Energy-Efficient Drone Communication Networks Using Hybrid K-Means and Adaptive Particle Swarm Optimization

Abstract

The Article proposes the design, development, and assessment of a novel energy-efficient drone communication network optimization framework that integrates K-Means clustering and Adaptive Particle Swarm Optimization (PSO). The solution addresses the crucial task of optimizing drone locations in dynamic settings with the dual objectives of maximizing network coverage and QoS while minimizing overall network energy. The focus is on battery-powered drone communications where energy efficiency is essential. This proposed methodology adopts a three-step process: (1) spatial grouping of mobile users with K-Means analysis for determining the best service areas, (2) dynamic PSO optimization for optimizing smart UAV positioning with adaptive exploration-exploitation rates, and (3) online re-optimization for handling mobile users and environmental dynamics. An integrated simulation platform with an interactive six-tab GUI has also been developed for simulating and assessing all six key performance aspects of this model: energy, coverage, network capacity, delay, and support for Quality of Service and active UAV. Experimental outputs indicate remarkable improvements over baseline solutions. "The algorithm provides 25-40% energy saving, 15-30% network coverage improvement, 25-37% relative throughput improvement, and 35% faster convergence time" than existing optimization algorithms." Adaptive optimization of parameters through linearly decreasing inertia weights and balancing cognitive and social parameters avoids local optimal points and enables full exploration of the solution space. Performance of the solution works effectively in small-scale and large-scale networks, and energy per user increases with the network size. The major contributions of this work are: (1) a hybrid optimization strategy combining efficiency and effectiveness, (2) dynamic parameter tuning of the PSO algorithm for faster convergence, (3) a multi-dimensional evaluation strategy for effective assessment, and (4) realistic deployment advice regarding optimal cluster sizes, frequency of re-optimization, and parameters. The proposed work offers both theoretical analysis and practical implementation advice and has a number of applications in emergency, smart, and rural networks and could enjoy the benefits of a heterogeneity-based strategy of spatial analysis and swarm intelligence for the next generation of drone networks.