FlexiGuard: Self-Adaptive and Dynamic Context-Based Access Control for Cross-Domain Data Sharing

Access control for data sharing among multiple autonomous organizations remains a challenging problem, as existing solutions typically rely on predefined static rules and cannot adapt well to such a scenario featured by hierarchical cross-domain data circulation with dynamic control contexts. While zero trust emerges as a promising tool to solve the problem, integrating it with a data-centric access control model that can dynamically generate context-aware policies and efficiently enforce trustworthy authorization and authentication is a non-trivial task. In this paper, we propose a self-adaptive and dynamic access control framework for cross-domain data sharing, which follows up zero-trust security principles and exploits blockchains coupled with rule learning to facilitate context-based access control. Specifically, our approach utilizes real-time contextual information for dynamic rule learning and compiles multiple rule clusters guided by Dempster-Shafer evidence theory to generate precise and adaptive control policies. Experimental results show that the proposed framework achieves desired security goals with a minimal performance overhead (approximately no more than 13 milliseconds), even for large-scale networks with high-concurrency tasks.