Post quantum blockchain framework using probabilistic hidden state deep learning for smart IoT systems
In recent years, smart IoT systems pose significant challenges regarding security, scalability, and intelligent decision-making for IoT data, particularly amid advances in quantum computing attacks. Traditional cryptographic and machine learning techniques seem insufficient for real-time IoT systems where data is dynamic and large-scale and security requirements are strict. In this paper, a novel Post-Quantum Probabilistic Hidden-State Deep Learning (PQP_HS_DL) framework is presented that comprises of effective probabilistic hidden state modelling, lattice-based post-quantum cryptography and blockchain technology to process IoT data securely and efficiently. In the proposed PQP_HS_DL framework, a probabilistic hidden state model is applied to learn temporal dynamics and uncertainty in IoT data streams to provide enhanced prediction and reliable anomaly detection capabilities. A lattice-based cryptographic scheme ensures quantum-resistant security, and blockchain provides data integrity, transparency, and decentralized trusted authority management. The system is further enhanced by edge computing to alleviate latency and realize real-time processing performance. The experimental evaluation of the proposed framework is carried out under 100 IoT nodes to evaluate its performance. The results of the PQP_HS_DL provide a high classification accuracy (97.6%), and higher precision, recall, and F1-score compared to the existing techniques. The latency (72 ms) is lower, the throughput (285 transactions per second) is higher, and energy consumption (0.91) is also effective in the PQP_HS_DL framework for real-time applications of IoT. Security analysis shows that the entropy (0.98) is very high, and the attack probability (0.01) is very low. The framework uses lattice-based post-quantum cryptographic mechanisms, which are effective against any classical attacker as well as against existing quantum cryptanalytic methods, based on standard computational assumptions. Scalability analysis demonstrates that the proposed framework, evaluated with run on IoT networks with over 1000 nodes, exhibits significant performance.