ECC-EXONUM-eVOTING: Enhancing Secure E-Voting with AI-Based Fraud Detection and Offline Voting Support

Abstract

As the world transitions toward digital-first governance and civic participation, ensuring the security and integrity of voting systems has become a critical concern. Traditional evoting mechanisms, although convenient, suffer from a range of vulnerabilities - including voter impersonation, double voting, identity leaks, and tampering by insiders or external adversaries. ECC-EXONUM-eVOTING was previously proposed to mitigate many of these issues through elliptic curve cryptography (ECC), Zero-Knowledge Proofs (ZKP), and Exonum private blockchain. In this paper, we extend the capabilities of ECC-EXONUMeVOTING by integrating two novel modules aimed at enhancing both system intelligence and accessibility. First, we implement an AI-based fraud detection system using unsupervised anomaly detection techniques that proactively identify and block fraudulent voting behaviors in real time. Second, we introduce a secure offline voting architecture designed for voters in remote or lowconnectivity regions, using QR-based tokenization and Merkle-root-based integrity proofs for delayed blockchain synchronization. Through simulations, algorithmic validation, and comparative analysis, we demonstrate how these enhancements significantly increase the robustness, scalability, and real-world applicability of blockchain-based voting systems.