DeepNT: Path-Centric Graph Neural Networks for Network Tomography
Keywords: Network Tomography, Graph Structure Learning
Abstract: Network tomography is a crucial problem in network monitoring, where observable path measurements are used to infer unmeasured network properties, making it essential for tasks such as route selection, fault diagnosis, and traffic control. However, most existing methods assume that the network topology is fully known—an assumption that rarely holds in practice. The incomplete topology introduces significant challenges in extracting path information for predicting path performance. Furthermore, these approaches are typically designed for a single path performance metric and lack the flexibility to handle multiple metrics that may be of interest in a single network. To address these limitations, we propose Deep Network Tomography (DeepNT), a new framework that simultaneously infers the adjacency matrix and predicts unmeasured paths by leveraging Graph Neural Networks (GNNs) as backbones to learn path-centric end-node pair representations under incomplete network topology. To ensure that the learned adjacency matrix aligns with the characteristics of real-world networks, we propose a novel learning objective that constrains the model in terms of connectivity, sparsity, and path performance bounds, enabling robust generalization across a variety of performance metrics. Extensive experiments on real-world and synthetic datasets demonstrate the superiority of DeepNT in predicting performance metrics and inferring graph topology compared to state-of-the-art methods.
Primary Area: learning on graphs and other geometries & topologies
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 2052
Loading