Malicious Node Detection in VANETs via Enhanced DSR and ML

Being vulnerable to various security threats like other wireless networks, Vehicular Ad-hoc Networks (VANETs) need an effective intrusion detection system to ensure security in VANETs. This paper proposes an intrusion detection system for VANETs using machine learning techniques. The system is imple...

Full description

Saved in:

Bibliographic Details
Published in:	2024 International Conference on Wireless Communications Signal Processing and Networking (WiSPNET) pp. 1 - 7
Main Authors:	AN, Palaniappan, J, Pranav, S, Shivamanikkavasakam, Kumar, Santosh, R, Gandhiraj
Format:	Conference Proceeding
Language:	English
Published:	IEEE 21-03-2024
Subjects:	Blackhole attack DSR protocol Intrusion detection Machine learning NetSim simulation Routing protocols Signal processing Support vector machines Transportation VAnet security Vehicular ad hoc networks Wireless networks
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	Being vulnerable to various security threats like other wireless networks, Vehicular Ad-hoc Networks (VANETs) need an effective intrusion detection system to ensure security in VANETs. This paper proposes an intrusion detection system for VANETs using machine learning techniques. The system is implemented in NetSim where a VANET is created with a malicious node causing blackhole/sinkhole attack. The malicious node is detected using enhanced Dynamic Source Routing (DSR) protocol and watchdog timer. Once detected, the route through the malicious node is avoided in future transmissions. The packet trace from NetSim simulation is used as input dataset for Support Vector Machine (SVM) and Random Forest models for classifying nodes as malicious or not. The models provide over 99% accuracy, precision, recall and F1-score in detecting malicious nodes, outperforming previous approaches. The system provides an effective intrusion detection framework for securing VANETs against attacks using simulations and machine learning.
DOI:	10.1109/WiSPNET61464.2024.10532957