Browsing by Author "Obonna, Ugochukwu Onyekachi"

Now showing 1 - 1 of 1
  • Thumbnail Image
    ItemOpen Access
    Detection and prevention of amorphous cyber-attacks in process control networks of oil and gas installations
    (Federal University of Technlogy, Owerri, 2023-09) Obonna, Ugochukwu Onyekachi
    Amorphous cyber-attacks in process control networks (PCN) of oil and gas installations have posed a major cyber security challenge to the industry, due to the consistent deployment of unpredictable dynamic attack strategies by the attackers, which has made it difficult to predict their next attack modes. The aim of this dissertation is to monitor, detect, prevent and mitigate the effect of these malicious attacks on PCN. To achieve this, standard engineering structured methods, tools and techniques were applied by developing and analyzing mathematical models of the attacks, designing secured, centralized and distributed process control network architecture, designing a defense system capable of detecting false data injection attacks. The engineering materials, principles and concepts applied in the simulation, testing and validation of the developed models include: top-down structural approach, block diagrams, Machine learning algorithms, Deep learning toolkits, structured programming languages and simulation packages, such as: Python 3.0 Libraries, MATLAB, Allen Bradley PLC RSLogix 5000 emulator software, flowcharts and algorithmic representations of normal as well as compromised plant operations. Modelling and simulation of a 3-phase separator under attack was used to showcase attacks on PCN, predictions and forecasting using different machine learning algorithms. Real-time 68,722 SCADA dataset used in this research helped to overcome some of the shortfalls of previous researchers who used identical and repeated datasets that affected the learning ability of their algorithms and their final outcome. Several other machine learning algorithms and analytical tools were explored using the same dataset, but the Coarse Tree algorithm produced the best results with 100% accuracy, zero False Alarm Rate, one million observations per second prediction speed and 0.45488 seconds computation time. The results obtained showed the quality and precision of attack detections, hence the model’s robust performance in detecting network intrusions. The various units and system's tests conducted showed highly improved results of about 95% comparatively with previous industry research results, thereby confirming the probability of useful contributions made in ameliorating theft in Oil and Gas sector of the economy. The successful integration of the developed models to the developed PCN architecture, shows that the cyber-attacks vulnerabilities on Oil and Gas infrastructures could be detected, prevented and reduced to barest minimum, thereby preventing production downtime, with adverse impact on the economy of the country, in general.