Scholarlink

<p>Network obfuscation is widely applied to protect sensitive layouts such as power grids and communication infrastructures from visual inspection and automated attacks. Traditional evaluations emphasize scrambling metrics that produce layouts with visual distortion. However, this creates a critical paradox: deterministic methods such as Arnold maps or cyclic shifts achieve near-perfect scrambling scores yet are reversible, providing no true privacy.</p><p>This study introduces a new evaluation framework that moves beyond simple scrambling metrics. It assesses three key dimensions: visual scrambling reliability, privacy reliability (resistance to attacks), and reversibility screening. Reversibility is treated as a non-negotiable failure mode: if a method can be reversed, it is considered unreliable regardless of other scores.</p><p>We evaluate ten obfuscation methods on four benchmark networks, including Zachary network and the three power grids, using Monte Carlo simulations and heatmap visualizations. Results confirm the paradox: reversible methods dominate in scrambling and global scores when reversibility is ignored, but collapse once reversibility is enforced. By contrast, non-reversible methods such as jittering, random displacement, and differential privacy perturbation emerge as reliable privacy-preserving strategies, even if their scrambling performance is moderate.</p><p>Our contributions are threefold: (1) a systematic framework that distinguishes aesthetic obfuscation from genuine privacy protection; (2) an empirical demonstration of the obfuscation paradox across multiple networks; and (3) the introduction of a reliability perspective that aligns obfuscation evaluation with principles of failure analysis. This work clarifies why visual scrambling alone is insufficient and establishes a rigorous foundation for selecting non-reversible obfuscation methods in privacy-sensitive applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Beyond visual scrambling: a reliability-oriented evaluation framework for non-reversible network obfuscation

  • 期刊论文

  • 发表日期：

    2026年2月4日

查看全文
  • Claudio M. Rocco

摘要

Network obfuscation is widely applied to protect sensitive layouts such as power grids and communication infrastructures from visual inspection and automated attacks. Traditional evaluations emphasize scrambling metrics that produce layouts with visual distortion. However, this creates a critical paradox: deterministic methods such as Arnold maps or cyclic shifts achieve near-perfect scrambling scores yet are reversible, providing no true privacy.

This study introduces a new evaluation framework that moves beyond simple scrambling metrics. It assesses three key dimensions: visual scrambling reliability, privacy reliability (resistance to attacks), and reversibility screening. Reversibility is treated as a non-negotiable failure mode: if a method can be reversed, it is considered unreliable regardless of other scores.

We evaluate ten obfuscation methods on four benchmark networks, including Zachary network and the three power grids, using Monte Carlo simulations and heatmap visualizations. Results confirm the paradox: reversible methods dominate in scrambling and global scores when reversibility is ignored, but collapse once reversibility is enforced. By contrast, non-reversible methods such as jittering, random displacement, and differential privacy perturbation emerge as reliable privacy-preserving strategies, even if their scrambling performance is moderate.

Our contributions are threefold: (1) a systematic framework that distinguishes aesthetic obfuscation from genuine privacy protection; (2) an empirical demonstration of the obfuscation paradox across multiple networks; and (3) the introduction of a reliability perspective that aligns obfuscation evaluation with principles of failure analysis. This work clarifies why visual scrambling alone is insufficient and establishes a rigorous foundation for selecting non-reversible obfuscation methods in privacy-sensitive applications.