Application of Active Learning Algorithm in Mobile Ad Hoc Network Intrusion Detection

Abstract

Network intrusion detection system is widely used in the security protection of network systems. It monitors network traffic, finds suspicious traffic, and actively responds. In recent years, the network intrusion detection system based on anomaly detection has attracted widespread attention because it can detect unknown attacks. Introduction Mobile ad hoc network is a wireless network, which is composed of mobile nodes. These nodes can communicate with each other without using fixed infrastructure. Communication between nodes is based on routing protocols such as distance vector protocol (DV), link state protocol (LSP) and hybrid protocol. Mobile ad hoc networks have applications in many applications, such as sensor networks, vehicle networks, wireless sensor networks, and so on. Due to the characteristics of open media, dynamic topology, interaction and limited resources, mobile ad hoc networks need more security than traditional networks. This paper will discuss the application of active learning algorithm in mobile ad hoc intrusion detection system. An integrated intrusion detection model is introduced. In this model, the classifier with supervised anomaly detection is based on support vector machine. At the same time, three pool-based active learning algorithms applied in the model are introduced. Compared with the traditional self-learning algorithm, the pool-based active learning algorithm can effectively reduce the dependence on training samples and reduce the impact of noise data on the performance of the intrusion detection system. It is suitable for the requirements of mobile ad hoc networks for high detection rate, high anti-noise ability and low computational delay of the intrusion detection system.