A Threshold-Cryptographic Framework for Anti-Leak Distribution of Digital Examination Papers
Abstract
This paper presents a threshold-cryptographic architecture for reducing the risk of premature leakage of digital examination papers during the interval between question-paper finalization and examination administration. The proposed design separates the data path from the control path. Examination content is encrypted using a fresh AES-256-GCM key, while the key is protected through envelope encryption under a key-release service. The capability to release that key is distributed using (k,n)-Shamir secret sharing across independent custodians, preventing any single custodian from unilaterally authorizing early release. At the scheduled release time, a quorum-based time authority provides an independently attested timestamp. Once the required time quorum and custodian threshold are satisfied, the key-release service reconstructs its private key within an HSM boundary, unwraps the examination key, and derives recipient-specific keys for individual examination centers. These keys are separately wrapped under each center's registered public key, limiting the impact of a compromise at any single examination center. The paper presents an actor and trust model, an explicit adversary model, a step-by-step release protocol, a threat-to-control security analysis, and a qualitative comparison with physical custody, blockchain-anchored distribution, and time-lock-puzzle-based timed-release cryptography. It also explicitly discusses residual risks, including custodian collusion, post-decryption optical or physical exfiltration, hardware and supply-chain trust, and compromise of the time-authority quorum. The architecture is presented as a research design rather than a claim of unconditional leak prevention. Future work includes implementing a prototype, evaluating quantitative performance, replacing reconstruct-and-zeroize key handling with threshold decryption, evaluating post-quantum key-encapsulation mechanisms, and conducting a formal mechanized security proof.
Community
0 commentsNo discussion yet
Be the first to share a question or observation.