Real-Time Intelligent Assignment of Controllers with Adaptive Load Balancing

Ensuring an adaptive and resilient Software Defined Network (SDN) is crucial for sustaining performance under dynamic load and faulty controllers. Most of the existing controller assignment methods use an evolutionary algorithm for optimal assignment of the switches of the failed controllers, considering latency and load distribution, thus lacking real-time adaptability. This article introduces a hybrid decision-making framework that integrates historical learning with real-time metrics to achieve intelligent assignment of switches for failed controllers. Using a modified KNN-based mechanism, the system identifies past configurations closest to the current network states. A selective fallback mechanism ensures immediate responsiveness, while an evolutionary algorithm refines assignments in the background when conditions diverge. This dual-path strategy enables both continuity and long-term adaptability. The proposed framework evolves over time by self-improving its assignment decisions. This framework has been evaluated and found to be more effective for real-time controller assignment with efficient load distribution in the presence of multi-controller failures.