|
Volume 11, Number 3, July 2026 , pages 249-275 📚 VSI DOI: https://doi.org/10.12989/acd.2026.11.3.249 |
||
|
Optimized secure routing with intrusion resilience via deep reinforcement learning for cyber threat detection |
||
Jayalakshmi Sambandam, S. Malathi, L. Logeshwari, J. Dillibabu, Lithin Kumble
|
||
| Abstract | ||
| With the increasing number of networked devices, elevated network speeds, and sophisticated cyber threats, cybersecurity has become a difficult problem to solve. Many current approaches fall short due to inefficient data preprocessing, redundant features, low adaptability, and failure to recognize complex behavioral patterns. To address these challenges, an intelligent hybrid framework is proposed to deal with cybersecurity problems in smart networks. An Intelligent Adaptive Routing–Density Fusion Model (IARDFM) is utilized for efficiently collecting network traffic data and preprocessing it to remove noise, redundant, and outlying data while increasing quality and balance of the data. The preoptimized features are then processed by a Deep Probabilistic Cluster-Aware Behavior Learning Network (DPCB-Net) that utilizes deep learning with probabilistic clustering to perform accurate multiclass behavioral learning. A Spectral Hierarchical Swarm Feature Optimization Framework (SHSFOF) is then used to select the most relevant features for training a learning-based approach. To detect various types of cyber-attacks, a Graph Reinforced Intrusion Detection System (GRIDS) is proposed based on graph-based learning and trust-aware evaluation. A Deep Reinforcement Learning–Based Intrusion-Resilient Secure Routing (DRL-ISR2) framework is used for selectively routing packets to different destinations to prevent intrusion and increase network resilience. Comprehensive experiments and comparisons validate improved accuracy, precision, recall, and robustness of the intelligent hybrid framework for smart network cybersecurity. | ||
| Key Words | ||
| adaptive routing; cyber threat detection and prevention; density-based clustering; feature optimization and selection techniques; graph-based intrusion detection systems; probabilistic learning methods; reinforcement learning techniques for network security problems; trust evaluation models | ||
| Address | ||
| Jayalakshmi Sambandam: Department of Computer Science and Engineering (Emerging Technologies), SRM Institute of Science and Technology (SRM IST), Vadapalani Campus, Chennai – 600026, Tamil Nadu, India S. Malathi: Department of Computer Science, St. Thomas College of Arts and Science, Chennai – 600107, Tamil Nadu, India L. Logeshwari: Department of Computer Science with Data Science, Tagore College of Arts and Science, Chennai – 600044, Tamil Nadu, India J. Dillibabu: Department of Computer Applications, St. Thomas College of Arts and Science, Chennai – 600107, Tamil Nadu, India Lithin Kumble: School of Computing Science and Engineering, REVA University, Bengaluru – 562157, Karnataka, India | ||
