Deep Learning on Graphs for Mobile Network Topology Generation

2025-04-18

AI Summary

Automating accurate modeling of dynamic neighbor relationships (mobility edges) in mobile network topologies prior to hardware deployment remains challenging. Method: This paper introduces graph neural networks (GNNs) to automated real-world telecom topology generation for the first time, proposing an end-to-end learning framework that jointly leverages ANR measurement data and base station configuration. It innovatively designs a geography-aware edge-sampling heuristic, imposing distance constraints to reduce graph scale and training overhead while preserving modeling fidelity. Results: On operational network datasets, the proposed GNN model achieves over 12% higher accuracy and precision than an MLP baseline; incorporating the distance-aware heuristic further improves precision by 18% and reduces training time by 37%. This work overcomes the static limitations of conventional manual, rule-based neighbor relation configuration, establishing an interpretable, deployable, data-driven paradigm for intelligent 5G/6G network planning.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Planning, Routing, and Scheduling: Planning with Language Models
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Mobile networks consist of interconnected radio nodes strategically positioned across various geographical regions to provide connectivity services. The set of relations between these radio nodes, referred to as the emph{mobile network topology}, is vital in the construction of the networking infrastructure. Typically, the connections between radio nodes and their associated cells are defined by software features that establish mobility relations (referred to as emph{edges} in this paper) within the mobile network graph through heuristic methods. Although these approaches are efficient, they encounter significant limitations, particularly since edges can only be established prior to the installation of physical hardware. In this work, we use graph-based deep learning methods to determine mobility relations (edges), trained on radio node configuration data and reliable mobility relations set by Automatic Neighbor Relations (ANR) in stable networks. This paper focuses on measuring the accuracy and precision of different graph-based deep learning approaches applied to real-world mobile networks. We evaluated two deep learning models. Our comprehensive experiments on Telecom datasets obtained from operational Telecom Networks demonstrate the effectiveness of the graph neural network (GNN) model and multilayer perceptron. Our evaluation showed that considering graph structure improves results, which motivates the use of GNNs. Additionally, we investigated the use of heuristics to reduce the training time based on the distance between radio nodes to eliminate irrelevant cases. Our investigation showed that the use of these heuristics improved precision and accuracy considerably.

Problem

• Predicting mobile network topology edges using graph deep learning
• Evaluating accuracy of GNNs and MLPs on real telecom data
• Reducing training time with distance-based heuristic methods

Innovation

• Graph-based deep learning for mobility relations
• Training on radio node and ANR data
• Heuristics to reduce training time