A Quantum-Inspired Dragonfly Optimization and Deep Reinforcement Learning Framework for Energy-Efficient Routing in Ad Hoc and IoT Networks

Abstract

This repository presents QADO-DRL, a novel energy-efficient routing framework for wireless ad hoc and IoT networks that combines Quantum-Inspired Adaptive Dragonfly Optimization (QADO) with a Dueling Double Deep Q-Network (D3QN) routing agent. QADO uses quantum-superposition initialization and quantum rotation-gate updates to generate high-quality candidate cluster-head and relay paths, while the D3QN agent adaptively selects among these candidates in real time, adjusting cluster-head selection, transmission power, and relay scheduling based on current network state. A Cross-Layer Energy Harvesting Module (CLEHM) further extends network lifetime by predicting and exploiting ambient RF, solar, and kinetic energy availability at each node. Evaluated across three simulation scenarios SynMobility, UbiSense-IoT, and DisasterNet spanning 100 to 500 heterogeneous nodes, QADO-DRL achieves substantial improvements in network lifetime, energy consumption per packet, and packet delivery ratio over established baselines including LEACH-C, EEHFR, DRL-EE, and Bio-EAODV, demonstrating strong scalability and robustness under dynamic mobility and disaster-response conditions. This repository includes the complete source code, datasets, and experiment pipeline needed to reproduce all reported results.