Intelligent Post-Quantum Cryptography Deployment in Enterprise Linux Infrastructure Using Machine Learning

Abstract

Enterprise Linux infrastructure forms the backbone of global digital operations powering banks, hospitals, government systems, and cloud platforms. The cryptographic algorithms protecting these systems, primarily RSA and Elliptic Curve Cryptography, face an existential threat from quantum computers. Algorithms that take classical computers billions of years to break can be defeated by sufficiently powerful quantum machines in hours, using Shor's algorithm. Despite this known threat, most enterprise Linux deployments have no migration plan, largely because the task of identifying, replacing, and validating cryptographic assets across thousands of servers is too complex to be done manually at scale. This paper presents an intelligent, machine-learning-driven framework for deploying Post-Quantum Cryptography (PQC) across enterprise Linux infrastructure. The proposed system ML-PQC combines workload-aware PQC algorithm selection, reinforcement learning-based parameter tuning, automated crypto inventory scanning, and zero-downtime orchestrated rollout using Ansible and Puppet. The ML engine uses an ensemble of XGBoost and Random Forest classifiers trained on system telemetry to recommend optimal NIST-standardized PQC algorithms (CRYSTALS-Kyber, CRYSTALS-Dilithium, SPHINCS+) for each server's specific workload profile. Evaluated on a 500-node testbed spanning RHEL 9 and Ubuntu 22.04 systems, ML-PQC achieves a 95.6% algorithm selection accuracy, reduces PQC deployment time across 5,000 nodes from 640 hours (manual) to 52 hours, and delivers a composite security score of 95 out of 100. Cryptographic overhead is kept within practical bounds, with TLS handshakes completing in 47 ms. The framework is validated for FIPS 140-3, CIS Benchmark Level 2, and NIST Cybersecurity Framework compliance. This work provides a practical, deployable pathway for enterprise Linux administrators to achieve quantum safety today.