HyRA-WSN: A Hybrid Reinforcement-Attention Model for Intelligent and Energy-Efficient Resource Allocation in Wireless Sensor Networks

Wireless Sensor Networks (WSNs) have emerged as a foundational technology enabling ubiquitous monitoring and control in diverse fields such as environmental surveillance, industrial automation, precision agriculture, and smart cities.HyRA-WSN, a new Hybrid Reinforcement-Attention model that aims to achieve optimal energy efficient resource allocation in large scale Wireless Sensor Networks (WSNs), is presented in this paper. Furthermore, traditional routing protocols may not be scaleable, may not easily enable node heterogeneity and may not address energy balancing in dynamic networks effectively. In order to overcome such shortcomings, to improve energy efficiency, and provide a utility-oriented strategy, HyRA-WSN introduces a Double Deep Q-Network (DDQN) reinforcement learning framework with a self-attention mechanism that assesses the significance of nodes using energy level, data significance, and congestion. The model encodes states of the environment in the form of features such as residual energy, the queue length of packets and status of neighbors to facilitate intelligent and adaptive decision making about routing, power control and duty cycling. Simulation results for different node densities and traffic conditions reveal that the HyRA-WSN outperforms pre-existing models including LEACH, HEED, Q-LEACH and RL-WSN. The proposed model will provide a packet delivery ratio (PDR) = 95.4%, network lifetime = 28.2% and energy savings = 46.8% over the conventional protocols. Moreover, end-to-end delay of HyRA-WSN is low average and it has high throughput even under high node mobility. These findings confirm the capability for intelligent, scalable, sustainable WSN operation for real-world deployments like environmental monitoring, industrial automation and precision agriculture.