Scholarlink

<p>Quantum computing threatens cryptographic techniques in cyber-physical systems (CPS), especially in vital infrastructures like smart grids, hospitals, and industrial automation. This study integrates post-quantum cryptography (PQC) with deep learning-based intrusion detection to strengthen CPS resilience against conventional and quantum cyber attacks. Graph Convolutional Network (GCN) and Variational Autoencoder (VAE) models for real-time anomaly detection are used in the hybrid framework, together with quantum-resistant lattice-based (Kyber), code-based (McEliece), and hash-based (SPHINCS+) cryptographic techniques. The system learns to adapt to changing assault patterns while preserving computational efficiency in resource-constrained contexts. Based on KDD Cup 1999 and UNSW-NB15 datasets, the framework achieves 99.7% reconstruction accuracy, 96% precision, 94% recall, and a 95% F1-score with 18.7 ms latency and 2.4&#xa0;mJ energy usage. Instead of treating cryptography and intrusion detection separately, the hybrid PQC–DL model provides an integrated, quantum-resilient, and adaptive security architecture to guard against quantum cyber attacks.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing cyber-physical system security: a hybrid post-quantum cryptography and deep learning framework for resilience against quantum cyber threats

  • 期刊论文

  • 发表日期：

    2026年6月17日

查看全文
  • Divya Priya Degala,
  • Senthil Athithan

摘要

Quantum computing threatens cryptographic techniques in cyber-physical systems (CPS), especially in vital infrastructures like smart grids, hospitals, and industrial automation. This study integrates post-quantum cryptography (PQC) with deep learning-based intrusion detection to strengthen CPS resilience against conventional and quantum cyber attacks. Graph Convolutional Network (GCN) and Variational Autoencoder (VAE) models for real-time anomaly detection are used in the hybrid framework, together with quantum-resistant lattice-based (Kyber), code-based (McEliece), and hash-based (SPHINCS+) cryptographic techniques. The system learns to adapt to changing assault patterns while preserving computational efficiency in resource-constrained contexts. Based on KDD Cup 1999 and UNSW-NB15 datasets, the framework achieves 99.7% reconstruction accuracy, 96% precision, 94% recall, and a 95% F1-score with 18.7 ms latency and 2.4 mJ energy usage. Instead of treating cryptography and intrusion detection separately, the hybrid PQC–DL model provides an integrated, quantum-resilient, and adaptive security architecture to guard against quantum cyber attacks.