DDRL‐AqP MSPP: Double Deep Reinforcement Learning With Aquila Pelican Optimization Based Energy Hole Prediction for Mobile Sink Path Planning

ABSTRACT A wireless sensor network is a collection of spatially distributed sensor nodes that wirelessly communicate to collect and transmit data. Static sink‐based routing involves sending data from sensor nodes to a fixed base station (sink node). Sensor nodes on the data path can run out of energy quickly, especially those closer to the sink, and uneven energy consumption may lead to the premature failure of nodes. Mobile sink path planning involves moving the sink node to different locations to collect data that balances energy consumption, extending network lifetime, improved scalability, and enhanced fault tolerance. Still, energy‐efficient routing is a challenging task. Thus, the mobile sink path planning (MSPP) by predicting the energy hole using the double deep reinforcement learning (DDRL) is introduced in this research, wherein the optimal action selection by considering the remaining energy and distance is employed using the Aquila pelican optimization (AqP) algorithm. The proposed AqP algorithm is designed by integrating the Pelican optimization‐based solution updation with the Aquila Optimization algorithm for enhancing the convergence rate. The proposed AqP‐DDRL MSPP accomplished delay, network lifetime, packet delivery ratio, residual energy, and throughput with 1.61 ms, 99.98%, 99.30%, 0.99 J, and 255.31 kbps respectively.