Efficient Real-Time Integrity Auditing With Privacy-Preserving Arbitration for Images in Cloud Storage System

Xin Tang, Yongfeng Huang, Chin-Chen Chang, Linna ZhouIEEE Access 2019 · DOI

Abstract

Cloud storage provides an inexpensive and effective means for the storage and management of images, which in turn occupy a huge proportion and are usually stored in an archived mode. Considering the security and efficiency requirements of cloud images, an efficient real-time integrity audit scheme is in urgent need. However, existing solutions cannot be directly applied since they do not take the characteristics of cloud images into account and thus take enormous computations, communications, and storage to generate, transfer, and store authentication data. Moreover, the result of auditing cannot be used as evidence to prove the guilt of cloud service provider since the verifier whom is specified by the client may hide its misbehavior. Reversible watermarking is a potential way to achieve lightweight real-time audit for cloud images without introducing permanent distortion. Nevertheless, existing algorithms cannot provide stable capacity for authentication data of fixed length. In addition, it entails security problems once it is used to solve the fairness problem. This paper proposes an efficient real-time integrity audit scheme specific to cloud images with fair arbitration support. The scheme is based on the presented adaptive reversible watermarking algorithm which provides a fixed embedding capacity for images to embed authentication data. To address fairness problem under the proposed mechanism, we adopt Diffie-Hellman key exchange scheme to design a new challenge-response protocol under the established simplified consensus mechanism, so that replay attack resistance and privacy-preserving fair arbitration are achieved. The security analysis and performance evaluation show that the proposed scheme has provable security and greatly enhanced efficiency comparing with the state of the art.

Introduction

As cloud storage becomes the standard for managing massive image archives, ensuring data integrity remains a critical challenge. Cloud Service Providers (CSPs) are not fully trusted entities; data corruption can occur due to hardware failures, management negligence, or external attacks. Traditional integrity auditing often falls into two categories: periodic epoch-based auditing, which may discover corruption too late, or high-overhead real-time auditing, which requires significant external metadata (tags or signatures) that consume bandwidth and storage.

Furthermore, resolving disputes between clients and CSPs introduces a "fairness" problem. A client might falsely claim corruption to seek compensation, while a CSP might attempt to hide a breach. While a Third-Party Arbitrator (TPAR) can resolve these issues, granting them full access to private images creates significant privacy risks.

Proposed Methodology

The proposed framework addresses these challenges through two main innovations: adaptive reversible watermarking and a privacy-preserving arbitration protocol.

Adaptive Reversible Watermarking

To eliminate the need for external metadata, the scheme uses reversible watermarking to embed authentication data directly into the image pixels. Unlike traditional methods where embedding capacity varies with image content, the authors developed an algorithm using Discrete Cosine Transform (DCT) coefficients. By reordering high-frequency coefficients and employing a median-based expansion technique, the system achieves a fixed embedding capacity of 26 bits per 8x8 block.

The scheme utilizes a two-tier security approach:

  • Type 1 Watermark: Embedded in every block to allow for precise tampering localization.
  • Type 2 Watermark: A global watermark for the entire image, enabling a rapid "yes/no" integrity check.

Privacy-Preserving Arbitration Protocol

To ensure fair dispute resolution without compromising privacy, the authors designed a challenge-response protocol based on Diffie-Hellman key exchange and bilinear pairings.

The system model illustrates the interaction between Cloud Clients, the Cloud Service Provider (CSP), and the Third-Party Arbitrator (TPAR) during the auditing and arbitration process.

When a dispute arises, the TPAR issues a random challenge. The prover (either the Client or the CSP) generates a proof based on image features and this challenge. The TPAR can then mathematically verify the proof against the original signature. Due to the complexity of the Discrete Logarithm Problem (DLP), the TPAR cannot reverse-engineer the proof to access the actual image content, ensuring privacy.

Experimental Evaluation

The scheme was evaluated using the Wang Dataset (1,000 low-resolution JPEG images) and the Oxford5k Dataset (5,062 high-resolution images) to test scalability and performance.

Key Results

  • Performance: Auditing a 600KB image took approximately 88.1 ms, significantly faster than the 244 ms required by previous state-of-the-art methods like Hwang’s scheme. This represents a 1.4x to 3.8x speedup.
  • Visual Quality: The watermarked images maintain high visual fidelity before restoration, with a Peak Signal-to-Noise Ratio (PSNR) exceeding 28.5 dB for low-resolution and 37.6 dB for high-resolution images.
  • Security: Formal analysis confirms the scheme's resistance to replay attacks, where an attacker might attempt to use an old valid proof for a corrupted image.

Conclusion and Future Work

By integrating authentication data directly into image content, this scheme eliminates the storage and bandwidth overhead associated with traditional metadata tags. The reversible nature of the watermark ensures that clients receive bit-perfect original images upon verification.

While the global integrity check is highly efficient, tampering localization remains computationally linear with image size. Future research will focus on extending this framework to multi-replica cloud storage environments and exploring identity-based sharing scenarios to further enhance cloud image security.

Summary updated

BibTeX
@article{tang2019efficient,
  author = {Xin Tang and Yongfeng Huang and Chin-Chen Chang and Linna Zhou},
  title = {{Efficient Real-Time Integrity Auditing With Privacy-Preserving Arbitration for Images in Cloud Storage System}},
  journal = {IEEE Access},
  year = {2019},
  doi = {10.1109/access.2019.2904040},
}
Figure