Access Control Encryption: Enforcing Information Flow with Cryptography

Ivan Damgård, Helene Haagh, Claudio Orlandi

Abstract

This paper introduces Access Control Encryption (ACE), a novel cryptographic primitive designed to enforce complex information flow policies, such as the Bell-Lapadula model's "no read-up" and "no write-down" rules. Unlike traditional encryption, which primarily manages receiver access, ACE controls both who can send and who can receive messages. By utilizing a semi-trusted, oblivious "sanitizer" to transform traffic, ACE ensures that unauthorized communication is mathematically blocked, even if senders and receivers collude. The authors provide formal security definitions and propose constructions based on both standard number-theoretic assumptions and advanced primitives like Indistinguishability Obfuscation.

Introduction

In traditional cryptographic systems, security is defined by the receiver's ability to decrypt a message. However, high-stakes environments—such as military or intelligence operations—require stricter control over the flow of information. Standard encryption effectively handles "No Read-Up" policies (preventing public users from reading top-secret data), but it fails to address "No Write-Down" policies, where a high-level user might attempt to leak secret data to a lower-level recipient. Historically, enforcing these restrictions required trusting the hardware or operating system of the sender. ACE aims to move this enforcement into the cryptographic layer itself, creating a "mathematical firewall."

Formal Security Framework

The core of ACE is defined by two fundamental security properties: the No-Read Rule and the No-Write Rule. These are formalized through cryptographic games where an adversary attempts to distinguish between different ciphertexts or message sources.

The No-Read Rule formalizes the requirement that unauthorized receivers cannot extract information from ciphertexts.

The No-Read Rule ensures that unless a receiver is explicitly authorized by the policy to read from a specific sender, the ciphertext remains indistinguishable from random noise. Conversely, the No-Write Rule prevents unauthorized senders from successfully transmitting information to a receiver, even if they attempt to use valid encryption keys.

The No-Write Rule ensures that unauthorized senders cannot leak information to receivers, even through collusion.

Methodology

The authors introduce a central, oblivious party known as the Sanitizer (San). All communication in an ACE system must pass through this Sanitizer, which performs a mathematical transformation on every ciphertext. Crucially, the Sanitizer does not need to know the underlying security policy, the identity of the participants, or the content of the messages; it simply applies a randomized algorithm that "refreshes" valid traffic and destroys unauthorized traffic.

Two primary constructions are proposed:

  1. Linear ACE: Built on standard assumptions like Decisional Diffie-Hellman (DDH) and the Paillier cryptosystem. This version is practical for systems with a limited number of roles, as ciphertext and key sizes grow linearly with the number of security levels.
  2. Polylogarithmic ACE: A theoretically scalable construction using Indistinguishability Obfuscation (iO) and Sanitizable Functional Encryption (sFE). This allows for efficient handling of millions of roles with minimal overhead.

Key Insights and Results

  • Oblivious Enforcement: The Sanitizer acts as a blind gatekeeper. Because it is oblivious, it cannot leak message content, yet its mathematical operations are sufficient to enforce the global flow policy.
  • Collusion Resistance: The system is designed to be robust against groups of malicious users. Unless the central policy explicitly allows a path between a specific sender and receiver, no information can be transferred between them.
  • Feasibility: The paper proves that ACE is achievable using current cryptographic tools, making it a viable candidate for specialized secure communication networks.

Limitations and Future Directions

While ACE offers a powerful new paradigm, it relies on the physical assumption that all traffic must pass through the Sanitizer. If a user can bypass this bottleneck, the "No-Write" guarantee is lost. Additionally, the most efficient constructions are currently vulnerable to quantum attacks, and the highly scalable polylogarithmic version relies on iO, which remains computationally expensive for real-world deployment. Future research may focus on post-quantum ACE constructions and distributing the Sanitizer's role via Multi-Party Computation (MPC) to eliminate single points of failure.

Summary updated

BibTeX
@inproceedings{damgard2016access,
  author = {Ivan Damgård and Helene Haagh and Claudio Orlandi},
  title = {{Access Control Encryption: Enforcing Information Flow with Cryptography}},
  booktitle = {TCC (B2)},
  year = {2016},
  doi = {10.1007/978-3-662-53644-5\_21},
}
Figure