A hybrid deep learning based enhanced and reliable approach for VANET intrusion detection system

Abstract

Advances in autonomous transportation technologies have profoundly influenced the evolution of daily commuting and travel. These innovations rely heavily on seamless connectivity, facilitated by applications within intelligent transportation systems that make effective use of vehicular Ad- hoc Network (VANET) technology. However, the susceptibility of VANETs to malicious activities necessitates the implementation of robust security measures, notably intrusion detection systems (IDS). The article proposed a model for an IDS capable of collaboratively collecting network data from both vehicular nodes and Roadside Units (RSUs). The proposed IDS makes use of the VANET distributed denial of service dataset. Additionally, the proposed IDS uses a K-means clustering method to find clear groups in the simulated VANET architecture. To mitigate the risk of model overfitting, we meticulously curated test data, ensuring its divergence from the training set. Consequently, a hybrid deep learning approach is proposed by integrating Convolutional Neural Networks (CNN) and Bidirectional Long Short-Term Memory (BiLSTM) networks. which results in the highest training, testing, and validation accuracy of 99.56, 99.49, and 99.65% respectively. The results of the proposed methodology is compared with the existing state-of-the-art in the same domain, the accuracy of the proposed method is raised by maximum of 4.65% and minimum by 0.20%.