An intrusion detection system for SDN-based tactical networks: a machine learning approach

Network security is increasingly becoming a critical and continuous issue due to ongoing advancements in Information and Communication Technologies (ICT) and the concomitant rise in the number of security threats. This is especially true for military communication networks as security breaches may have detrimental effects. However, over the years, it has become increasingly difficult to attain high levels of detection accuracy in military tactical networks with conventional anomaly detection systems due to the dynamic nature of network traffic in the battlefield, special operations, and the harsh environment where they operate. Intrusion detection systems (IDS) have emerged as essential countermeasures to preserve network security. In addition, the introduction of software-defined networks (SDN) in tactical networks presents countless opportunities for security. This study developed an IDS model for military tactical networks that utilizes Machine Learning (ML) techniques for high detection rates and SDN for network global view and centralised data collection. Following the Design Science methodology, the model was designed based on guidelines from related literature, and a proof-of-concept prototype of the model was implemented to assess its effectiveness. The experimental results indicated that Machine Learning using network flow data collected via SDN can improve intrusion detection rates in tactical networks. Among the machine learning techniques, ensemble learning methods utilising Decision Tree classification methods, namely Random Forest and Adaptive Boosting, obtained high recall and precision when detecting DDoS attacks, malicious, and misbehaving nodes in an SDN-enabled tactical network.

Grantor dc:publisher.institution

University of Zululand

Year dc:date.issued

2020

Author dc:creator

• Zwane, Skhumbuzo Goodwill

• Network security
• Security threats
• Intrusion detection systems (IDS)

Licence

• https://devspace.unizulu.ac.za/bitstreams/f65dd6a4-e0d7-4fda-a4ea-854257ec9666/download