Learning from Routing Information for Detecting Routing Misbehavior in Ad Hoc Networks

Area of Science:

• Computer Science
• Network Security
• Wireless Communication

Background:

• Ad hoc wireless networks face security challenges due to limited resources, making complex security systems difficult to implement.
• Anomaly and intrusion detection systems are crucial for securing these networks, with host-based and cluster-based approaches having distinct advantages and disadvantages.
• Existing heuristic methods for intrusion detection in wireless ad hoc networks still have room for improvement.

Purpose of the Study:

• To propose a novel intrusion detection system adaptable for both host-based and cluster-based implementations in ad hoc wireless networks.
• To detect routing misbehavior attacks by leveraging shared routing information and advanced learning techniques.
• To evaluate the system's performance under varying network conditions, including mobility and size.

Main Methods:

• Development of a detection system capable of operating in either host-based or cluster-based modes.
• Creation of a dataset using novel routing-information-sharing algorithms tailored for wireless ad hoc networks.
• Application of both supervised learning (using historical or exploratory network data) and unsupervised learning for anomaly detection.
• Extension of the testbed to assess the impact of node mobility and network scale on detection accuracy.

Main Results:

• The proposed detection system demonstrated promising performance in identifying routing misbehavior attacks.
• The system's adaptability to both host-based and cluster-based configurations was validated.
• Simulation results indicated effective operation even under challenging conditions like high mobility and large network sizes.
• The use of shared routing information and flexible learning approaches proved effective for intrusion detection.

Conclusions:

• The developed detection system offers a viable and flexible solution for enhancing security in ad hoc wireless networks.
• The proposed routing-information-sharing algorithms and learning methods contribute to robust detection of routing attacks.
• The system's performance suggests its potential for practical deployment in resource-constrained wireless environments.
• Further research can explore optimizations for specific attack vectors and network dynamics.