(2025). Conditional Invertible Neural Networks for Traffic Matrix Estimation from Link Loads. In 2025 IEEE International Mediterranean Conference on Communications and Networking (MeditCom) (pp. 1–6). IEEE. https://doi.org/10.1109/meditcom64437.2025.11104479
Περίληψη
Traffic matrices (TMs) are essential to many network management and optimization tasks. However, direct measurement of traffic flows is impractical, and a more viable alternative is traffic matrix estimation (TME) via network tomography, inferring TMs from readily available link load measurements and routing information. In this paper, we address this ill-posed linear inverse problem by leveraging conditional invertible neural networks (cINNs), a recent neural architecture that introduces conditioning into its building blocks to approximate the posterior conditional distribution of TMs given the observed link loads. Specifically, we explore two alternative conditioning mechanisms: directly integrating raw link loads or incorporating extracted features generated by a dedicated conditioning neural network. The trained cINN enables efficient sampling from the posterior conditional distribution, allowing accurate TM estimation via either computing the conditional mean or iteratively solving a minimization problem to ensure data fidelity with the link measurements. The proposed TME approach is evaluated on a dataset obtained from a real-world backbone network and compared against three baseline methods across a comprehensive set of performance metrics. A prototype implementation is publicly available under an open-source license.
- DOI
- 10.1109/meditcom64437.2025.11104479
- Τύπος
- Άρθρο σε Πρακτικά Συνεδρίου
- Έτος
- 2025
Σύνδεσμοι
BibTeX
@inproceedings{kakkavas2025conditional,
title = {Conditional Invertible Neural Networks for Traffic Matrix Estimation from Link Loads},
author = {Grigorios Kakkavas and Petros Maratos and Vasileios Karyotis and Symeon Papavassiliou},
url = {https://doi.org/10.1109/meditcom64437.2025.11104479},
doi = {10.1109/meditcom64437.2025.11104479},
year = {2025},
date = {2025-01-01},
booktitle = {2025 IEEE International Mediterranean Conference on Communications and Networking (MeditCom)},
pages = {1–6},
publisher = {IEEE},
abstract = {Traffic matrices (TMs) are essential to many network management and optimization tasks. However, direct measurement of traffic flows is impractical, and a more viable alternative is traffic matrix estimation (TME) via network tomography, inferring TMs from readily available link load measurements and routing information. In this paper, we address this ill-posed linear inverse problem by leveraging conditional invertible neural networks (cINNs), a recent neural architecture that introduces conditioning into its building blocks to approximate the posterior conditional distribution of TMs given the observed link loads. Specifically, we explore two alternative conditioning mechanisms: directly integrating raw link loads or incorporating extracted features generated by a dedicated conditioning neural network. The trained cINN enables efficient sampling from the posterior conditional distribution, allowing accurate TM estimation via either computing the conditional mean or iteratively solving a minimization problem to ensure data fidelity with the link measurements. The proposed TME approach is evaluated on a dataset obtained from a real-world backbone network and compared against three baseline methods across a comprehensive set of performance metrics. A prototype implementation is publicly available under an open-source license.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}