DTR-GAT-OPAL: Digital Twin reinforcement learning and graph attention-based openflow prediction for secure cross-cloud SDN resource orchestration
Year: 2026
Type: article
Abstract: Intelligent resource orchestration in large-scale distributed computing systems can be facilitated via Software-Defined Networking (SDN)-enabled cross-cloud infrastructures, which allow programmability, elasticity, and service-centric administration. OpenFlow prediction accuracy, resource utilization efficiency, and Quality of Service (QoS) stability are significantly reduced by network heterogeneity, dynamic traffic fluctuations, controller congestion, and sophisticated cyberattacks. Optimization and learning-based techniques mostly estimate traffic or balance loads without topology-aware intelligence, adaptive decision-making, or synchronized cyber-physical representations for robust cross-cloud orchestration. DTR-GAT-OPAL (Digital Twin and Graph Attention Reinforcement Learning Framework for Intelligent OpenFlow Prediction and Secure Cross-Cloud Resource Orchestration) addresses these existing limitations. The proposed framework includes a Digital Twin (DT) layer for real-time synchronization of cyber-physical states, a Graph Attention Network (GAT) for topology-aware learning of flow representations, and a Proximal Policy Optimization (PPO)-based reinforcement learning engine for adaptive controller and virtual machine orchestration. Likewise included are a federated trust-security module and an uncertainty-aware QoS reasoning method for robust decision support under dynamic workloads. The same multi-objective optimization strategy maximizes service reliability and orchestration robustness while minimizing control-plane instability, migration overhead, energy consumption, carbon intensity, and security vulnerability. DTR-GAT-OPAL outperforms state-of-the-art methods in heterogeneous CloudSim-SDN and Mininet environments in OpenFlow prediction accuracy (0.99%), controller response latency (31.4%), energy consumption (34.2%), migration overhead (32.8%), and security assurance (29.6%). These results prove DTR-GAT-OPAL can orchestrate scalable, robust, and intelligent cross-cloud SDN resources for next-generation distributed computing ecosystems.
Source: Scientific Reports
Authors: S. Manikandan, R. Nadhiya Banu
Institution: E.G.S. Pillay Engineering College
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,690
OpenAlex ID: W7225617783
FWCI:
Cites:
Cited by:
Related to: