A Deep Learning Based Intrusion Detection Approach in Manets for Detecting Black Hole Attacks

Abstract

Mobile ad hoc networks (MANETs) have attracted increasing attention in the field of computer science due to their many civil and military applications. However, these applications pose significant security challenges. To effectively protect mobile wireless networks, it is crucial to implement an intrusion detection system (IDS) that is tailored to the specific characteristics of these networks and capable of identifying various types of threats. This article proposes an approach for designing an intrusion detection system, in which a dataset specifically designed for mobile ad hoc networks (MANETs) is generated to improve black hole attack detection. This approach exploits the functionalities of different layers of the OSI model, such as the physical, MAC, and network layers, to extract relevant features. A scheme was developed to collect data from the Network Simulator 2 (NS-2) simulator and process it to form a dataset called MANET-DataSet. An artificial neural network (ANN) was then trained with this dataset to detect black hole attacks. The results show that using MANET-DataSet improved the accuracy of the intrusion detection system while reducing the false positive rate.