Hybrid Post-Quantum Secure Wireless Healthcare Network with Audit Logging and AI-Based Intrusion Detection

Abstract

Remote patient monitoring over Wireless Body Area Networks (WBANs) continuously transports physiological measurements such as heart rate, SpO₂, blood pressure and temperature over wireless links that may be exposed to eavesdropping, replay, manipulation, injection and flooding attacks. Long-lived healthcare records also require protection against future quantum attacks on classical public-key mechanisms. This work presents a hybrid security framework that combines NIST-standardized ML-KEM for post-quantum key encapsulation, ML-DSA for digital signatures, AES-256-GCM for authenticated encryption, HMAC-SHA256 for packet-level integrity verification, a blockchain-inspired hash-chained audit ledger for tamper evidence, and a Random Forest intrusion detection subsystem for behavioral anomaly detection. The prototype integrates a Python TCP gateway, SQLite-based storage with role-based access control, a Streamlit clinician dashboard, and a React/Node.js web client. Functional evaluation covers six security test scenarios involving valid traffic, ciphertext manipulation, replay, signature forgery, malformed input, and packet flooding. The reported prototype measurements are 452 ms for ML-KEM key generation, 318 ms for ML-DSA key generation, 94 μs for AES-256-GCM encryption of a 512-byte payload, 88 ms average gateway processing time, 147 ms end-to-end latency, 342 packets/s sustained throughput, 96% packet delivery ratio, 87% five-class IDS accuracy, and 100% verified audit-chain integrity. A supplementary Random Forest experiment on the BLEWBAN dataset achieved 76.9% accuracy for twelve-class sensor-position classification. The study explicitly distinguishes functional prototype results from production security claims and identifies limitations related to the PQC backend, synthetic attack traffic, hardware platform, scalability and formal security analysis.