On the basis of the analysis of clustering algorithm that had been proposed for MANET, a novel clustering strategy was
proposed in this paper. With the trust defined by statistical hypothesis in probability theory and the cluster head selected
by node trust and node mobility, this strategy can realize the function of the malicious nodes detection which was
neglected by other clustering algorithms and overcome the deficiency of being incapable of implementing the relative
mobility metric of corresponding nodes in the MOBIC algorithm caused by the fact that the receiving power of two
consecutive HELLO packet cannot be measured. It's an effective solution to cluster MANET securely.
Proc. SPIE. 5284, Wireless Communications and Networks
KEYWORDS: Detection and tracking algorithms, Networks, Data processing, Telecommunications, Computer intrusion detection, Data communications, Information security, Computer security, Mobile communications, Network security
Ad hoc networks do not depend on any predefined infrastructure or centralized administration to operate. Their security characters require more complex security preventions. As the second line of defense, Intrusion detection is the necessary means of getting the high survivability. In this paper the security characters of ad hoc networks and the related contents of intrusion detection are discussed. Mobile Agent and Multi-layer Integrated Distributed Intrusion Detection Model (MAMIDIDM) and a heuristic global detection algorithm are proposed tentatively by combining the mobile agent technology with the multi-layer conception. This heuristic global detection algorithm combines the mobile agent detection engine with the multi-layer detection engines and analyzes the results obtained by the corresponding detection engines. MAMIDIDM has the better flexibility and extensibility, can execute the intrusion detection in clustering ad hoc networks effectively.