Mobile Intelligence

17.1 INTRODUCTION

In this chapter, we will introduce an online scheme for threat detection within mobile ad hoc networks (MANET) using a single-cycle associative memory algorithm. The scheme implements a pattern-recognition approach, where the states of the network are considered as patterns. These patterns are collected and analyzed in real-time for discovering network intrusions and threat detection.

MANET is a decentralized network that comprises wireless mobile nodes that form a network through self-cooperation [1]. In MANET, there is no node configuration and coordination. The nodes are self-organized to form a mobile network. There are several issues related to MANET, these include routing in a changing topology, wireless communications, energy constraints, and a general lack of computational resource in the nodes. Being a wireless and mobile network, MANET is prone to security threats such as selfish node, distributed denial-of-service (DDoS), and traffic jamming. Intrusion detection is one of the security measures generally adopted for overcoming these threats. However, implementing intrusion detection within MANET requires a very different strategy to the standard IP-based wired network. There are several intrusion detection schemes (IDS) reported in the literature, which deal with the distributed and decentralized aspects of MANET. Huang and Lee [2] have proposed a cooperative approach that uses a cluster-based detection scheme. Sun et al. [3] have introduced the zone-based ...