Enhancing network performance in IoT-WSNs through federated deep reinforcement learning-based intelligent routing

Abstract

By making it possible to collect and monitor data in real time, the installation of wireless sensor networks (WSNs) in the era of the internet of things (IoT) has brought about a transformation in a number of different industries. On the other hand, efficient data routing continues to be a significant challenge because of the dynamic nature of WSNs and the limitations of electronic devices connected to the IoT. This research presents an intelligent data routing technique that is based on federated deep reinforcement learning (FDRL), with the intention of optimizing data transmission in WSNs that are enabled by the IoT. In order to increase the efficiency of data routing, decrease latency, and extend the lifetime of the network, the model that has been recommended makes use of the collaborative learning skills of federated learning (FL) and the decision-making features of deep reinforcement learning (DRL). This is done in order to achieve these goals. It has been demonstrated through extensive simulations that the FDRL-based routing approach is superior to the methods that were previously utilized in terms of energy efficiency, the accuracy of data transfer, and the responsiveness to changes in the network. As a consequence of this, it is a solution that is conceivable for future applications including the IoT.