Eliminating Mixnet Overhead in Blockchain Voting: A Scalable Privacy-Preserving Protocol Using Zero-Knowledge Proofs, Homomorphic Encryption, and Selective Metadata Mixing

Abstract

Blockchain-based voting systems provide transparency and auditability but introduce significant privacy risks due to publicly observable metadata. Existing approaches rely on mixnets or heavy cryptographic primitives to achieve anonymity, resulting in high computational overhead and limited scalability. In this paper, we propose a novel privacy-preserving voting protocol that eliminates the need for full ciphertext mixnets by introducing a selective metadata mixing mechanism. Our protocol combines zero-knowledge proofs for vote validity, homomorphic encryption for confidential aggregation, and randomized metadata transformations to achieve unlinkability. We formalize security properties including ballot secrecy, unlinkability, and end-to-end verifiability, and prove security under standard cryptographic assumptions. We further provide a gas-aware smart contract model and evaluate scalability for elections with one million voters under Layer-2 rollup deployment. Our results show that the proposed protocol reduces anonymization complexity from O(n log n) to O(n) while maintaining strong privacy guarantees.