Skip navigation.

Contact Us

Department of Computer Science

Janssen Engineering
Room 236
PO Box 441010
Moscow, Idaho
83844-1010

phone: 208-885-6592
fax: 208-885-9052

e-mail:
dept chair
graduate info
undergrad info
sys admin
webmaster

CS Department Banner Graphic

Ondrej Linda
MS Thesis Defense

APPLICATIONS OF COMPUTATIONAL INTELLIGENCE IN CRITICAL INFRASTRUCTURES: NETWORK SECURITY, ROBOTICS, AND SYSTEM MODELING ENHANCEMENTS

Major Professor: Dr. Milos Manic

Abstract:

Critical infrastructures comprise essential components of industry, energy generation, security, defense, transportation and public services. Their protection and effective maintenance is one of today's most relevant concerns. Due to the recent development in information technology, modern critical infrastructures have become substantially computerized and automated. This thesis discusses the use of computational intelligence techniques for support and enhancements of modern critical infrastructures. Computational intelligence techniques, such as artificial neural network, fuzzy logic systems or unsupervised clustering, are well established and widely used, and are capable of machine learning, pattern recognition or non-linear intelligent control. In this thesis, these techniques were utilized in three specific areas of critical infrastructures: network security, robotics and system modeling. Firstly, the artificial neural networks were applied to the problem of network security and an anomaly-based intrusion detection system was implemented. The development and testing of this system was based on network data recorded from a control system of an existing critical infrastructure. Further, the intrusion detection system was enhanced by an improved training data generation technique. Secondly, specific control architecture of multi-robot system for emergency response was designed. This architecture combined a swarm behavior model with fuzzy logic control into an effective single-operator control system. Moreover, an algorithm for autonomous navigation of mobile robots inspired by the support vector machines was developed. Finally, this thesis describes enhanced system modeling using the growing neural gas algorithm. This self-organizing neural network was used for problem complexity reduction and for topology learning.