Energy-Efficient Routing in Software-Defined Wireless Sensor NetworksUsing Genetic Algorithms

Abstract

Energy-efficient routing remains a major challenge in Software-Defined Wireless Sensor Networks (SDWSNs),where existing approaches typically optimize routing paths without explicitly integrating routingdecisions into SDN flow table generation. This paper proposes a multi-objective Genetic Algorithm (GA)-based framework that formulates routing as a flow table optimization problem executed at the SDN controller.The proposed framework evolves network-wide next-hop assignments using feasibility-preservingcrossover and mutation operators combined with tournament selection and elitist replacement. A normalizedmulti-objective fitness function jointly optimizes energy consumption, end-to-end delay, packet deliveryratio (PDR), and hop count using Analytical Hierarchy Process (AHP)-derived weights. The proposed approachis evaluated through extensive OMNeT++ simulations using IEEE 802.15.4 MAC modeling acrossnetwork sizes ranging from 200 to 2000 sensor nodes. Experimental results over 30 independent runsdemonstrate that the proposed framework significantly outperforms shortest-path routing, LEACH, andGA-energy baselines in terms of network lifetime, delivery reliability, throughput, and latency. In particular,the framework reduces dead-node progression by up to 50.5%, improves packet delivery ratio by13%, increases throughput by 50%, and decreases end-to-end delay by 40% compared with shortest-pathrouting. Scalability analysis further shows stable performance under large-scale deployment conditions,retaining 76% of peak energy efficiency at 2000 nodes. The results indicate that GA-driven flow table optimizationprovides a scalable, transparent, and reproducible alternative to reinforcement learning-basedrouting approaches for energy-aware SD-WSNs.