Double Q‐learning‐based adaptive trajectory selection for energy‐efficient data collection in wireless sensor networks

Summary This paper proposes a novel distributed stochastic routing strategy using mobile sink based on double Q‐learning algorithm to improve the network performance in wireless sensor network with uncertain communication links. Furthermore, in order to extend network lifetime, a modified leach‐based clustering technique is proposed. To balance the energy dissipation between nodes, the selected cluster head nodes are then rotated based on the newly suggested threshold energy value. The simulation results demonstrate that the proposed algorithms outperform the QWRP, QLMS, ESRP and HACDC in terms of network lifetime by 18.33%, 35.1%, 39.7% and 44.7%, respectively. Moreover, the proposed algorithms considerably enhances the learning rate and hence reduces the data collection latency. The IoT‐based WSN network is considered as an environment and the cluster heads as states. Mobile sink acts as an agent, gathers state information from the environment, and finds the optimal path for data collection using the proposed double Q‐learning‐based algorithm. According to the proposed reward function, the environment rewards the mobile sink for the computed path. The above process is repeated to determine the best trajectory for the MS.