João Pires
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
João Pires has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Alexandre Freitas
Department of Electrical and Computer Engineering, Instituto Superior Técnico, Universidade de Lisboa, Avenida Rovisco Pais 1, 1049-001 Lisboa, Portugal
5 shared publications - João Pedro
Instituto de Telecomunicaes, Instituto Superior Tcnico, Lisbon, Portugal
5 shared publications - António Eira
Infinera Portugal, R. da Garagem 1, 2790-078 Carnaxide, Portugal
3 shared publications - Luis Goncalo C. Cancela
Department of Information Science and Technology, ISCTE—Instituto Universitário de Lisboa, 1649-026 Lisboa, Portugal
3 shared publications - Paulo Monteiro
Instituto de Telecomunicações, University of Aveiro
2 shared publications - Tomás Maia
Department of Electrical and Computer Engineering, Instituto Superior Técnico, Universidade de Lisboa, Avenida Rovisco Pais 1, 1049-001 Lisboa, Portugal
2 shared publications
Latest Publications
- Open Access
- 21 July 2026
- Published by Elsevier in Optical Fiber Technology
- Vol. 102
Security-aware non-uniform multipath traffic partitioning for QKD networks: A MILP-based squeezing protection approach
- 21 July 2026
- Published by Elsevier in Optical Fiber Technology
- Vol. 102
Open AccessSecurity-aware non-uniform multipath traffic partitioning for QKD networks: A MILP-based squeezing protection approach
- Open Access
- 14 December 2025
- Published by MDPI AG in Electronics
- Vol. 14 (24) , 4911
Topological Design Aspects of Super C+L-Band Optical Backbone Networks Using Machine Learning
A promising approach to alleviate the emerging capacity limitations in backbone optical networks is to employ the Super C+L-band, which provides an available spectrum of roughly 12 THz. Network throughput is a key metric for analyzing the performance of such networks; however, evaluating this metric is a complex task due to the interplay between physical and network layer aspects. Physical modeling, which accounts for signal impairments, is particularly complex in these scenarios due to the presence of Stimulated Raman Scattering (SRS), which transfers energy from the C to the L band. On the other hand, network layer modeling is also challenging due to the influence of numerous factors, including physical topology, routing, and traffic characteristics. For the networks considered here, we propose a machine learning approach to predict both the network throughput and the average channel capacity for the Shannon and real cases, and to investigate how these metrics depend on various physical topology parameters. The approach relies on an Artificial Neural Network (ANN) model, whose predictions are interpreted using the SHapley Additive exPlanations (SHAP) method to identify the importance of various topological parameters. Furthermore, the ANN is trained using data obtained from a previously developed simulator that takes into account both physical and network aspects. The analysis provides valuable insights for designing future ultra-high-capacity optical backbone networks.
- 14 December 2025
- Published by MDPI AG in Electronics
- Vol. 14 (24) , 4911
Open AccessTopological Design Aspects of Super C+L-Band Optical Backbone Networks Using Machine Learning
A promising approach to alleviate the emerging capacity limitations in backbone optical networks is to employ the Super C+L-band, which provides an available spectrum of roughly 12 THz. Network throughput is a key metric for analyzing the performance of such networks; however, evaluating this metric is a complex task due to the interplay between physical and network layer aspects. Physical modeling, which accounts for signal impairments, is particularly complex in these scenarios due to the presence of Stimulated Raman Scattering (SRS), which transfers energy from the C to the L band. On the other hand, network layer modeling is also challenging due to the influence of numerous factors, including physical topology, routing, and traffic characteristics. For the networks considered here, we propose a machine learning approach to predict both the network throughput and the average channel capacity for the Shannon and real cases, and to investigate how these metrics depend on various physical topology parameters. The approach relies on an Artificial Neural Network (ANN) model, whose predictions are interpreted using the SHapley Additive exPlanations (SHAP) method to identify the importance of various topological parameters. Furthermore, the ANN is trained using data obtained from a previously developed simulator that takes into account both physical and network aspects. The analysis provides valuable insights for designing future ultra-high-capacity optical backbone networks.
- Open Access
Multipath Interference Impact Due to Fiber Mode Coupling in C+L+S Multiband Transmission Reach
Multiband transmission is, nowadays, being implemented worldwide to increase the optical transport network capacity, mainly because it uses the already-installed single-mode fiber (SMF). The G.654E SMF, due to its attributes (e.g., low-loss, and large-effective area in comparison with the standard G.652 SMF), can also increase network capacity and can also be used for multiband (MB) transmission. Nevertheless, in MB transmission, power mode coupling arises when bands with wavelengths below the cut-off wavelength are used, inducing multipath interference (MPI). This work investigates the impact of the MPI, due to mode coupling from G.654E SMF, in the transmission reach of a C+L+S band transmission system. Our results indicate that for the S-band scenario, the band below the wavelength cut-off, an approximately 25% reach decrease is observed when the MPI/span increases to −26 dB/span, considering quadrature phase-shift keying (QPSK) signals with a 64 GBaud symbol rate. We also concluded that if the L-band were not above the wavelength cut-off, it would be much more affected than the S-band, with an approximately 52% reach decrease due to MPI impact.
Open AccessMultipath Interference Impact Due to Fiber Mode Coupling in C+L+S Multiband Transmission Reach
Multiband transmission is, nowadays, being implemented worldwide to increase the optical transport network capacity, mainly because it uses the already-installed single-mode fiber (SMF). The G.654E SMF, due to its attributes (e.g., low-loss, and large-effective area in comparison with the standard G.652 SMF), can also increase network capacity and can also be used for multiband (MB) transmission. Nevertheless, in MB transmission, power mode coupling arises when bands with wavelengths below the cut-off wavelength are used, inducing multipath interference (MPI). This work investigates the impact of the MPI, due to mode coupling from G.654E SMF, in the transmission reach of a C+L+S band transmission system. Our results indicate that for the S-band scenario, the band below the wavelength cut-off, an approximately 25% reach decrease is observed when the MPI/span increases to −26 dB/span, considering quadrature phase-shift keying (QPSK) signals with a 64 GBaud symbol rate. We also concluded that if the L-band were not above the wavelength cut-off, it would be much more affected than the S-band, with an approximately 52% reach decrease due to MPI impact.
- Open Access
- 8 July 2025
- Published by MDPI AG in Preprints.org
Multipath Interference Impact Due to Fiber Mode Coupling in C+L+S Multiband Transmission Reach
Multiband transmission is, nowadays, being implemented worldwide to increase the optical transport network capacity, mainly because it uses the already installed single mode fiber (SMF). The G.654E SMF, due to its attributes, e.g. low-loss, and large-effective area in comparison with the standard G.652 SMF, can also increase network capacity and can also be used for MB transmission. Nevertheless, in MB transmission power mode coupling arises when bands with wavelengths below the cut-off wavelength are used, inducing multipath interference (MPI). This work investigates the impact of the MPI, due mode coupling from G.654E SMF, in the transmission reach of a C+L+S band transmission system. Our results indicate that for the S-band scenario, the band below the wavelength cut-off, an approximately 25% reach decrease is observed when the MPI/span increases to -26 dB/span considering quadrature phase-shift keying (QPSK) signals with a 64 GBaud symbol rate. We also concluded that if the L-band were not above the wavelength cut-off it would be much more affected than the S-band, with an approximately 55% reach decrease due to MPI impact.
- 8 July 2025
- Published by MDPI AG in Preprints.org
Open AccessMultipath Interference Impact Due to Fiber Mode Coupling in C+L+S Multiband Transmission Reach
Multiband transmission is, nowadays, being implemented worldwide to increase the optical transport network capacity, mainly because it uses the already installed single mode fiber (SMF). The G.654E SMF, due to its attributes, e.g. low-loss, and large-effective area in comparison with the standard G.652 SMF, can also increase network capacity and can also be used for MB transmission. Nevertheless, in MB transmission power mode coupling arises when bands with wavelengths below the cut-off wavelength are used, inducing multipath interference (MPI). This work investigates the impact of the MPI, due mode coupling from G.654E SMF, in the transmission reach of a C+L+S band transmission system. Our results indicate that for the S-band scenario, the band below the wavelength cut-off, an approximately 25% reach decrease is observed when the MPI/span increases to -26 dB/span considering quadrature phase-shift keying (QPSK) signals with a 64 GBaud symbol rate. We also concluded that if the L-band were not above the wavelength cut-off it would be much more affected than the S-band, with an approximately 55% reach decrease due to MPI impact.
- Open Access
- 6 July 2025
- Published by MDPI AG in Electronics
- Vol. 14 (13) , 2723
A Throughput Analysis of C+L-Band Optical Networks: A Comparison Between the Use of Band-Dedicated and Single-Wideband Amplification
Optical networks today constitute the fundamental backbone infrastructure of telecom and cloud operators. A possible medium-term solution to address the enormous increase in traffic demands faced by these operators is to rely on Super C+ L transmission optical bands, which can offer a bandwidth of about 12 THz. In this paper, we propose a methodology to compute the throughput of an optical network based on this solution. The methodology involves detailed physical layer modeling, including the impact of stimulated Raman scattering, which is responsible for energy transfer between the two bands. Two approaches are implemented for throughput evaluation: one assuming idealized Gaussian-modulated signals and the other using real modulation formats. For designing such networks, it is crucial to choose the most appropriate technological solution for optical amplification. This could either be a band-dedicated scheme, which uses a separate amplifier for each of the two bands, or a single-wideband amplifier capable of amplifying both bands simultaneously. The simulation results show that the single-wideband scheme provides an average throughput improvement of about 18% compared to the dedicated scheme when using the Gaussian modulation approach. However, with the real modulation approach, the improvement increases significantly to about 32%, highlighting the benefit in developing single-wideband amplifiers for future applications in Super C+L-band networks.
- 6 July 2025
- Published by MDPI AG in Electronics
- Vol. 14 (13) , 2723
Open AccessA Throughput Analysis of C+L-Band Optical Networks: A Comparison Between the Use of Band-Dedicated and Single-Wideband Amplification
Optical networks today constitute the fundamental backbone infrastructure of telecom and cloud operators. A possible medium-term solution to address the enormous increase in traffic demands faced by these operators is to rely on Super C+ L transmission optical bands, which can offer a bandwidth of about 12 THz. In this paper, we propose a methodology to compute the throughput of an optical network based on this solution. The methodology involves detailed physical layer modeling, including the impact of stimulated Raman scattering, which is responsible for energy transfer between the two bands. Two approaches are implemented for throughput evaluation: one assuming idealized Gaussian-modulated signals and the other using real modulation formats. For designing such networks, it is crucial to choose the most appropriate technological solution for optical amplification. This could either be a band-dedicated scheme, which uses a separate amplifier for each of the two bands, or a single-wideband amplifier capable of amplifying both bands simultaneously. The simulation results show that the single-wideband scheme provides an average throughput improvement of about 18% compared to the dedicated scheme when using the Gaussian modulation approach. However, with the real modulation approach, the improvement increases significantly to about 32%, highlighting the benefit in developing single-wideband amplifiers for future applications in Super C+L-band networks.
- Open Access
The Role of Topological Parameters in Wavelength Requirements for Survivable Optical Backbone Networks
As optical networks operate using light-based transmission, assigning wavelengths to the paths taken by traffic demands is a key aspect of their design. This paper revisits the wavelength assignment problem in optical backbone networks, focusing on survivability via 1 + 1 Optical Chanel (OCh) protection, which ensures fault tolerance by duplicating data over two disjoint optical paths. The analysis gives great emphasis to studying the influence of topological parameters on wavelength requirements, with algebraic connectivity being identified as the most significant parameter. The results show that, across a set of 27 real-world networks, the wavelength increment factor, defined as the ratio between the number of wavelengths required with protection and without protection, ranges from 1.49 to 3.07, with a mean value of 2.26. Using synthetic data, formulas were derived to estimate this factor from network parameters, resulting in a mean relative error of 12.7% and errors below 15% in 70% of the real-world cases studied.
Open AccessThe Role of Topological Parameters in Wavelength Requirements for Survivable Optical Backbone Networks
As optical networks operate using light-based transmission, assigning wavelengths to the paths taken by traffic demands is a key aspect of their design. This paper revisits the wavelength assignment problem in optical backbone networks, focusing on survivability via 1 + 1 Optical Chanel (OCh) protection, which ensures fault tolerance by duplicating data over two disjoint optical paths. The analysis gives great emphasis to studying the influence of topological parameters on wavelength requirements, with algebraic connectivity being identified as the most significant parameter. The results show that, across a set of 27 real-world networks, the wavelength increment factor, defined as the ratio between the number of wavelengths required with protection and without protection, ranges from 1.49 to 3.07, with a mean value of 2.26. Using synthetic data, formulas were derived to estimate this factor from network parameters, resulting in a mean relative error of 12.7% and errors below 15% in 70% of the real-world cases studied.
- Open Access
- 16 January 2025
- Published by Universal Wiser Publisher Pte. Ltd in Computer Networks and Communications
- p. 61–74
Estimating Throughput in Optical Backbone Networks Using Deep Neural Networks
Optical backbone networks, which use optical fibres as the transmission medium, form the core infrastructure used by network operators to deliver services to users, as well as by Internet companies to route traffic between data centres. The network throughput is a key parameter in the analysis of the networks' performance. However, its determination can be a complex process that involves long computation times, since aspects related to both the physical and network layers need to be accounted for. To face this challenge, we propose a machine learning solution: a deep neural network (DNN) model, that has the goal of estimating the values of the network throughput and of a closely related parameter, average channel capacity, accurately and with short computation times. The simulation results indicate that the DNN model accurately predicts both outputs, with mean relative errors of 6.17% for the network throughput and 2.84% for the average channel capacity. These predictions are made in just a few milliseconds, providing a significant advantage over the heuristic routing algorithms, which can take up to tens of seconds in larger networks.
- 16 January 2025
- Published by Universal Wiser Publisher Pte. Ltd in Computer Networks and Communications
- p. 61–74
Open AccessEstimating Throughput in Optical Backbone Networks Using Deep Neural Networks
Optical backbone networks, which use optical fibres as the transmission medium, form the core infrastructure used by network operators to deliver services to users, as well as by Internet companies to route traffic between data centres. The network throughput is a key parameter in the analysis of the networks' performance. However, its determination can be a complex process that involves long computation times, since aspects related to both the physical and network layers need to be accounted for. To face this challenge, we propose a machine learning solution: a deep neural network (DNN) model, that has the goal of estimating the values of the network throughput and of a closely related parameter, average channel capacity, accurately and with short computation times. The simulation results indicate that the DNN model accurately predicts both outputs, with mean relative errors of 6.17% for the network throughput and 2.84% for the average channel capacity. These predictions are made in just a few milliseconds, providing a significant advantage over the heuristic routing algorithms, which can take up to tens of seconds in larger networks.
- Open Access
Using Artificial Neural Networks to Evaluate the Capacity and Cost of Multi-Fiber Optical Backbone Networks
A possible solution to address the enormous increase in traffic demands faced by network operators is to rely on multi-fiber optical backbone networks. These networks use multiple optical fibers between adjacent nodes, and, when properly designed, they are capable of handling petabits of data per second (Pbit/s). In this paper, an artificial neural network (ANN) model is investigated to estimate both the capacity and cost of a multi-fiber optical network. Furthermore, a fiber assignment algorithm is also proposed to complement the network design, enabling the generation of datasets for training and testing of the developed ANN model. The model consists of three layers, including one hidden layer with 50 hidden units. The results show that for a large network, such as one with 100 nodes, the model can estimate performance metrics with an average relative error of less than 0.4% for capacity and 4% for cost, while achieving a computation time nearly 800 times faster than the heuristic approach used in network simulation. Additionally, the network capacity is around 5 Pbit/s.
Open AccessUsing Artificial Neural Networks to Evaluate the Capacity and Cost of Multi-Fiber Optical Backbone Networks
A possible solution to address the enormous increase in traffic demands faced by network operators is to rely on multi-fiber optical backbone networks. These networks use multiple optical fibers between adjacent nodes, and, when properly designed, they are capable of handling petabits of data per second (Pbit/s). In this paper, an artificial neural network (ANN) model is investigated to estimate both the capacity and cost of a multi-fiber optical network. Furthermore, a fiber assignment algorithm is also proposed to complement the network design, enabling the generation of datasets for training and testing of the developed ANN model. The model consists of three layers, including one hidden layer with 50 hidden units. The results show that for a large network, such as one with 100 nodes, the model can estimate performance metrics with an average relative error of less than 0.4% for capacity and 4% for cost, while achieving a computation time nearly 800 times faster than the heuristic approach used in network simulation. Additionally, the network capacity is around 5 Pbit/s.
- Open Access
- 18 October 2024
- Published by MDPI AG in Preprints.org
Using Artificial Neural Networks to Evaluate the Capacity and Cost of Multi-Fiber Optical Backbone Networks
A possible solution to address the enormous increase in traffic demands faced by network operators is to rely on multi-fiber optical backbone networks. These networks use multiple optical fibers between adjacent nodes, and, when properly designed, they are capable of handling petabits of data per second (Pbit/s). In this paper, an artificial neural network (ANN) model is investigated to estimate both the capacity and cost of a multi-fiber optical network. Furthermore, a fiber assignment algorithm is also proposed to complement the network design, enabling the generation of datasets for training and testing the developed ANN model. The model consists of three layers, including one hidden layer with 50 hidden units. The results show that for a large network, such as one with 100 nodes, the model can estimate performance metrics with an average relative error of less than 0.4% for capacity and 4% for cost, while achieving a computation time nearly 800 times faster than the heuristic approach used in network simulation. Additionally, the network capacity is around 5 Pbit/s.
- 18 October 2024
- Published by MDPI AG in Preprints.org
Open AccessUsing Artificial Neural Networks to Evaluate the Capacity and Cost of Multi-Fiber Optical Backbone Networks
A possible solution to address the enormous increase in traffic demands faced by network operators is to rely on multi-fiber optical backbone networks. These networks use multiple optical fibers between adjacent nodes, and, when properly designed, they are capable of handling petabits of data per second (Pbit/s). In this paper, an artificial neural network (ANN) model is investigated to estimate both the capacity and cost of a multi-fiber optical network. Furthermore, a fiber assignment algorithm is also proposed to complement the network design, enabling the generation of datasets for training and testing the developed ANN model. The model consists of three layers, including one hidden layer with 50 hidden units. The results show that for a large network, such as one with 100 nodes, the model can estimate performance metrics with an average relative error of less than 0.4% for capacity and 4% for cost, while achieving a computation time nearly 800 times faster than the heuristic approach used in network simulation. Additionally, the network capacity is around 5 Pbit/s.