Adaptive Dynamic Graph Modeling for Unmanned Aerial Vehicles (UAVs) Swarm Clustering and Path Selection Optimization
International Journal of Computer Networks and Applications · 30.06.2026 · 13 (3) · 655-680
DOI: 10.22247/ijcna/2026/34 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Computer Networks and Applications, online erschienen am 30.06.2026. Verfasst von Maneet Kaur und 2 weiteren Autor:innen.
Abstract (Originalsprache)
– Unmanned Aerial Vehicle (UAV) swarms are widely used in applications such as surveillance, disaster response, environmental monitoring, and intelligent communication systems. However, reliable communication, stable clustering, and efficient path selection remain challenging due to UAV mobility, energy limitations, and dynamic network topology changes. Existing methods, including SC, FSC, PSO, and recent GNN/GNA-based approaches, either rely on static graph structures or lack effective global optimization for stable and energy-aware clustering. To address these challenges, this paper proposes an Adaptive Dynamic Graph embedding-based Particle Swarm Optimization (ADG – PSO) framework for UAV swarm clustering and trajectory optimization. The proposed method integrates graph attention-driven spectral embedding with PSO-based cluster refinement. A graph attention mechanism first learns adaptive similarity relationships among UAVs to construct an attention-weighted Laplacian matrix, whose eigen decomposition generates dynamic clusters in a low-dimensional embedding space. PSO is then applied to refine cluster centroids, optimize cluster head selection, and improve clustering quality. Unlike conventional graph-based or GNN/GNA-based models, ADG – PSO combines adaptive graph learning, swarm intelligence optimization, and on-demand cluster maintenance within a unified framework, enabling balanced clustering and robust trajectory planning in dynamic environments. Simulation results show that ADG – PSO outperforms SC, FSC, GA, GNA, and PSO in terms of clustering stability, silhouette score, energy deviation, convergence speed, travel time, and mission completion rate. The proposed framework achieves improved energy efficiency, stable cluster formation, and a 100% mission completion rate, demonstrating its suitability for real-world UAV swarm applications.
Zitieren
APA
Kaur, M., Kaur, A., & Singh, P. (2026). Adaptive Dynamic Graph Modeling for Unmanned Aerial Vehicles (UAVs) Swarm Clustering and Path Selection Optimization. International Journal of Computer Networks and Applications, 13(3), 655-680. https://doi.org/10.22247/ijcna/2026/34
Vancouver
Kaur M, Kaur A, Singh P. Adaptive Dynamic Graph Modeling for Unmanned Aerial Vehicles (UAVs) Swarm Clustering and Path Selection Optimization. International Journal of Computer Networks and Applications. 2026;13(3):655-680. doi:10.22247/ijcna/2026/34
BibTeX
@article{Kaur2026,
author = {Kaur, Maneet and Kaur, Avinash and Singh, Parminder},
title = {Adaptive Dynamic Graph Modeling for Unmanned Aerial Vehicles (UAVs) Swarm Clustering and Path Selection Optimization},
journal = {International Journal of Computer Networks and Applications},
year = {2026},
volume = {13},
number = {3},
pages = {655--680},
doi = {10.22247/ijcna/2026/34}
}
EUVIMED