A Statistical Framework for Intrusion Detection in Ad Hoc Networks

Abstract

We focus on detecting intrusions in ad hoc networks using the misuse detection technique. We allow for detection modules that periodically fail to detect attacks and also generate false positives. Combining theories of hypothesis testing and approximation algorithms, we develop a framework to counter different threats while minimizing the resource consumption. We obtain computationally simple optimal rules for aggregating and thereby minimizing the errors in the decisions of the nodes executing the intrusion detection software (IDS) modules. But, we show that the selection of the optimal set of nodes for executing the IDS is an NPhard problem. We describe a polynomial complexity, distributed selection algorithm, "Maximum Unsatisfied Neighbors in Extended Neighborhood" (MUNEN) that attains the best possible approximation ratio. The aggregation rules and MUNEN can be executed by mobile nodes with limited processing power. The overall framework provides a good balance between complexity and performance for attaining robust intrusion detection in ad hoc networks. This material is posted here with permission of the IEEE. Such permission of the IEEE does not in any way imply IEEE endorsement of any of the University of Pennsylvania's products or services. Internal or personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution must be obtained from the IEEE by writing to pubs-permissions@ieee.org. By choosing to view this document, you agree to all provisions of the copyright laws protecting it. Abstract-We focus on detecting intrusions in ad hoc networks using the misuse detection technique. We allow for detection modules that periodically fail to detect attacks and also generate false positives. Combining theories of hypothesis testing and approximation algorithms, we develop a framework to counter different threats while minimizing the resource consumption. We obtain computationally simple optimal rules for aggregating and thereby minimizing the errors in the decisions of the nodes executing the intrusion detection software (IDS) modules. But, we show that the selection of the optimal set of nodes for executing the IDS is an NP-hard problem. We describe a polynomial complexity, distributed selection algorithm, "Maximum Unsatisfied Neighbors in Extended Neighborhood" (MUNEN) that attains the best possible approximation ratio. The aggregation rules and MUNEN can be executed by mobile nodes with limited processing power. The overall framework provides a good balance between complexity and performance for attaining robust intrusion detection in ad hoc networks.

Cluster based Statistical Anomaly Intrusion Detection for Varied Attack Intensities

“…In a later work [8], Yi et al extends previous work on the detection of local anomalies and conducted a comparative analysis of the function to explore the relationships between functions and other functions using a classification algorithm, decision tree. In [9], the authors took a different line of research by the application of theories of hypothesis testing and approximation of a statistical framework for intrusion detection in ad hoc networks to develop.…”