Graph Neural Networks for Network Topology Optimization
Dr. Luca Bianchi, Dr. Elena Rossi, Prof. Marco Ferrari
Politecnico di Milano
Abstract
This paper presents a graph neural network (GNN) approach for optimizing network topology in large-scale mobile networks. By representing the network as a graph with base stations as nodes and interference relationships as edges, our GNN model learns to predict optimal resource allocation strategies. The approach reduces interference by 35% and increases network capacity by 22% compared to traditional optimization methods.
AI Summary
- GNN-based approach for network topology optimization treating base stations as graph nodes.
- Reduces inter-cell interference by 35% compared to traditional methods.
- Increases overall network capacity by 22%.
- Scales to networks with thousands of base stations.
Key Findings
- 1Graph representation naturally captures the spatial relationships in mobile networks.
- 2Message passing between GNN layers mimics distributed coordination between cells.
- 3The model adapts to topology changes without retraining.
Industry Implications
Enables real-time network reconfiguration for dynamic 6G environments.
Reduces operational costs through automated topology management.
Applicable to heterogeneous networks combining terrestrial and non-terrestrial nodes.
Read the Original Paper
Access the full paper on arXiv for complete methodology, results, and references.
Open on arXivRelated Papers
Federated Reinforcement Learning for Distributed Network Optimization
Stanford University — 8 citations
AI/ML PapersTransformer-Based Channel Estimation for Massive MIMO Systems
Tsinghua University — 12 citations
AI/ML PapersNeural Architecture Search for Efficient Edge AI in Wireless Networks
Samsung AI Center Seoul — 5 citations