VANET Denial of Service Attack Detection in Using Optimization and Deep Feedforward Neural Networks

Vehicular Ad Hoc Networks (VANETs) importance lies in being responsible for supporting secure and reliable communication for intelligent transportation systems. Though imperative to the operations, VANETs are subject to Denial-of Service attacks, with detrimental effects causing an erosion in the integrity of networks and an advancement in risking roads. The present work posits an Optimization-Based method against detection of DoS attacks in cooperation with the Cyber twin paradigm paired with Deep Feedforward Neural Networks (DFNNs). Through this method, real-time monitoring of the VANET is permitted with the activation of alarms whenever anomalies are sensed and imitations of potential candidate attacks in a simulated virtual laboratory. The deep learning nature of DFNN can be beneficial in analyzing complex vehicular traffic patterns, allowing its accurate identification of malicious behavior associated with DoS. For performance enhancement, Genetic Algorithm (GA) is applied for feature subset selection and hyper-parameter optimization that enhances the detection accuracy by reducing the computational cost. The experiment results are evaluated on actual and simulated vehicular communication datasets with excellent detection performance and low false positives. This validates the performance of our framework in dealing with unpredictable changes in distributed dynamic VANET environments. The platform provides a scalable, effective, and proactive way of monitoring the secure operation of VANETs under active DoS attacks via real-time observation, deep learning, and optimization, thus ensuring network reliability and facilitating road safety.