Abstract:Today's network is notorious for its complexity and uncertainty. Network operators often rely on network models for efficient network planning, operation, and optimization. The network model is responsible for understanding the complex relationships between network performance metrics (e.g., delay and jitter) and network characteristics (e.g., traffic and configuration). However, we still lack a systematic approach to developing accurate and lightweight network models that are aware of the impact of network configurations (i.e., expressiveness) and provide fine-grained flow-level temporal predictions (i.e., granularity). In this paper, we propose xNet, a data-driven network modeling framework based on graph neural networks (GNN). It is worth noting that xNet is not a dedicated network model designed for a specific network scenario with constraint considerations. On the contrary, xNet provides a general approach to modeling the network characteristics of concern with relation graph representations and configurable GNN blocks. xNet learns the state transition functions between time steps and rolls them out to obtain the full fine-grained prediction trajectory. We implement and instantiate xNet with three use cases. The experimental results show that xNet can accurately predict different performance metrics (i.e. temporal and steady-state QoS) in different scenarios, with performance comparable to state-of-the-art domain-specific models. Compared with traditional packet-level simulators, xNet achieves a speed improvement of more than two orders of magnitude, demonstrating its promising application in real-time optimization of network configurations.

EAGLE: Heterogeneous GNN-based Network Performance Analysis.

Hybrid Performance Modeling And Analyzing Of Parallel Systems

NGAT: Attention in Breadth and Depth Exploration for Semi-Supervised Graph Representation Learning

Xnet: Modeling Network Performance with Graph Neural Networks

CEA-GNN: Performance Prediction of Hybrid Network Traffic with Deterministic and Best Effort

Deeper and Wider Networks for Performance Metrics Prediction in Communication Networks

Hyperbolic Benchmarking Unveils Network Topology-Feature Relationship in GNN Performance

Performance Analysis of Graph Neural Network Frameworks

Traffic Analysis of Partially Observable Bearer Network under RouteNet Based GNN Model

Heterogeneous graph neural networks analysis: a survey of techniques, evaluations and applications

Performance Analysis in Wireless Heterogeneous Networks Based on Stochastic Geometry Theory

HTNet: Dynamic WLAN Performance Prediction using Heterogenous Temporal GNN

Learning Load Balancing with GNN in MPTCP-Enabled Heterogeneous Networks

STAG: Enabling Low Latency and Low Staleness of GNN-based Services with Dynamic Graphs

Addressing Heterogeneity and Heterophily in Graphs: A Heterogeneous Heterophilic Spectral Graph Neural Network

Hetero^2Net: Heterophily-aware Representation Learning on Heterogenerous Graphs

Xnet: Improving Expressiveness and Granularity for Network Modeling with Graph Neural Networks

Heterogeneous Graph Propagation Network

Are we really making much progress?: Revisiting, benchmarking and refining heterogeneous graph neural networks

GNN at the Edge: Cost-Efficient Graph Neural Network Processing over Distributed Edge Servers

DeepComNet: Performance evaluation of network topologies using graph-based deep learning