Revisiting Access Control in Complex Environments: Advancing Decision Making and Explainability of Access Control Systems
Output:
Thesis
Phd Thesis 1 (Research TU/e / Graduation TU/e)
Academic
Open access
Information
Output type
Output:
Thesis
Phd Thesis 1 (Research TU/e / Graduation TU/e)
Academic
Original language
EnglishQualification
Doctor of PhilosophyPublication milestones
- Published - 19/10/2026
Publication status
Published - 19/10/2026
Supervisors/Advisors
- N. Zannone (Promotor)
- E. Zambon (Copromotor)
Place of publication
EindhovenAward date
19/10/2026Publisher
Eindhoven University of TechnologyISBN (Print)
978-90-386-6795-9Abstract
Large volumes of data are collected and exchanged daily due to the deep integration of digital technologies into personal and organizational life. Protecting such data has become a central concern with significant economic and personal implications. Access control plays a fundamental role in providing this protection by regulating which actions a subject can perform on a resource. It therefore serves as a primary line of defense for preserving the confidentiality and integrity of digital assets and for ensuring the secure operation of information systems. Weaknesses in access control and authentication, however, are among the leading causes of security incidents. In particular, ensuring effective access control is challenging in complex environments, with many involved stakeholders and dynamic conditions. In collaborative scenarios, resources may be owned and managed by multiple stakeholders whose authorization preferences over a shared object can conflict. Existing work on multi-party access control often fails to capture the dynamics of real-world decision making, does not account for fairness, or relies on user intervention that is impractical in many settings. Complexity can also arise when granted permissions are exercised in unintended or harmful ways. In mission-critical environments such as hospitals, protocols like Break the Glass can be invoked to bypass access control mechanisms, further enabling potential misuse of privileges. Once permissions are granted, access control alone cannot cope with the misuse, and existing approaches still struggle to detect and prevent it in an effective and interpretable manner. These challenges compromise the protection of sensitive information and weaken trust in the system. At the same time, while access control has become more expressive and capable of capturing nuanced privacy requirements, it has also become harder to reason about. Besides the challenges of decision making, important gaps remain in how stakeholders comprehend the rationale behind the resulting decisions. Despite several techniques aimed at making access decisions more understandable, the field still lacks a clear conceptualization and a structured framework for helping stakeholders understand how decisions are reached. In this thesis, we therefore examine the following main research question: RQ: How can we support effective access decision making and policy comprehension in complex environments? We approach this question by investigating access control in complex environments from three perspectives, covering access decision making in collaborative environments, anomaly detection for identifying privilege misuse, and explainable access control to improve decision comprehension. We study access decision making in collaborative environments and propose a game-theoretic approach that resolves conflicts between parties across different strategic environments. Our solution accounts for key characteristics that affect the acceptability of collective decisions, such as fairness and optimality. We also tackle the challenges of collaborative environments from an alternative perspective that allows modifying shared content to regulate access in online social networks. We use state-of-the-art deep learning techniques to shift privacy enforcement to the level of individual stakeholders, while preserving the fidelity of the modified content. Evaluation reveals that participants are generally willing to adopt our fully automated privacy tool for context-aware image editing based on personalized privacy preferences within the study setting and tasks considered. Following this, we investigate how anomaly detection can complement access control to detect potential misuse of legitimately assigned privileges in complex environments. We develop profiling techniques that enable anomaly detection by flagging requests deviating from established profiles as potential privilege misuse. In particular, we propose data-driven methods that draw on historical access events to characterize typical user behavior for profiles. We further examine how domain-informed feature engineering and data scope choices impact the resulting profiles. Our findings confirm that data-driven profiling, supported by expert validation, enables construction of representative and interpretable profiles of access behavior. Finally, we argue that enabling effective access control in complex environments can increase the complexity of access decision making and reduce its transparency. In such settings, the reasoning behind obtained decisions may need to be explained to the involved stakeholders for tasks such as policy refinement in order to improve transparency and trust in the system. To address the comprehension of access decision making and underlying policies, we introduce a conceptual model of explainable access control that structures explanation in terms of the relevant actors, purposes, and forms of explanation. We propose techniques for explaining access decisions, including a causality-based framework for identifying the reasons behind decisions and a visualization framework to support comprehension of XACML policies and their evaluation. In conclusion, this thesis shows that effective access control in modern environments requires more than enforcing authorization rules. It requires mechanisms that can support the complexities inherent in collaborative settings, methods that complement policy enforcement through behavioral analysis, and techniques that help stakeholders comprehend the reasoning behind the obtained access decisions. By addressing these challenges, this thesis contributes methods and foundations for access control that support effective decision making and policy comprehension in complex environments.
Bibliographical Note
Proefschrift.
Documents and Links
Final published version, 30.25 MB
Metrics
Metrics
Download statistics
Download count
75
