EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Intelligent Routing for MANETs Using Deep Neural Networks and Multi-Metric Decision Models

Basheer A. Al-Qassab, Diary R. Sulaiman

Zanco Journal of Pure and Applied Sciences · 30.06.2026 · 38 (3) · 169-185

DOI: 10.21271/zjpas.38.3.13 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Zanco Journal of Pure and Applied Sciences, online erschienen am 30.06.2026. Verfasst von Basheer A. Al-Qassab und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Mobile Ad Hoc Networks (MANETs) demand efficient routing to ensure reliable communication under dynamic conditions. Traditional routing protocols like DSDV rely on static hop-count metrics, often neglecting crucial factors such as link quality, congestion, and node energy, leading to suboptimal performance in terms of packet delivery and delay. This paper introduces a Machine Learning-based Multi-Objective Routing Protocol (MLMORP), a proactive routing framework that integrates a pre-trained Deep Neural Network (DNN) into the routing decision process. MLMORP periodically broadcasts lightweight beacon messages to collect multi-metric features – including residual energy, data rate, signal strength, queue length, and node degree – from neighbors. These features are fed into the DNN to predict a route cost, enabling each node to select next hops based on a learned composite metric rather than simple hop-count. This work details the design of MLMORP’s routing architecture and DNN model, and formulates the multi-objective optimization problem it addresses. Extensive OMNeT++/INET simulations evaluate MLMORP against a hop-count-based routing protocol under various network loads. The results demonstrate that MLMORP consistently achieves higher throughput (~24% gain on average), lower end-to-end delay (~27% reduction), improved packet delivery ratio (~21% improvement under heavy traffic), and reduced energy consumption (up to ~39% savings) compared to traditional routing protocol. These gains underscore the novelty of employing a fully embedded DNN-driven, multi-metric routing framework in MANETs marking a significant advancement over previous ML-based approaches that did not integrate learning models into the core routing decision process.

Zitieren

APA

Al-Qassab, B. A., & Sulaiman, D. R. (2026). Intelligent Routing for MANETs Using Deep Neural Networks and Multi-Metric Decision Models. Zanco Journal of Pure and Applied Sciences, 38(3), 169-185. https://doi.org/10.21271/zjpas.38.3.13

Vancouver

Al-Qassab BA, Sulaiman DR. Intelligent Routing for MANETs Using Deep Neural Networks and Multi-Metric Decision Models. Zanco Journal of Pure and Applied Sciences. 2026;38(3):169-185. doi:10.21271/zjpas.38.3.13

BibTeX
@article{AlQassab2026,
  author = {Al-Qassab, Basheer A. and Sulaiman, Diary R.},
  title = {Intelligent Routing for MANETs Using Deep Neural Networks and Multi-Metric Decision Models},
  journal = {Zanco Journal of Pure and Applied Sciences},
  year = {2026},
  volume = {38},
  number = {3},
  pages = {169--185},
  doi = {10.21271/zjpas.38.3.13}
}
Kostenlos starten →