Low-Cost Anti-Copying 2D Barcode by Exploiting Channel Noise Characteristics

Ning Xie, Qiqi Zhang, Yicong Chen, Ji Hu, Gang Luo, Changsheng Chen

TL;DR

This paper introduces a novel Low-Cost Anti-Copying (LCAC) 2D barcode system that leverages differences in noise characteristics between legitimate and illegal copying channels. By embedding an authentication message and modeling the complex noise introduced by a double print-scan process, the LCAC system effectively detects direct copies and significantly increases the cost for sophisticated synthesized copying attacks.

What Problem Does This Paper Address?

The widespread use of 2D barcodes for applications like product authentication and mobile payments has unfortunately led to increased security risks, particularly the threat of illegal copying. Existing anti-copying methods often suffer from high costs (special materials/techniques), limited applicability, or high overhead (online databases). The core challenge is to create a robust and cost-effective way to prevent unauthorized replication of 2D barcodes, especially when attackers can use readily available off-the-shelf printers and scanners to create near-perfect replicas. This paper specifically targets the problem of preventing illegal replication of 2D barcodes.

Methods

The proposed LCAC system embeds a secret authentication message into the original 2D barcode data. This embedding is designed to be covert, meaning it doesn't significantly degrade the original barcode's readability. The key innovation lies in exploiting the fact that an "illegal channel" (involving multiple print-scan cycles) introduces significantly more noise and distortion than a "legal channel" (a single print-scan cycle).

The authors developed a theoretical model for this "illegal channel" noise using a generalized Gaussian distribution (GGD). This model was validated against experimental data from various printers and scanners. Based on this model, they devised a prediction function to optimize the embedding parameters (specifically, the size and error correction capability of the authentication message). This optimization aims to balance the covertness of the embedded message, the robustness of the original data, and the cost of both producing the barcode and performing an illegal copy.

Two embedding strategies were proposed: Strategy 1 embeds the authentication message into the redundant bits of the source message, while Strategy 2 embeds it into the entire source message bits. The paper analyzes the trade-offs between these strategies in terms of covertness, robustness, and security.

Datasets

The paper does not introduce new datasets. Instead, it relies on experimental data collected using a variety of commercially available printers (HP LaserJet P1108, FUJI P355D), scanners (BENQ K810, EPSON V330), and a mobile phone (HONOR V20). These devices were used to simulate both legal (single print-scan) and illegal (double print-scan) channels to gather data for modeling and evaluation.

Key Insights and Ideas

  • Exploiting Channel Noise: The fundamental idea is that the complex, multi-stage process of illegal copying (print-scan-print-scan) inherently introduces more noise and distortion than a legitimate print-scan process. This difference is the cornerstone of the LCAC system.
  • GGD for Illegal Channel Modeling: The generalized Gaussian distribution (GGD) proves to be an effective model for characterizing the complex noise introduced by the double print-scan (DPS) process, accurately matching experimental observations.
  • Trade-off Optimization: The paper emphasizes the importance of optimizing parameters not just for security, but also for a balance between production cost and the cost an attacker incurs to bypass the security.
  • Synthesized Copying (SC) Attack: The authors go beyond simple direct copying (DC) to consider a more sophisticated SC attack, where an attacker averages multiple copies to improve quality. Their LCAC system is designed to resist this as well.
  • Authentication Message as a "Canary": The authentication message's error rate is designed to be highly sensitive to the noise introduced by illegal copying. This makes it a more reliable indicator of an illegal copy than simply checking the source message's integrity.

Key Results

  • The proposed LCAC code, using two printers and two scanners, effectively detects direct copying (DC) attacks.
  • The system demonstrates resistance to synthesized copying (SC) attacks, requiring an attacker to access up to 14 legal copies to potentially bypass the authentication.
  • The theoretical GGD model for the illegal channel closely matches experimental results from various devices, validating the model's accuracy.
  • The parameter optimization strategy, based on the GGD model, successfully increases the cost for attackers to perform successful synthesized copying.
  • The LCAC approach offers better covertness (less visual modification) compared to existing methods like the Two-Level QR (2LQR) code.

Limitations and Future Work

  • Complexity of SC Attack Prediction: While the paper models SC attacks, it acknowledges that mathematically intractable expressions arise when summing multiple GGD random variables. Their approach uses approximations and curve fitting.
  • Parameter Optimization Trade-offs: The paper discusses the trade-off between production cost and copying cost, but the "optimal" balance might depend on specific application requirements.
  • Future Work:
    • Extending the LCAC concept to color barcodes.
  • Developing more advanced detection techniques for anti-copying barcodes.
  • Exploring machine learning or deep learning approaches to further enhance performance.
  • Investigating the impact of occlusion on different embedding strategies.

Summary updated

Abstract

In this paper, to overcome the drawbacks of prior approaches for defending against Illegally-Copying (IC) attacks, such as low generality, high cost, and high overhead, we propose a Low-Cost Anti-Copying (LCAC) 2D barcode by exploiting the difference between the noise characteristics of legal and illegal channels. An embedding strategy is proposed, and for a variant of this strategy, we also perform a corresponding analysis. To accurately evaluate the performance of our approach, a theoretical model of the noise in an illegal channel is established by using a generalized Gaussian distribution. By comparing the experimental results based on various printers, scanners, and mobile phones, it can be found that the sample histogram and fitting curve of the theoretical model match well, so it can be concluded that the theoretical model works well. To evaluate the security of the proposed LCAC code, in addition to the Direct-Copying (DC) attack, an improved version, called the Synthesized-Copying (SC) attack, is also considered in this paper. Based on the theoretical model, we build a prediction function to optimize the parameters of our approach. Parameters optimization seeks a tradeoff between the production cost and the cost of IC attacks. The experimental results show that the proposed LCAC code with two printers and two scanners can detect DC attacks effectively and resist SC attacks up to the access of 14 legal copies.
BibTeX
@article{xie2021low,
  author = {Ning Xie and Qiqi Zhang and Yicong Chen and Ji Hu and Gang Luo and Changsheng Chen},
  title = {{Low-Cost Anti-Copying 2D Barcode by Exploiting Channel Noise Characteristics}},
  journal = {IEEE Trans. Multimedia},
  year = {2021},
  url = {https://openreview.net/forum?id=8sUmVtxJ3f},
}
Figure