Runtime support for rule-based access-control evaluation through model-transformation

Salvador Martínez, Jokin García, Jordi Cabot

TL;DR

This paper proposes a novel approach to building robust and adaptable runtime access-control evaluation engines by leveraging model transformation frameworks. By treating access-control policies and requests as models, they can be automatically transformed into executable code, significantly reducing the development burden for integrating fine-grained security into applications, especially DSLs.

What Problem Does This Paper Address?

Implementing access-control in software applications is crucial for security, but building the underlying runtime evaluation engine (the Policy Decision Point or PDP) is a complex, time-consuming, and error-prone task. While enterprise applications often have built-in solutions, smaller frameworks and Domain-Specific Languages (DSLs) lack readily available, reusable, and adaptable PDP infrastructures. This forces developers to either build these critical components from scratch or rely on specialized, often inflexible, solutions. The paper aims to bridge this gap by providing a generic, model-driven approach to generate these PDPs.

Methods

The core of the paper's approach lies in using model transformation frameworks, specifically ATL (Atlas Transformation Language), to automate the creation of PDPs. The process involves:

  • Policy Injection: Translating existing access-control policies (written in a specific language) into a model conforming to a defined policy metamodel.
  • Model Transformation Generation (HOT): Using a Higher-Order Transformation (HOT) to automatically generate an ATL model transformation specification from the policy metamodel. This HOT acts as a "policy-to-transformation" generator.
  • Access Request Evaluation: The generated ATL transformation then takes an access request (also modeled) as input and produces an evaluation model, effectively acting as the PDP.
  • Conflict Resolution: An optional, language-independent refining transformation can be applied to resolve conflicts between rules, ensuring a single, definitive access decision.

This approach separates the access-control logic from the application, making it more modular and reusable.

Datasets

Not discussed in the paper. The paper focuses on a prototype implementation and its evaluation, not on training models with specific datasets.

Key Insights and Ideas

  • Model Transformation as a PDP: The fundamental insight is that the process of evaluating an access-control request against a policy can be elegantly modeled as a model transformation. This allows leveraging mature model transformation technologies for security infrastructure.
  • Automated PDP Generation: The use of HOTs to generate the ATL transformation specifications from policy metamodels significantly automates the creation of PDPs, reducing manual effort and potential errors.
  • Language Independence: By working with metamodels and transformations, the approach offers a degree of independence from specific access-control languages, making it more adaptable.
  • Separation of Concerns: The model-driven approach naturally enforces a separation between the access-control infrastructure and the application logic, improving maintainability and testability.
  • Flexibility and Extensibility: The model-based nature allows for easier modification of policy and request metamodels to incorporate new features like traceability or support for obligations.

Key Results

The paper demonstrates the feasibility of their approach with a prototype PDP for a rule and attribute-based access-control language. The performance evaluation shows that the PDP performs reasonably well, with execution times comparable to existing XACML PDP implementations. Specifically, for policies with up to 700 rules, execution times were below one second, and for policies with up to 1000 rules, it was just over one second. This indicates that the model transformation approach is efficient enough for practical use.

Limitations and Future Work

Limitations:

  • The initial generation of the HOT might require some manual adaptation for highly domain-specific access-control languages.
  • The current prototype's request filtering primarily deals with string comparisons and could be extended to support other data types and integrate with Policy Information Points (PIPs) for retrieving attributes.

Future Work:

  • Extending the approach to manage systems with multiple, potentially related, access-control policies, including methods for deciding the appropriate transformation for a given request.
  • Investigating the benefits of experimental transformation execution modes like lazy evaluation and incremental propagation for access-control policy evaluation.
  • Applying model transformation validation and verification techniques to HOTs and generated ATL transformations to ensure correctness and potentially uncover policy flaws.

Summary updated

Abstract

International audience
BibTeX
@inproceedings{martinez2016runtime,
  author = {Salvador Martínez and Jokin García and Jordi Cabot},
  title = {{Runtime support for rule-based access-control evaluation through model-transformation}},
  booktitle = {SLE},
  year = {2016},
  url = {https://openreview.net/forum?id=xyQCvsmFJW},
}
Figure