Machine Learning Based Intrusion Detection of Wormhole Attack in Mobile Ad-Hoc Networks

Mobile Ad hoc networks are transient, infrastructure-free networks that self-organize. A group of peer nodes that interact with one another without the aid of any infrastructure make up an ad hoc network. The most serious security risk to an ad hoc network is a wormhole attack, which is not resistant to standard security measures. A simulation of a wormhole attack in an environment of an ad hoc network with numerous wormhole tunnels is one of our objectives. The characterization of packet attributes that influence feature selection is the following job. In order to produce a large volume of dataset, we conduct data generation and data collection operations. The final job involves using machine learning to detect wormhole attacks. Previously, a wormhole assault was discovered using conventional methods. The most serious security risk to an ad hoc network is a wormhole attack, which is not resistant to standard security measures. A network layer attack that mimics routing methods is known as a wormhole attack. Several machine learning techniques, including the support vector machine (SVM), decision tree (DT), and convolutional neural network, are used to accomplish the classification. (CNN). Additionally, we used the MANET's node's speed as well as other node characteristics for feature extraction. We have gathered 3997 unique examples that include both benign and malicious nodes (normal 3781 and malicious 216). The classification findings demonstrate that the SVM, DT, and CNN methods, respectively, have accuracy rates of 98.2%, 98.9%, and 96.4%. According to our research, the DT method's accuracy is better than other methods at 98.9%. SVM and CNN both show high accuracy for the following priority