The development of Internet of Things (IoT) devices leads to increased tasks and resource workload imbalance, which leads to a decrease in Quality of Service (QoS). In order to control resource workload, the use of Software-Defined Networking (SDN) provides efficient solutions for load-balancing with the aim of improving the QoS in IoT applications. Task growth can lead to overloading of some SDN controllers and reduced resource efficiency, necessitating the use of effective load-balancing approaches. In this research, a task migration-based load-balancing approach is presented, in which overloaded controller tasks are selected based on the least internal dependency and assigned to underloaded controllers with the most external dependency and the least communication cost.

The proposed method is implemented in the Mininet environment with the RYU controller and compared with the conventional SMSC, DLBM, DHA, and CAMD approaches. The evaluation criteria include response time, migration cost, migration count, energy consumption, and processor resource utilization in ARN, Atlanta, and NSF topologies. Simulation results show that the proposed method reduces the controller response time by 36.3%, migration cost by 43%, migration count by 26%, energy consumption by 21%, and resource utilization by 19.7% on average compared to the compared approaches. These results indicate that dependency-based task migration for load-balancing can effectively improve QoS parameters and increase user satisfaction in real IoT networks.