BLOCKCHAIN AND AI-ENABLED EXAMINATION INTEGRITY VERIFICATION: A DECENTRALIZED FORENSIC READINESS FRAMEWORK USING DEEP LEARNING

Abstract

University examination systems represent high-value targets for sophisticated cyber threats, including identity fraud, result tampering, AI-assisted cheating, and insider log manipulation — attacks whose detection and legal prosecution require both real-time intelligence and forensically admissible, tamper-proof evidence chains. Existing frameworks address either blockchain-based integrity or AI-driven detection in isolation; none provide an integrated, examination-specific, legally defensible architecture. This paper presents DAIF (Decentralized AI-Integrated Forensic Readiness Framework), a novel six-layer architecture that fuses a permissioned Hyperledger Fabric blockchain, a five-model Deep Learning ensemble (FaceNet, LSTM-Keystroke Dynamics, BERT-based AI-text detection, pruned ResNet-18, and Graph Neural Network Sybil scoring), IPFS-based distributed forensic evidence storage, and RFC 3161-compliant cryptographic timestamping to produce legally admissible digital evidence aligned with India's DPDP Act 2023 and Bharatiya Sakshya Adhiniyam 2023. Evaluated on simulated 5,000-concurrent-user examination environments across three university campuses, DAIF achieves an ensemble detection accuracy of 97.2%, F1 score of 0.972, smart contract transaction throughput of 161–210 TPS, and 100% blockchain-verified evidence integrity up to 2,500 concurrent sessions. The framework establishes a new benchmark for examination integrity combining decentralization, deep intelligence, and forensic legal readiness.