Work

Optimizing Energy-Cognitive Data Routing in Wireless Sensor Networks Using a Hybrid DRL-SI Approach with Levy Flight-Tuned Red Deer Optimization

Year: 2026

Type: article

Abstract: Energy efficiency and site reliability in data provision are two issues that are pinpointed as vital concerns to Wireless Sensor Networks (WSNs) because of limited resources and dynamic environment. A multimodel is hybrid Deep Reinforcement Learning (DRL) and Swarm Intelligence (SI) with a Levy Flight-tuned Red Deer Optimization (LFRDO) algorithm is proposed to meet these challenges. The DRL component is a tool used in real-time routing decisions depending on network states, e.g. residual energy and link quality. The SI module (Particle Swarm Optimization and Ant Colony Optimization) is applied in long term, global optimization of paths. The deployment of the LFRDO algorithm is made to provide exploration functions of the SI and to avoid an early convergence. An intensive simulation is conducted to check the model and comparisons are made with the A-Leach, Pegasus and DSO-EHO protocols. The results also show that the proposed approach is properly optimized and has an energy conservation value of 98%, 96 % packet delivery ratio and a network throughput of 0.9 Mbps in a 500-node network. End to end delay is also reduced. The paper notes that the proposed hybrid DRL-SI-LFRDO framework demonstrates significant improvement compared to the existing methods and can be an optimal robust and energy-cognitive routing solution to be used in the challenging WSNs that require, e.g., environmental monitoring and industrial automation.

Source: Journal of Robotics and Control (JRC)

DOI: 10.18196/jrc.v7i3.26189