real-time traffic. Multiple metrics - including throughput, dropped packet ratio, and overhead traffic ratio are incorporated - to the collected dataset which is of real-time nature. To achieve a reasonable model performance, the collection set is normalized using three scaling rescaling techniques. Anomaly detection is then implemented by DMSA, which includes innovative attention mechanisms to trace even the most complex deviations of patterns in traffic data. Optimizing DMSA model using the Planet Optimization (PO), fine-tuning the hyperparameter, so that detection is enhanced, and at the same time, computational efficiency gets improved to detect the fine anomalies which might have remained undetected in the previously used traditional methods while offering high efficiency in dynamic vehicular networks. The results of the conducted experiments show that the DMSA-PO framework was able to achieve detection accuracy in the range of 98.2%, precision of 97.5% and an F1-score of 97.2%, higher than others in use. This quantum enhanced model offers a strong, scalable, and efficient anomaly detection solution for VANETs to guarantee their security and reliability in dynamic, real-time environments.