An Improved Deep Learning-Based Intrusion Detection for Reliable Communication in VANET

Abstract

The research area of Vehicular Ad-Hoc Network (VANET) is rapidly expanding for different reasons. The main factors responsible for this are the unstable topology, fast vehicle movement, and security vulnerabilities. Identifying malicious attacks in vehicular ad-hoc networks is necessary to enhance the security and reliability of communication between all vehicles in the system. The intruders carry out numerous malicious attacks. The focus of this paper is to identify and mitigate attacks like Botnet, Sybil, DoS, Wormhole, PortScan, Blackhole, and BruteForce. The present work proposes an Improved LeeNET (I-LeeNet) architecture to identify and mitigate unidentified attack types. The suggested architecture intelligently blends Convolutional Neural Networks (CNN) and Adaptive Neuro-Fuzzy Inference Systems (ANFIS) to provide real-time solutions to unidentified attacks. The proposed approach includes a module <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$K_{IDS}$ </tex-math></inline-formula> for known attack detection and another module <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$U_{IDS}$ </tex-math></inline-formula> for learning and identifying previously unidentified attacks. The <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$K_{IDS}$ </tex-math></inline-formula> module uses ANFIS classification to identify destructive attacks. ANFIS and CNN use the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$U_{IDS}$ </tex-math></inline-formula> module to identify unknown attacks on VANET further. The efficacy of the proposed I-LeeNet is tested on three different datasets, such as i-VANET, ToN-IoT, and CIC-IDS 2017. The results section discusses the testing in a regressive manner. In addition, the experimental results are compared to other advanced methods. The average accuracy achieved by the proposed method is 97.21% on i-VANET, 97.75% on ToN-IoT, and 96.66% on the CIC-IDS 2017 dataset. The analysis suggests the proposed method is promising and can be applied in real-time.