Reinforcement learning-based dynamic routing using mobile sink for data collection in WSNs and IoT applications

Energy is one of the most critical resources for sensor devices that decides the network lifetime of the wireless sensor networks. In many circumstances, sensor devices consume more energy for data transmission, reception, and forwarding operations. The major challenge is to increase the network lifetime by implementing the latest research models to reduce the deployment and operational cost. Many existing methods address the application of the static sink with multi-hop routing. But most of them suffer from energy-hole issues and inefficient data collection due to the early death of sensor nodes. Most of the existing methods of learning require massive data with feature engineering which eventually increases the learning complexity. In order to avoid these issues, a robust reinforcement learning-based mobile sink model is proposed for dynamic routing with efficient data collection. In addition, the Q-Learning approach is implemented to induce automatic learning through the shortest route. Combining these strategies preserves network stability and efficiently improves the routing performance as well as the reward. The simulation results reveal that the proposed reinforcement learning-based mobile sink model extends the network lifetime, provides an improved learning time with more reward, and results in high efficiency when compared with existing methods.