Verifiable Text Watermarking Detection to Improve Security

Zhangjie Fu, Xingming Sun, Jiangang Shu, Lu Zhou, Jin WangInternational Journal of Security and Its Applications 2014 · DOI

Abstract

Digital watermarking technology plays an important role in the areas of copyright protection and identity tracing for owners of digital mediums. At present, the security of the watermarking scheme is facing a great threat. The security of a digital watermarking scheme must not depend on the scheme being kept secret. Zero knowledge-based watermark detection scheme (ZKWD) can achieve this aim. For ZKWD scheme, an owner can provide proof to a verifier that a digital medium in question indeed contains the owner's watermark information without revealing any secret key and watermark-related information. However, the existing ZKWD protocols are still facing some challenging problems, such as ambiguity attacks. In this paper, a public ZKWD protocol is proposed for plain text, and the homomorphic property of asymmetric encryption algorithm in the multiplication operation is used to prevent the owner from cheating by ambiguity attacks. Compared with existing methods, the security of our proposed ZKWD scheme is improved by using the improved feature extraction algorithm.

Introduction

Digital watermarking is essential for proving ownership of digital media, but it faces a "catch-22" during legal disputes. In a traditional "symmetric" watermarking scheme, the owner must reveal their secret key to a verifier (like a judge or an untrusted third party) to prove the watermark exists. Once that key is revealed, the security is compromised; an untrusted verifier could use the key to remove the watermark entirely.

While "Zero-Knowledge" proofs—where you prove you know a secret without revealing the secret itself—offer a solution, existing methods are often vulnerable to ambiguity attacks. In these attacks, an adversary derives a "fake" watermark from a public work and claims they were the original creator. Furthermore, most ZKWD research has focused on images and video, leaving plain text—the most common medium on the internet—relatively unprotected.

The Proposed Framework

The authors propose a novel Zero-Knowledge Watermark Detection (ZKWD) protocol specifically designed for plain text. By combining chaotic mapping, synonym-based embedding, and homomorphic encryption, the system allows an owner to prove they authored a document without ever revealing the secret watermark.

Flowchart of the proposed watermarking and ZKWD protocol.

The pipeline consists of four primary stages:

  1. Robust Feature Extraction: Unique features are extracted from the plain text to serve as the "seed" for the watermark.
  2. Watermark Generation via Chaos: A logistic chaotic map transforms these features into a binary watermark sequence. Chaotic systems are highly sensitive to initial conditions, making it computationally impossible for an attacker to reverse-engineer the original key.
  3. Natural Language Embedding: To hide the watermark, the authors use a synonym substitution method based on "semantic adjacent words" to maintain readability while ensuring robustness.
  4. Public ZKWD Protocol: Using the homomorphic property of asymmetric encryption, a verifier can perform calculations on encrypted data to confirm the watermark's presence without seeing the raw watermark or secret key.

Conceptual Foundation: Zero-Knowledge Proofs

To understand how an owner can prove ownership without revealing the secret, it is helpful to consider the classic "Ali Baba Cave" analogy for Zero-Knowledge proofs.

The Ali Baba Cave analogy illustrating the interaction between a Prover (Peggy) and a Verifier (Victor).

In this analogy, Peggy (the Prover) wants to prove to Victor (the Verifier) that she knows the secret code to open a magic door inside a circular cave. Victor stands at the entrance. Peggy enters and goes to either path A or B. Victor then calls out a path for Peggy to emerge from. If Peggy knows the secret, she can always emerge from the requested path, even if she has to pass through the magic door. By repeating this many times, Victor becomes convinced Peggy knows the secret without Peggy ever revealing the code itself.

Security Analysis and Results

The paper provides a qualitative security analysis based on the three pillars of Zero-Knowledge proofs:

  • Completeness: An honest prover will always be able to convince an honest verifier.
  • Soundness: A cheating prover has only a $1/2$ chance of successfully faking a watermark in a single round. By repeating the protocol $n$ times, the probability of a successful cheat drops to $1/2^n$, which becomes infinitesimally small very quickly.
  • Zero-Knowledge: The verifier learns absolutely nothing about the watermark or the secret key during the process, other than the fact that the watermark exists.

Conclusion and Future Work

This research represents a pioneering application of ZKWD for plain text, which is notoriously difficult to watermark because even small changes can be glaringly obvious to a reader. By using chaotic mapping and semantic intelligence, the method ensures the watermark is both unique and mathematically "hard" to crack.

While the current protocol is interactive, future work aims to develop a non-interactive version. Such a system would allow an owner to publish a single "proof" that anyone could verify at any time without further input from the owner. They also plan to improve the system's resistance to "sensitivity attacks," where an attacker tries to break the watermark by making tiny, iterative changes to the text.

Summary updated

Abstract

Digital watermarking technology plays an important role in the areas of copyright protection and identity tracing for owners of digital mediums. At present, the security of the watermarking scheme is facing a great threat. The security of a digital watermarking scheme must not depend on the scheme being kept secret. Zero knowledge-based watermark detection scheme (ZKWD) can achieve this aim. For ZKWD scheme, an owner can provide prove to a verifier that a digital medium in question indeed contains the owner's watermark information without revealing any secret key and watermark-related information. However, the existing ZKWD protocols are still facing some challenging problems, such as ambiguity attacks. In this paper, a public ZKWD protocol is proposed for plain text, and the homomorphic property of asymmetric encryption algorithm in the multiplication operation is used to prevent the owner from cheating by ambiguity attacks. Compared with existing methods, the security of our proposed ZKWD scheme is improved by using the improved feature extraction algorithm.
BibTeX
@article{fu2014verifiable,
  author = {Zhangjie Fu and Xingming Sun and Jiangang Shu and Lu Zhou and Jin Wang},
  title = {{Verifiable Text Watermarking Detection to Improve Security}},
  journal = {International Journal of Security and Its Applications},
  year = {2014},
  doi = {10.14257/ijsia.2014.8.5.01},
}
Figure