Sybil attack detection and secure data transmission in VANET using CMEHA-DNN and MD5-ECC

TLDR: A deep learning-centered intrusion detections system (IDS) is proposed utilizing CMEHA-DNN for detecting the SA in VANET and attains an accuracy of 98.37% and a security level of 99.2%, which is better compared to existing methods.

Abstract: Vehicular ad-hoc networks (VANET) technology, which is an open-access network, renders a quick, simple to deploy, and cheap solution for intelligent traffic control as well as traffic disaster preventive measure but it is prone to disparate sorts of attacks. The Sybil attack (SA) is the most harmful attacks that the VANET has to face. In this, the attacker generates manifold identities to fake manifold nodes. It is extremely onerous to defend as well as detect, especially if it is commenced by means of some connived attackers utilizing their genuine identities. Here, a deep learning-centered intrusion detections system (IDS) is proposed utilizing CMEHA-DNN for detecting the SA in VANET. The proposed technique encompasses ‘4’ steps: (i) cluster formation (CF), (ii) cluster head (CH) selection, (iii) attack detection, and (iv) security of VANET. Initially, the MKHM clustering algorithm clusters the vehicles. Next, the Floyd–Warshall algorithm (FWA) selects the CH as of the clusters. Subsequent to CH selection, the malevolent CH is identified utilizing the deep leaning model CMEHA-DNN by means of extracting the pertinent features as of the CH. Lastly, subsequent to detection, if the CH is a normal one, the information contained by means of the CH is securely sent to the cloud utilizing the MD5-ECC. The proposed work attains an accuracy of 98.37% and a security level of 98.2%, which is better compared to existing methods.