A Comparative Evaluation of Machine Learning Models for Forwarding Decisions in Delay-Tolerant Networks: A Unified Framework Running head: ML Models for DTN Forwarding
Abstract Routing in Delay-Tolerant Networks (DTNs) requires forwarding decisions under intermittent connectivity and without global topology knowledge. While machine learning has been applied to DTN routing, many studies evaluate a single model in isolation. This paper presents a unified framework that compares four machine-learning models — Decision Tree, Random Forest, XGBoost, and Support Vector Machine — for the DTN forwarding decision under identical conditions, using multiple metrics and computational cost. The contribution is methodological rather than algorithmic: we do not propose a new routing algorithm but establish a controlled, reproducible procedure for comparing existing models. Models are trained on node-encounter traces from the ONE simulator, with statistical significance testing across multiple runs, and the learned policy is integrated into the simulator and compared with classical protocols. Results indicate that tree-based models, particularly XGBoost, provide the best balance of effectiveness and computational cost under the evaluated conditions.
Authors
- Hasan Omar Ali
- Mohamed Farag Agleiwan
Publication Details
- Journal
- Research Square
- Published
- 2026-11-17
- DOI
- https://doi.org/10.21203/rs.3.rs-11333462/v1
- Primary Topic
- Opportunistic and Delay-Tolerant Networks
- Type
- preprint