RouteNet-Fermi: Network Modeling With Graph Neural Networks

Abstract

Network models are an essential block of modern networks. For example, they are widely used in network planning and optimization. However, as networks increase in scale and complexity, some models present limitations, such as the assumption of Markovian traffic in queuing theory models, or the high computational cost of network simulators. Recent advances in machine learning, such as Graph Neural Networks (GNN), are enabling a new generation of network models that are datadriven and can learn complex non-linear behaviors. In this paper, we present RouteNet-Fermi, a custom GNN model that shares the same goals as Queuing Theory, while being considerably more accurate in the presence of realistic traffic models. The proposed model predicts accurately the delay, jitter, and packet loss of a network. We have tested RouteNet-Fermi in networks of increasing size (up to 300 nodes), including samples with mixed traffic profiles -e.g., with complex non-Markovian modelsand arbitrary routing and queue scheduling configurations. Our experimental results show that RouteNet-Fermi achieves similar accuracy as computationally-expensive packet-level simulators and scales accurately to larger networks. Our model produces delay estimates with a mean relative error of 6.24% when applied to a test dataset of 1,000 samples, including network topologies one order of magnitude larger than those seen during training. Finally, we have also evaluated RouteNet-Fermi with measurements from a physical testbed and packet traces from a real-life network.

Reinforcement learning-based SDN routing scheme empowered by causality detection and GNN

“…Supervised learning algorithms (Rusek et al, 2019 ; Xie et al, 2019 ; Ferriol-Galm's et al, 2023 ) rely on labeled training datasets, where the model take network and traffic information as input to generate routing scheme. One major challenge of supervised learning methods is feature extraction, and the existing extraction methods generally perceive the network topological structure based on Graph Neural Network (GNN).…”

Oracle-based data generation for highly efficient digital twin network training

“…The recently proposed RouteNet-Fermi model [4], is a SOTA GNN model for KPI prediction. In this model, the network is described by its topology, low-level traf ic, routing and interface-level con iguration.…”

“…From the provided overview, we can see that all current SOTA solutions that address the networking KPI prediction problem are deep learning-based models. In the proposed solution, we used RouteNet-Fermi [4] as the base model to present the effectiveness of our novel data generation and ef icient training paradigm. This model was chosen as it is the current SOTA solution for the steadystate KPI prediction problem.…”

“…Typically, the quality of an ML-based model depends on two main factors: the model architecture and the training dataset. However, while the research community has shown signi icant progress in producing accurate and robust DT architectures [3,4,5], the data challenges have not yet been explored.…”

“…Particularly, the training set should enable the GNN model to effectively scale to large network topologies, much larger than the small simulated samples seen during training, potentially having different traf ic distribution. Participants are given a State-Of-The-Art (SOTA) GNN model, RouteNet-Fermi [4,8], for network performance evaluation and a packet-level network simulator, OMNeT++ [6], to generate the training dataset.…”