Facial reconstruction

Search LJMU Research Online

Browse Repository | Browse E-Theses

Behavioural Observation for Critical Infrastructure Security Support

Hurst, William (2014) Behavioural Observation for Critical Infrastructure Security Support. Doctoral thesis, Liverpool John Moores University.

[img] Text
157627_2014-WHurst-PhD.pdf - Published Version

Download (4MB)

Abstract

Critical infrastructures include sectors such as energy resources, finance, food and water distribution, health, manufacturing and government services. In recent years, critical infrastructures have become increasingly dependent on ICT; more interconnected and are often, as a result, linked to the Internet. Consequently, this makes these systems more vulnerable and increases the threat of cyber-attack. In addition, the growing use of wireless networks means that infrastructures can be more susceptible to a direct digital attack than ever before.
Traditionally, protecting against environmental threats was the main focus of critical infrastructure preservation. Now, however, with the emergence of cyber-attacks, the focus has changed and infrastructures are facing a different danger with potentially debilitating consequences. Current security techniques are struggling to keep up to date with the sheer volume of innovative and emerging attacks; therefore, considering fresh and adaptive solutions to existing computer security approaches is crucial.
The research presented in this thesis, details the use of behavioural observation for critical infrastructure security support. Our observer system monitors an infrastructure’s behaviour and detects abnormalities, which are the result of a cyber-attack taking place. By observing subtle changes in system behaviours, an additional level of support for critical infrastructure security is provided through a plug-in device, which operates autonomously and has no negative impact on data flow.
Behaviour is evaluated using mathematical classifications to assess the data and detect changes. The subsequent results achieved during the data classification process were high and successful. Our observer approach was able to accurately classify 98.138 % of the normal and abnormal system behaviours produced by a simulation of a critical infrastructure, using nine data classifiers.

Item Type: Thesis (Doctoral)
Uncontrolled Keywords: Critical Infrastructure, Behavioural Observation, Cyber-Attack, Security, Data Analysis, Data Classification
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Q Science > QA Mathematics > QA76 Computer software
Divisions: Computer Science & Mathematics
Date Deposited: 19 Oct 2016 13:19
Last Modified: 03 Sep 2021 23:26
DOI or ID number: 10.24377/LJMU.t.00004382
Supervisors: Merabti, M and Fergus, P
URI: https://researchonline.ljmu.ac.uk/id/eprint/4382
View Item View Item