2025, Vol. 6, Issue 2, Part A
Post-quantum federated anomaly detection for zero-trust storage: A graph neural network framework with GDPR-compliant differential privacy
Author(s): Hasan Jameel Azooz
Abstract: Zero-trust storage architectures require continuous verification of access requests, yet traditional centralized anomaly detection systems face quantum vulnerabilities and violate data sovereignty principles. This paper presents Post-Quantum Federated Anomaly Detection for Zero-Trust PQFAD-ZT, a novel framework that integrates post-quantum cryptography (CRYSTALS-Dilithium), federated Graph SAGE learning, and Rényi differential privacy to detect Advanced Persistent Threats (APTs) while maintaining data locality. Our approach addresses three critical gaps: quantum-resistant authentication for federated updates, privacy-preserving graph-based anomaly detection, and GDPR compliance for cross-border data processing. Through comprehensive evaluation on CICIDS-2017 and Edge-IIoTset datasets with 1,000 federated clients, PQFAD-ZT achieves an F1-score of 0.923 (±0.012) with ? = 1.18 differential privacy guarantee, reducing mean-time-to-detect by 28% compared to centralized baselines while maintaining communication overhead below 42MB per round. Theoretical analysis provides formal security proofs under Module-LWE assumptions and (?,?)-differential privacy guarant A comprehensive GDPR compliance mapping demonstrates adherence to Articles 5, 25, and 32 requirements.
DOI: 10.33545/27075923.2025.v6.i2a.99Pages: 26-33 | Views: 65 | Downloads: 30Download Full Article: Click Here
How to cite this article:
Hasan Jameel Azooz.
Post-quantum federated anomaly detection for zero-trust storage: A graph neural network framework with GDPR-compliant differential privacy. Int J Circuit Comput Networking 2025;6(2):26-33. DOI:
10.33545/27075923.2025.v6.i2a.99
Research Journals