FRLTC: A new topology control algorithm using federated reinforcement learning for software-defined wireless sensor network in IoT

Topology is one of the most important factors influencing the performance of software-defined wireless sensor networks (SDWSN) in the Internet of Things (IoT). This paper presents a novel topology control algorithm using federated reinforcement learning, proposed for SDWSN in the IoT. Our approach represents the SDWSN as a federated learning system in which each sensor node runs a reinforcement learning model to change its communication range such that the node degree approaches the desired value. Another reinforcement learning model was applied to the SDN controller to assign the sensor node for learning, bringing the average degree of the entire network closer to the desired value. Simulation results reveal that the proposed approach outperforms well-known topology control algorithms in terms of the desired node degree, energy consumption, and quality of transmission.