Introduction
In the realm of sensitive communications—such as medical consultations or legal discussions—traditional encryption methods often fall short. While they provide confidentiality, they frequently lack "deniability," a feature that allows a sender to plausibly deny having sent a message to a third party, even if the intended recipient knows it is authentic. This paper, published in IEEE Access, introduces a novel Deniable Authenticated Encryption (DAE) scheme designed to provide high security with significantly reduced computational overhead.
The Problem: Balancing Security and Privacy
Sensitive scenarios require a delicate balance of four security features:
- Confidentiality: Ensuring only the intended recipient can read the message.
- Integrity: Guaranteeing the message has not been altered.
- Deniable Authentication: Allowing the receiver to verify the sender's identity without being able to prove that identity to a third party (e.g., a judge).
- Non-repudiation for Content: Ensuring the receiver is certain the content is exactly what the sender intended.
Existing DAE schemes are often computationally expensive or require multiple interactive "handshakes," making them unsuitable for resource-constrained devices like wearable medical sensors.
System Architecture
The proposed system involves three primary entities: the medical sensor (sender), the data processing center (receiver), and a certification authority (CA) that manages public keys. This structure ensures that while the sender's identity is authenticated, the proof of that identity remains private between the two communicating parties.
The Proposed DAE Scheme
The authors developed a non-interactive DAE scheme that achieves all four security goals in a single mathematical operation. Grounded in the Computational Diffie-Hellman (CDH) assumption and the Discrete Logarithm (DL) problem, the scheme utilizes two random hash functions ($H_1$ and $H_2$) and modular exponentiation.
The core innovation is the "deniability trick": the mathematical structure allows a receiver to generate a ciphertext that is indistinguishable from one produced by the sender. Because the receiver could have forged the message, they cannot prove to an outsider that they didn't. However, since the receiver knows they did not forge it, they can trust the sender's identity.
Performance Evaluation
The scheme was benchmarked against the previous state-of-the-art (Li et al., 2016) using the MIRACL library and 1,000 real-world medical data samples. The results demonstrate significant efficiency gains across the board:
- Encryption Speed: Approximately 30% faster.
- Decryption Speed: Approximately 50% faster, driven by a 2/3 reduction in required modular multiplications.
- Bandwidth Efficiency: Ciphertext size was reduced by 33%.
These improvements become more pronounced as key lengths increase, suggesting the protocol is well-suited for future security standards and low-power IoT devices.
Conclusion
By streamlining the mathematical operations required for deniable authentication, the authors have created a protocol that is both highly secure and exceptionally efficient. While the reliance on a centralized Certification Authority remains a potential bottleneck for decentralized networks, the scheme offers a robust solution for privacy-preserving IoT applications and medical monitoring systems.
Summary updated