Predictive Multi-Weight Switch Migration for Distributed SDN Controllers in IoT

Abstract

The expansion of IoT ecosystems surges control-plane traffic, causing uneven load distribution in distributed Software-Defined Networking (SDN). Existing reactive load-balancing methods activate only after congestion has set in, leading to higher latency. To address these limitations, a proactive, traffic prediction-driven load-balancing framework is implemented. A load-sharing mechanism selectively synchronizes controller workload information with peer controllers to support proactive migration decisions. A decision-making module then determines optimal balancing actions using predicted load conditions. Finally, a proactive switch migration strategy reallocates switches from overloaded controllers to underloaded ones ahead of congestion peaks. Extensive simulations validate the effectiveness of the proposed approach. The method minimized switch migrations to only two, compared to five existing strategies. Furthermore, overall controller load balance improved by 6.59%, demonstrating the effectiveness of the proposed framework. These findings demonstrate that the solution significantly enhances the responsiveness of distributed SDN-based IoT networks.