An End-to-End Intelligent Control Architecture for 6G Networks: A Dual-SDN Approach With DRL, QNN and Cognitive Interface

The forthcoming 6G networks demand ultra-reliable, low-latency, and context-aware orchestration of heterogeneous resources across both radio access and core domains. This paper introduces an end-to-end (E2E) intelligent control architecture that integrates two cooperative Software-Defined Networking (SDN) planes: the SDN-enabled Radio Access Network (SDN-RAN) and the SDN-based Core Network (SDN-CN), connected through a cognitive East–West interface. Within the SDN-RAN, a Deep Reinforcement Learning (DRL) agent built upon Proximal Policy Optimization (PPO) and Graph Neural Networks (GNN)—optimizes dynamic resource allocation and adaptive access mode selection (NOMA/RSMA). In parallel, the SDN-CN, empowered by a Quantum-enabled RAN Intelligent Controller (Q-RIC), leverages DRL and Quantum Neural Networks (QNN) for virtualized core function orchestration and predictive decision-making. A digital twin continuously monitors and simulates network dynamics to ensure real-time validation and proactive control. Experimental results demonstrate that the proposed architecture achieves seamless service continuity, proactive load balancing, and intelligent task migration toward Multi-access Edge Computing (MEC) resources under stringent 6G requirements.