Usage control platformization via trustworthy SELinux

Masoom Alam, Jean-Pierre Seifert, Qi Li, Xinwen ZhangAsiaCCS 2008 · DOI

TL;DR

This paper tackles the challenge of enforcing usage control policies on distributed, untrusted platforms. They propose a platform architecture that leverages SELinux for mandatory access control and integrates hardware-based trust (TPM) and integrity measurement to create a trustworthy subsystem for policy enforcement. This approach aims to provide high assurance for data and service usage even when the data resides outside the owner's control.

What Problem Does This Paper Address?

The paper addresses the "usage control problem," which arises in increasingly distributed and heterogeneous computing environments. Unlike traditional access control in closed systems, usage control deals with enforcing policies on objects (data, services) that have been released to other, potentially untrusted, platforms. The core challenge is ensuring that a policy defined by a stakeholder (e.g., a data owner) is reliably enforced on a target platform, even when that platform is not under the stakeholder's direct control. Existing solutions like DRM or basic cryptography often fall short because they struggle with general attributes, dynamic conditions (like location or time), and, crucially, providing high assurance of enforcement on untrusted remote systems. The paper highlights the need for a trustworthy mechanism to enforce these policies in such environments.

Methods

The authors propose a platform architecture for trustworthy usage control enforcement. Key components and techniques include:

  • Trusted Subsystem: The core idea is to build a "virtually closed" and trusted subsystem on the target platform. This subsystem is designed to be tamper-proof and to reliably enforce usage control policies.
  • Hardware-based Root-of-Trust: They leverage a Trusted Platform Module (TPM) and a Core Root of Trust Measurement (CRTM) to establish a hardware-based root of trust. This is crucial for the integrity of the entire system.
  • Integrity Measurement and Verification: During the boot process and at runtime, critical components (BIOS, boot loader, OS kernel, reference monitor, client applications, policies) are measured. These integrity values are stored in the TPM's Platform Configuration Registers (PCRs). An Integrity Measurement Service (IMS) manages this process.
  • Mandatory Access Control (MAC) with SELinux: To enforce policies, they utilize SELinux, a MAC mechanism. SELinux allows for fine-grained labeling of subjects and objects, enabling the definition of strict policies that control interactions.
  • Policy Transformation: Usage control policies, often expressed in high-level formats like XACML, need to be translated into concrete SELinux policies. A policy transformation service handles this, ensuring that security properties are preserved.
  • Environmental Factors: The proposed architecture considers environmental factors (e.g., location, time) as part of the usage control policies, which are fed into the system via sensors.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Trustworthiness is Paramount: The paper strongly emphasizes that for usage control to be effective in distributed environments, the enforcement mechanism itself must be trustworthy. This goes beyond just authenticating users or objects; it requires assurance that the policy is being executed as intended.
  • Integrity as a Foundation: The authors highlight the importance of integrity verification not just for the data itself, but for all components involved in the policy enforcement chain, from the hardware root of trust to the policy transformation service.
  • SELinux as a Powerful Enforcement Tool: Leveraging SELinux for MAC provides a robust and fine-grained mechanism to enforce complex policies, especially when combined with integrity measurements.
  • Bridging High-Level Policies and Low-Level Enforcement: The policy transformation mechanism is a critical insight, enabling stakeholders to express policies in a more abstract and manageable way while still achieving precise enforcement on the target platform.
  • Beyond Traditional Access Control: The paper clearly articulates why traditional access control models and even DRM are insufficient for the dynamic and distributed nature of modern computing, necessitating a new approach like usage control.

Key Results

The paper presents a proposed architecture and design principles. While it doesn't detail specific quantitative results from a fully implemented system, the key qualitative findings are:

  • A comprehensive framework for building a trustworthy subsystem for usage control enforcement.
  • A clear set of design principles for such systems, including the need for high assurance, comprehensive policy models, MAC mechanisms, and policy transformation.
  • A proposed architecture that integrates hardware trust, integrity measurement, SELinux-based MAC, and policy transformation to meet these principles.

Limitations and Future Work

  • Implementation Prototype: The authors mention they are implementing a prototype system on a mobile reference platform. This implies that the full system's performance and effectiveness are still under evaluation.
  • Automated Policy Transformation: While a policy transformation mechanism is proposed, further work is needed on automating this process, especially in mobile computing environments.
  • Obligation Policies: The paper states they are investigating how to enforce "obligation policies" (actions that must be performed) within their architecture, indicating this is an area for future development.
  • Scalability and Performance: The paper doesn't delve into the performance implications or scalability of their proposed architecture, which would be crucial for real-world deployment.

Summary updated

BibTeX
@inproceedings{alam2008usage,
  author = {Masoom Alam and Jean-Pierre Seifert and Qi Li and Xinwen Zhang},
  title = {{Usage control platformization via trustworthy SELinux}},
  booktitle = {AsiaCCS},
  year = {2008},
  doi = {10.1145/1368310.1368345},
}
Figure