Inference Attacks on Encrypted Online Voting via Traffic Analysis

2025-09-19

AI Summary

This work exposes a novel privacy threat in encrypted online voting systems arising from network traffic metadata—specifically packet sizes, timing patterns, and directionality—enabling adversaries to infer, without decryption, whether a voter cast a ballot, when they voted, and whether their ballot is valid, thereby compromising ballot secrecy and electoral integrity. We propose the first systematic metadata side-channel attack model that integrates rule-based reasoning with machine learning, achieving 99.5% inference accuracy on two widely used voting platforms. To counter this threat, we introduce a lightweight defense mechanism based on payload padding and timestamp equalization, empirically reducing attack success rates significantly. Our study establishes a new evaluation paradigm and a reproducible benchmark methodology for assessing the practical privacy guarantees of end-to-end encrypted voting systems.

Technology Category

• Humans and AI: Voting
• Machine Learning: Privacy
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Online voting enables individuals to participate in elections remotely, offering greater efficiency and accessibility in both governmental and organizational settings. As this method gains popularity, ensuring the security of online voting systems becomes increasingly vital, as the systems supporting it must satisfy a demanding set of security requirements. Most research in this area emphasizes the design and verification of cryptographic protocols to protect voter integrity and system confidentiality. However, other vectors, such as network traffic analysis, remain relatively understudied, even though they may pose significant threats to voter privacy and the overall trustworthiness of the system. In this paper, we examine how adversaries can exploit metadata from encrypted network traffic to uncover sensitive information during online voting. Our analysis reveals that, even without accessing the encrypted content, it is possible to infer critical voter actions, such as whether a person votes, the exact moment a ballot is submitted, and whether the ballot is valid or spoiled. We test these attacks with both rule-based techniques and machine learning methods. We evaluate our attacks on two widely used online voting platforms, one proprietary and one partially open source, achieving classification accuracy as high as 99.5%. These results expose a significant privacy vulnerability that threatens key properties of secure elections, including voter secrecy and protection against coercion or vote-buying. We explore mitigations to our attacks, demonstrating that countermeasures such as payload padding and timestamp equalization can substantially limit their effectiveness.

Problem

• Examining encrypted traffic analysis threats to online voting privacy
• Inferring voter actions from metadata without decrypting content
• Testing attacks on proprietary and open-source voting platforms

Innovation

• Traffic analysis on encrypted voting
• Rule-based and machine learning methods
• Payload padding and timestamp equalization