ElectS: A secure web-based e-voting system with biometric authentication

Abstract

Sri Lanka’s elections face high administrative costs, low turnout among older and migrant voters, and persistent trust deficits. This work presents ElectS, a secure, open e-voting system that couples real-time facial verification (AWS Rekognition, with a 90% similarity threshold) with National Identity Card (NIC) checks to ensure single-ballot eligibility. Compared with prior internet-voting pilots that reported challenges with ballot secrecy and voter authentication, ElectS emphasizes auditability and operational security. Built on the MERN stack with a modular back end and an intuitive web interface, ElectS provides separate dashboards for administrators, candidates, and voters; supports time-bounded elections; and includes integrated complaint intake and feedback workflows. Privacy is embedded by design: facial images are processed transiently and not stored; only the embeddings necessary for matching are retained. All stored artefacts (embeddings and any NIC images for manual review) are encrypted at rest and transmitted over TLS, governed by least-privilege access controls with comprehensive audit logging, and subject to data-minimisation and retention policies (NIC images removed within 14 days of verification). Functional testing across 50 scenario cases demonstrated correct tallying and timely complaint handling, while usability testing indicated high responsiveness and ease of use. Planned enhancements include blockchain-based audit trails, remote voting, and multilingual interfaces. Overall, ElectS indicates that biometric-enabled e-voting can reduce costs, improve accessibility and trust, and scale to the needs of elections in developing contexts.