The Scheme for Recharging Nodes in Surveillance Wireless Sensor Networks Using Dynamic Resource Location

Sensor nodes in wireless sensor networks are dispersed at random and are used to keep tabs on whatever is of interest. Due to the restricted battery capacity of each sensor, it might be difficult to keep the network running for continuous observation. Using a mobile charger to recharge the sensor wirelessly is becoming a viable option for powering the sensors. However, because to uncertainty in the networks, choosing an effective charging route for the MC is difficult. Due to unpredictability in the network architecture, such as node failures, the pace at which the sensors use energy may vary dramatically. Because of these modifications, the relative significance of the various sensors, often treated as equivalent in previous publications, varies. We suggest a novel adaptive pricing mechanism in this study, one that makes use of deep reinforcement learning (DRL) to address these problems. In specifically, the MC uses a charging technique that decides which sensor should be charged next depending on the overall network's health. We then use a deep neural network, taught with RL techniques, to determine appropriate values for the price variables. Our model is flexible enough to accommodate changes in the architecture of a network. Results show that the proposed method greatly outperforms the best existing ondemand alternatives.