Construction of a Distributed Control Plane in SDN Based on Speculative Prediction and Random Early Migration Control

This study addresses the Switch Migration Problem (SMP) in distributed Software‐Defined Networking (SDN) control planes and proposes a speculative predictive control‐based load balancing approach. Conventional methods usually migrate switches only after controllers are overloaded, leading to response delays and network inefficiencies. To mitigate this issue, this study introduces a time‐series forecasting model based on the Adaptive Exponential Smoothing Moving Average (AESMA) method, enabling proactive switch migration through the forecasting of controller load variations. The proposed method adopts an orchestration‐based architecture, categorizing controller loads into three states: Green (no migration required), Yellow (stochastic migration), and Red (immediate migration). This adaptive approach minimizes unnecessary switch migrations while maintaining efficient load distribution. Additionally, by leveraging Little's Law for load estimation, the proposed method anticipates overload conditions and initiates switch migration preemptively. Performance evaluations confirm that the proposed method improves network throughput, reduces response time, enhances load balancing fairness, and stabilizes migration behavior under dynamic traffic conditions. These improvements are achieved through the synergistic integration of AESMA‐based prediction and Random Early Migration (REM)‐based probabilistic control, enabling both proactive responsiveness and migration stability.