Quality-of-service class specific traffic matrices in ip/mpls networks

Schnitter, Stefan; Hartleb, Franz; Horneffer, Martin

doi:10.1145/1298306.1298341

Cited by 8 publications

(5 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We then scale the inter-arrival times to match the values in the traffic data. Additionally, we used a distribution of source-destination flows from a real internet service provider [54] to map flows to different ToS classes. Our dataset includes 4 real-world network topologies, including one previously used (GEANT) and three new topologies that the model has never seen before (ABILENE, NOBEL-GBN, and GERMANY50).…”

Section: B Generalization and Scalabilitymentioning

confidence: 99%

RouteNet-Fermi: Network Modeling With Graph Neural Networks

Ferriol-Galmés¹,

Paillissé²,

Suárez‐Varela³

et al. 2023

IEEE/ACM Trans. Networking

View full text Add to dashboard Cite

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.

show abstract

Section: B Generalization and Scalabilitymentioning

confidence: 99%

RouteNet-Fermi: Network Modeling With Graph Neural Networks

Ferriol-Galmés¹,

Paillissé²,

Suárez‐Varela³

et al. 2023

IEEE/ACM Trans. Networking

View full text Add to dashboard Cite

show abstract

“…Then, we scale these inter-arrivals according to the values in the traffic matrices. Regarding the mapping of sourcedestination flows to ToS classes, we follow the same distribution from a real ISP [58].…”

Section: Experiments With Real Trafficmentioning

confidence: 99%

“…Finally, we evaluate TwinNet using a dataset that combines the 106 real network topologies previously used (Section 7.4), and with traffic matrices that follow the inter-arrivals times and ToS classes mentioned used in the previous experiment [57,58]. Fig.…”

Section: Experiments With Real Trafficmentioning

confidence: 99%

Building a Digital Twin for Network Optimization Using Graph Neural Networks

Ferriol-Galmés¹,

Suárez-Varela²,

Paillise³

et al. 2021

SSRN Journal

View full text Add to dashboard Cite

“…To ensure a smooth change of the ability vector, it is necessary to adjust the time interval, during which the traffic matrix can be considered as unchanged. Previous studies have shown that the time interval is from 5 to 15 minutes [12].…”

Section: Qos-aware Dynamic Map Selection While Retaining the Balance ...mentioning

confidence: 99%

QoS-Aware Dynamic MAP Selection in HMIPv6 Architectures

Kadusic¹,

Živić

Kos

2016

IEEE Access

View full text Add to dashboard Cite

The main problem dealt with in this paper is the creation of a protocol for improved QoS-aware mobility management support in cellular all-IP networks, whereby we propose a new algorithm for QoS-aware mobility management, based on multidimensional QoS metrics. An analytical framework for performance evaluation was presented as well. The proposed algorithm for QoS-aware dynamic MAP selection relies on multidimensional QoS metrics, defined in QoS-preference spaces of the mobile node and QoS-ability spaces of MAP candidates, in the decision-making process. The metric is chosen to achieve the desired QoS level through three parameters: bandwidth, delay, and reliability, while retaining the balance of MAP's loads in the entire network. For purposes of performance evaluation of the proposed model, we used: algorithm convergence, traffic class distribution by MAP's, and handover delay. Results showed that the standard deviation for each component of the QoS-ability vector is two orders of magnitude smaller than the deviation in the static MAP selection scenario. We achieved a total handover delay decrease from 20 ms to several hundred milliseconds, by simplifying DAD procedures preserving the simplicity of architecture.INDEX TERMS Handover delay, hierarchical mobile IPv6 (HMIPv6), MAP selection, mobility anchor point.

show abstract

Quality-of-service class specific traffic matrices in ip/mpls networks

Cited by 8 publications

References 11 publications

RouteNet-Fermi: Network Modeling With Graph Neural Networks

RouteNet-Fermi: Network Modeling With Graph Neural Networks

Building a Digital Twin for Network Optimization Using Graph Neural Networks

QoS-Aware Dynamic MAP Selection in HMIPv6 Architectures

Contact Info

Product

Resources

About