Lluís Fàbrega
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Lluís Fàbrega has a total of 42 co-authors. Here are the top co-authors by number of shared publications:
- Pere Vilà
Institute of Informatics and Applications, University of Girona, Girona, Spain
20 shared publications - Jose Luis Marzo
Universitat de Girona
9 shared publications Dimitri Papadimitriou
7 shared publicationsTeodor Jové
University of Girona
7 shared publications- Davide Careglio
Department of Computer Architecture, Technical University of Catalonia—BarcelonaTech (UPC), 08034 Barcelona, Spain
4 shared publications - Miquel Farreras
University of Girona, Girona, Spain
4 shared publications
Latest Publications
- Open Access
- 11 January 2025
- Published by Elsevier in Computer Communications
- Vol. 232
GNNetSlice: A GNN-based performance model to support network slicing in B5G networks
- 11 January 2025
- Published by Elsevier in Computer Communications
- Vol. 232
Open AccessGNNetSlice: A GNN-based performance model to support network slicing in B5G networks
- Open Access
- 15 July 2024
- Published by Elsevier in Data in Brief
- Vol. 55, 110738
Generation of a network slicing dataset: The foundations for AI-based B5G resource management
- 15 July 2024
- Published by Elsevier in Data in Brief
- Vol. 55, 110738
Open AccessGeneration of a network slicing dataset: The foundations for AI-based B5G resource management
- Open Access
- 20 July 2023
- Published by Springer Nature in Journal of Network and Systems Management
- Vol. 31 (4) , 1-36
Improving Network Delay Predictions Using GNNs
Autonomous network management is crucial for Fifth Generation (5G) and Beyond 5G (B5G) networks, where a constantly changing environment is expected and network configuration must adapt accordingly. Modeling tools are required to predict the impact on performance (packet and delay loss) when new traffic demands arrives and when changes in routing paths are applied in the network. Mathematical analysis and network simulators are techniques for modeling networks but both have limitations, as the former provides low accuracy and the latter requires high execution times. To overcome these limitations, machine learning (ML) algorithms, and more specifically, graph neural networks (GNNs), are proposed for network modeling due to their ability to capture complex relationships from graph-like data while predicting network properties with high accuracy and low computational requirements. However, one of the main issues when using GNNs is their lack of generalization capability to larger networks, i.e., when trained in small networks (in number of nodes, paths length, links capacity), the accuracy of predictions on larger networks is poor. This paper addresses the GNN generalization problem by the use of fundamental networking concepts. Our solution is built from a baseline GNN model called RouteNet (developed by Barcelona Neural Networking Center-Universitat Politècnica de Catalunya (BNN-UPC)) that predicts the average delay in network paths, and through a number of simple additions significantly improves the prediction accuracy in larger networks. The improvement ratio compared to the baseline model is 101, from a 187.28% to a 1.828%, measured by the Mean Average Percentage Error (MAPE). In addition, we propose a closed-loop control context where the resulting GNN model could be potentially used in different use cases.
- 20 July 2023
- Published by Springer Nature in Journal of Network and Systems Management
- Vol. 31 (4) , 1-36
Open AccessImproving Network Delay Predictions Using GNNs
Autonomous network management is crucial for Fifth Generation (5G) and Beyond 5G (B5G) networks, where a constantly changing environment is expected and network configuration must adapt accordingly. Modeling tools are required to predict the impact on performance (packet and delay loss) when new traffic demands arrives and when changes in routing paths are applied in the network. Mathematical analysis and network simulators are techniques for modeling networks but both have limitations, as the former provides low accuracy and the latter requires high execution times. To overcome these limitations, machine learning (ML) algorithms, and more specifically, graph neural networks (GNNs), are proposed for network modeling due to their ability to capture complex relationships from graph-like data while predicting network properties with high accuracy and low computational requirements. However, one of the main issues when using GNNs is their lack of generalization capability to larger networks, i.e., when trained in small networks (in number of nodes, paths length, links capacity), the accuracy of predictions on larger networks is poor. This paper addresses the GNN generalization problem by the use of fundamental networking concepts. Our solution is built from a baseline GNN model called RouteNet (developed by Barcelona Neural Networking Center-Universitat Politècnica de Catalunya (BNN-UPC)) that predicts the average delay in network paths, and through a number of simple additions significantly improves the prediction accuracy in larger networks. The improvement ratio compared to the baseline model is 101, from a 187.28% to a 1.828%, measured by the Mean Average Percentage Error (MAPE). In addition, we propose a closed-loop control context where the resulting GNN model could be potentially used in different use cases.
- Open Access
- 29 November 2022
- Published by Springer Nature in Research Square
Improving network delay predictions using GNNs
Autonomous network management is crucial for Fifth Generation (5G) and Beyond 5G (B5G) networks, where constant environmental changes must be predicted, adapting network configurations accordingly. Model-ing tools are required to predict the impact on performance when new traffic combinations and routings are applied in the network. Mathematical analysis and network simulators are techniques to model modern networks but have limitations of low accuracy or high execution times, respectively. To overcome these limitations, Machine Learning (ML) algorithms, and more concretely, Graph Neural Networks (GNNs), are proposed to model modern networks due to their ability to capture complex relationships from graph-like data while predicting network properties with high accuracy. However, one of the main issues when using GNNs is that they cannot generalize to larger networks, when different input data (i.e. smaller network topologies, and shorter paths) are provided for training and predictions (e.g., larger network topolo-gies, larger paths). This paper addresses the GNN generalization problem by leveraging networking concepts and analysis to improve a base-line model. The results applied on the RouteNet baseline developed by Barcelona Neural Networking Center-Universitat Politècnica de Catalunya (BNN-UPC) significantly improve the model’s accuracy in larger graphs. In addition, we propose a closed-loop control context where the resulting GNN model could be potentially used in different use cases.
- 29 November 2022
- Published by Springer Nature in Research Square
Open AccessImproving network delay predictions using GNNs
Autonomous network management is crucial for Fifth Generation (5G) and Beyond 5G (B5G) networks, where constant environmental changes must be predicted, adapting network configurations accordingly. Model-ing tools are required to predict the impact on performance when new traffic combinations and routings are applied in the network. Mathematical analysis and network simulators are techniques to model modern networks but have limitations of low accuracy or high execution times, respectively. To overcome these limitations, Machine Learning (ML) algorithms, and more concretely, Graph Neural Networks (GNNs), are proposed to model modern networks due to their ability to capture complex relationships from graph-like data while predicting network properties with high accuracy. However, one of the main issues when using GNNs is that they cannot generalize to larger networks, when different input data (i.e. smaller network topologies, and shorter paths) are provided for training and predictions (e.g., larger network topolo-gies, larger paths). This paper addresses the GNN generalization problem by leveraging networking concepts and analysis to improve a base-line model. The results applied on the RouteNet baseline developed by Barcelona Neural Networking Center-Universitat Politècnica de Catalunya (BNN-UPC) significantly improve the model’s accuracy in larger graphs. In addition, we propose a closed-loop control context where the resulting GNN model could be potentially used in different use cases.
- 1 April 2019
- Published by Elsevier in Discrete Applied Mathematics
- Vol. 259, 218-225
Low time complexity algorithms for path computation in Cayley Graphs
- 1 April 2019
- Published by Elsevier in Discrete Applied Mathematics
- Vol. 259, 218-225
Low time complexity algorithms for path computation in Cayley Graphs
- 1 July 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Evaluation of Cayley Graphs for Parallel Computer Systems
Some topology properties that determine the performance and robustness of parallel computer systems include fault-tolerance, latency, load-balancing and support for a high number of end points. This paper analyses the impact of graph symmetry on these properties. Through a comparison between Cayley Graphs (CGs) and classical topologies for parallel computer systems, we show that many families of CGs provide optimal values for the mentioned properties.
- 1 July 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Evaluation of Cayley Graphs for Parallel Computer Systems
Some topology properties that determine the performance and robustness of parallel computer systems include fault-tolerance, latency, load-balancing and support for a high number of end points. This paper analyses the impact of graph symmetry on these properties. Through a comparison between Cayley Graphs (CGs) and classical topologies for parallel computer systems, we show that many families of CGs provide optimal values for the mentioned properties.
- 23 June 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Network
- Vol. 31 (4) , 108-118
Routing at Large Scale: Advances and Challenges for Complex Networks
A wide range of social, technological and communication systems can be described as complex networks. Scale-free networks are one of the well known classes of complex networks in which nodes' degrees follow a power-law distribution. The design of scalable, adaptive and resilient routing schemes in such networks is very challenging. In this article we present an overview of required routing functionality, categorize the potential design dimensions of routing protocols among existing routing schemes, and analyze experimental results and analytical studies performed so far to identify the main trends/trade-offs and draw main conclusions. Besides traditional schemes such as hierarchical/shortest-path path-vector routing, the article pays attention to advances in compact routing and geometric routing since they are known to significantly improve scalability in terms of memory space. The identified trade-offs and the outcomes of this overview enable more careful conclusions regarding the (un-)suitability of different routing schemes to large-scale complex networks and provide a guideline for future routing research.
- 23 June 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Network
- Vol. 31 (4) , 108-118
Routing at Large Scale: Advances and Challenges for Complex Networks
A wide range of social, technological and communication systems can be described as complex networks. Scale-free networks are one of the well known classes of complex networks in which nodes' degrees follow a power-law distribution. The design of scalable, adaptive and resilient routing schemes in such networks is very challenging. In this article we present an overview of required routing functionality, categorize the potential design dimensions of routing protocols among existing routing schemes, and analyze experimental results and analytical studies performed so far to identify the main trends/trade-offs and draw main conclusions. Besides traditional schemes such as hierarchical/shortest-path path-vector routing, the article pays attention to advances in compact routing and geometric routing since they are known to significantly improve scalability in terms of memory space. The identified trade-offs and the outcomes of this overview enable more careful conclusions regarding the (un-)suitability of different routing schemes to large-scale complex networks and provide a guideline for future routing research.
- 21 October 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Communications Letters
- Vol. 18 (12) , 2125-2128
Geometric Routing With Word-Metric Spaces
As a potential solution to the compact routing problem, geometric routing has been proven to be both simple and heuristically effective. These routing schemes assign some (virtual) coordinates in a metric space to each network vertex through the process called embedding. By forwarding packets to the closest neighbor to the destination, they ensure a completely local process with the routing table bounded in size by the maximum vertex degree. In this letter, we present an embedding of any finite connected graph into a metric space generated by algebraic groups, and we prove that it is greedy (guaranteed packet delivery). Then, we present a specialized compact routing scheme relying on this embedding for scale-free graphs. Evaluation through simulation on several Internet topologies shows that the resulting stretch remains below the theoretical upper bounds.
- 21 October 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Communications Letters
- Vol. 18 (12) , 2125-2128
Geometric Routing With Word-Metric Spaces
As a potential solution to the compact routing problem, geometric routing has been proven to be both simple and heuristically effective. These routing schemes assign some (virtual) coordinates in a metric space to each network vertex through the process called embedding. By forwarding packets to the closest neighbor to the destination, they ensure a completely local process with the routing table bounded in size by the maximum vertex degree. In this letter, we present an embedding of any finite connected graph into a metric space generated by algebraic groups, and we prove that it is greedy (guaranteed packet delivery). Then, we present a specialized compact routing scheme relying on this embedding for scale-free graphs. Evaluation through simulation on several Internet topologies shows that the resulting stretch remains below the theoretical upper bounds.
- 1 June 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 63-69
Cayley-Graph-Based Data Centers and Space Requirements of a Routing Scheme Using Automata
Modern data centers connect tens of thousands of computers by an interconnection network. The design of such networks implies the selection of an appropriate routing scheme for them. Those schemes need to be efficient with respect to time and space requirements. Cayley Graphs (CG) has been proposed as models for large-scale interconnection networks with excellent properties and very efficient routing schemes. In a previous work, we presented a fast general-purpose shortest path routing scheme for CG with compact routing tables. The scheme uses the concept the Automatic Structures (AS) of a group. However, the size of such structures was not considered into the complexity analysis. Therefore, this paper evaluates the required space to keep such structures and the several intermediate finite state automata that arise during the process of constructing such AS. We perform the evaluation on six well-known families of CG. The results show which structures are space-efficient to implement the scheme, and how the size of such structures depends on the so-called k-fellow traveler property.
- 1 June 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 63-69
Cayley-Graph-Based Data Centers and Space Requirements of a Routing Scheme Using Automata
Modern data centers connect tens of thousands of computers by an interconnection network. The design of such networks implies the selection of an appropriate routing scheme for them. Those schemes need to be efficient with respect to time and space requirements. Cayley Graphs (CG) has been proposed as models for large-scale interconnection networks with excellent properties and very efficient routing schemes. In a previous work, we presented a fast general-purpose shortest path routing scheme for CG with compact routing tables. The scheme uses the concept the Automatic Structures (AS) of a group. However, the size of such structures was not considered into the complexity analysis. Therefore, this paper evaluates the required space to keep such structures and the several intermediate finite state automata that arise during the process of constructing such AS. We perform the evaluation on six well-known families of CG. The results show which structures are space-efficient to implement the scheme, and how the size of such structures depends on the so-called k-fellow traveler property.
- 1 January 2014
- Published by Springer Nature
- p. 189-197
Efficient Routing in Data Center with Underlying Cayley Graph
- 1 January 2014
- Published by Springer Nature
- p. 189-197
Efficient Routing in Data Center with Underlying Cayley Graph