Conference/ Workshop/ Seminar/ Proceedings/ Lecture Series
Permanent URI for this community
Browse
Browsing Conference/ Workshop/ Seminar/ Proceedings/ Lecture Series by Author "Ayuya, Caroline"
Now showing 1 - 1 of 1
Results Per Page
Sort Options
Item Behavioral Analysis and Predictive User Authentication in Zero Trust Architectures: A Systematic Review.(Daystar University, Global Cybershield, 2024) Ayuya, Caroline; Ayiro, Laban Peter; Osoro, Ragira EdwinBackground: Conventional biometric and authentication systems such as fingerprints, face recognition, and passwords are vulnerable to shoulder heat, surfing, and smudge attacks. This has resulted in the use of Behavioral analysis and predictive user authentication in the protection of Zero Trust Architectures (ZTAs). These approaches improve user identification and authentication, reducing cyber threats with continuous verification of access requests regardless of the user's location or device. Objectives: The review shows behavioral analysis and predictive user authentication as the main tools of ZTA, determining the trends, patterns and anomalies. Methodology: Databases such as Google Scholar, IEEE Xplore, Science Direct, Scopus, Embase and ACM were included in the search strategy. The articles from the period of 20202024 and peer-reviewed were chosen and reviewed according to the inclusion and exclusion criteria. The Critical Appraisal Skills Program (CASP) tool was employed to detect the quality of the studies and the extent of their risks of bias. Results: 12 studies met the criteria and were included in the final analysis. Those adopting supervised machine learning techniques like random forests (25%), support vector machines (17%), and neural networks (17%) for predictive user modeling. Approaches like clustering accounted for 17%. The most common data sources were network traffic (17%), application logs (17%), and user activity monitoring (9%). F1-score was the predominant evaluation metric (42%), followed by accuracy (33%). Reported F1 scores ranged from 0.71 to 0.94 across various techniques. Conclusions: This review confirms that behavioral analysis and predictive user authentication are ZTA's defenses against attacks. The results provide evidence-based proof of the effectiveness of these methods. They also contribute to the safety of the networks, data and systems in multiple industries.