Periodic Reporting for period 1 - REAL-NET (REAL-time monitoring and mitigation of nonlinear effects in optical NETworks)
Período documentado: 2019-01-01 hasta 2020-12-31
"It is hard to overstate the impact that fibre communications have made on the economy, society, and almost all aspects of our lives. In view of the COVID-19 consequences, the value of connectivity and stable high-speed Internet availability becomes vitally important, defining the new economical and societal functioning regulations. REAL-NET (EID) project objectives refer to developing the new artificial intelligence-based approach and related signal processing techniques, and their successive utilisation for the new nonlinearity and noise-immune optical transmission systems. Thus, all project ROs addresses the paramount insufficient-fibre-capacity (“nonlinear capacity crunch”) problem. We believe that the project WPs outcomes will bring about benefits for the development of communications sector and the whole digital economy in general, giving the new nontrivial but efficient solutions for improving the connectivity.
The EU telecommunication’s sector is featuring a rapid growth and changes based on innovations and use of the leading edge technology. Our vision is that the telecommunication industry is largely moving from a hardware-dominated to a software-dominated environment, incorporating novel intelligent low-complexity optical signal processing tools. This change in the approach is creating exceptional demand for optical engineers equipped with expertise in machine learning and big data science methods, and the project objectives, aiming at training the six ESRs (a) to acquire advanced multi-disciplinary knowledge at frontiers of communications engineering and nonlinear optics; (b) to obtain practical and hands-on skills in conducting experiments; (c) to improve employability chances for leading roles in the industry; are a direct response of the aforementioned challenge.
The importance of the project’s subject area is evidenced by the numbers presented in the recent report ""Global Photonics Industry"" (https://www.reportlinker.com/) illustrating the overall steady growth of the photonics-related industry and gross product costs in spite of the pandemic economy slowdown. After an early analysis of the business implications of the pandemic and its induced economic crisis, growth in the Lasers, Sensors, Detectors & Imaging Devices segment in 2020 was readjusted to a revised 6.7% (CAGR) for the next 7-year period. In the global Optical Communication Systems & Components segment, USA, Canada, Japan, China and Europe will drive the 8% compound annual growth rate estimated for this segment."""
The EU telecommunication’s sector is featuring a rapid growth and changes based on innovations and use of the leading edge technology. Our vision is that the telecommunication industry is largely moving from a hardware-dominated to a software-dominated environment, incorporating novel intelligent low-complexity optical signal processing tools. This change in the approach is creating exceptional demand for optical engineers equipped with expertise in machine learning and big data science methods, and the project objectives, aiming at training the six ESRs (a) to acquire advanced multi-disciplinary knowledge at frontiers of communications engineering and nonlinear optics; (b) to obtain practical and hands-on skills in conducting experiments; (c) to improve employability chances for leading roles in the industry; are a direct response of the aforementioned challenge.
The importance of the project’s subject area is evidenced by the numbers presented in the recent report ""Global Photonics Industry"" (https://www.reportlinker.com/) illustrating the overall steady growth of the photonics-related industry and gross product costs in spite of the pandemic economy slowdown. After an early analysis of the business implications of the pandemic and its induced economic crisis, growth in the Lasers, Sensors, Detectors & Imaging Devices segment in 2020 was readjusted to a revised 6.7% (CAGR) for the next 7-year period. In the global Optical Communication Systems & Components segment, USA, Canada, Japan, China and Europe will drive the 8% compound annual growth rate estimated for this segment."""
Despite the delay in the recruitment of some of the ESRs and despite the COVID-19 pandemic, the overall project implementation has progressed well and accordingly the Annex I of the GA. All ESRs are enrolled into a PhD programme and have started their secondments (November 2020).
ESR1, Mr Mohammad M. Hosseini, is primarily dealing with the research relevant to optical networking and traffic optimisation problems. His personal approach includes the special techniques that bring more efficiency and flexibility to the networks in terms of cost, power, and quality of service. One conference contribution has been submitted so far. ESR2, Mr Pedro J. Freire, is dealing with the development of low complexity equalizers for the high-speed optical communication systems. Pedro published his results in the Journal of Lightwave Technology (DOI: 10.1109/JLT.2020.3042414) and in two oral presentations at ECOC 2020. ESR3, Mr Mohannad Abu-Romoh, is working on designing realistically-implementable neural networks for mitigation of the nonlinear distortions in optical fiber, focusing on reducing the complexity of the equalization. He has submitted his results to SPPCom conference. The work-focus of Mr Jamal Darweesh, ESR4, has been on the compensation of nonlinear distortions in dual polarization coherent fiber-optic communication. ESR5´s work (Mr Masab Iqbal) is focused on security at the optical layer, in particular on data encryption and key exchange through the optical channel. Masab, during the secondment, is studying the feasibility of the security optical layer that he is proposing through simulation. ESR6, Diogo Sequeira, is working on the exploitation of Nonlinearities in Optical Communications. During the secondment he has been working on digital subcarrier multiplexing systems (DSCM).
In general, the project has progressed well in all the six WPs, out of 17 deliverables 16 have been achieved. Out of 19 milestones 17 have been completed. To date REAL-NET has achieved to present 3 conference papers and to obtain 2 Journal publications.
All the ESRs have participated to different training event, including: i) TSWI, a two-day training organised at Aston University on 27-28th February 2020, featuring external speakers (among which the editor of Nature and the editor of The Conversation), addressed topics such as: engaging with social media, writing for a non-academic public, team working, and open access in H2020 (Fig.1); ii) Mini-symposium on introduction to Machine Learning delivered virtually on 7-8th September 2020, including external academic and industrial speakers; and iii) 2nd REAL-NET workshop, held on 5th November 2020 online, where the WP leaders inform the consortium about their progress and the ESRs gave technical presentation about their work (Fig 2).
REAL-NET has different management tools implemented, like the project website and the Twitter account, used for external communication and dissemination of the project to reach the widest possible audience.
ESR1, Mr Mohammad M. Hosseini, is primarily dealing with the research relevant to optical networking and traffic optimisation problems. His personal approach includes the special techniques that bring more efficiency and flexibility to the networks in terms of cost, power, and quality of service. One conference contribution has been submitted so far. ESR2, Mr Pedro J. Freire, is dealing with the development of low complexity equalizers for the high-speed optical communication systems. Pedro published his results in the Journal of Lightwave Technology (DOI: 10.1109/JLT.2020.3042414) and in two oral presentations at ECOC 2020. ESR3, Mr Mohannad Abu-Romoh, is working on designing realistically-implementable neural networks for mitigation of the nonlinear distortions in optical fiber, focusing on reducing the complexity of the equalization. He has submitted his results to SPPCom conference. The work-focus of Mr Jamal Darweesh, ESR4, has been on the compensation of nonlinear distortions in dual polarization coherent fiber-optic communication. ESR5´s work (Mr Masab Iqbal) is focused on security at the optical layer, in particular on data encryption and key exchange through the optical channel. Masab, during the secondment, is studying the feasibility of the security optical layer that he is proposing through simulation. ESR6, Diogo Sequeira, is working on the exploitation of Nonlinearities in Optical Communications. During the secondment he has been working on digital subcarrier multiplexing systems (DSCM).
In general, the project has progressed well in all the six WPs, out of 17 deliverables 16 have been achieved. Out of 19 milestones 17 have been completed. To date REAL-NET has achieved to present 3 conference papers and to obtain 2 Journal publications.
All the ESRs have participated to different training event, including: i) TSWI, a two-day training organised at Aston University on 27-28th February 2020, featuring external speakers (among which the editor of Nature and the editor of The Conversation), addressed topics such as: engaging with social media, writing for a non-academic public, team working, and open access in H2020 (Fig.1); ii) Mini-symposium on introduction to Machine Learning delivered virtually on 7-8th September 2020, including external academic and industrial speakers; and iii) 2nd REAL-NET workshop, held on 5th November 2020 online, where the WP leaders inform the consortium about their progress and the ESRs gave technical presentation about their work (Fig 2).
REAL-NET has different management tools implemented, like the project website and the Twitter account, used for external communication and dissemination of the project to reach the widest possible audience.
WP1 is focused on the realistically implementable nonlinearity mitigation techniques. TPT, responsible of this WP, has been working on the low-complexity nonlinearity mitigation in optical fiber transmission using dense neural networks, managing to design a number of signal-signal dense neural networks, in which the network is trained to recover the modulated signal. Besides, they have managed to improve the LDBP propagation in order to obtain a neural network with significantly lower complexity.
WP2, with Aston as lead beneficiary, has been working on the development of few deep neural network-based equalizers for the mitigation of fibre- and component-induced impairments. By analysing the symmetries of the optical channel and their interplay with the digital signal processing operations, they have demonstrated that it is possible to reduce the amount of data required for the training to 2, 4, and 6 times, depending on the level of augmentation used. Moreover, they argue that the complexity issue, associated with different types of neural networks-based equalizers, plays a very important role in the neural-network-based equalizer performance.
WP3 has been working on the optical performance monitoring and trying to apply it to modern optical networks supporting 5G and beyond. In this context, the overall telecom infrastructure will require high capacity and low latency connectivity.
WP4, managed to develop and assessed an algorithm to enable autonomous lightpath operation with a 100% accuracy in presence of nonlinear noise. UPC, responsible for this WP, aims to determine the minimum channel spacing to avoid optical filtering penalties. Finally, the characterization of the NLI noise has been used to train machine learning-based algorithms to detect optical tampering attacks and estimate the distance from the receiver.
WP2, with Aston as lead beneficiary, has been working on the development of few deep neural network-based equalizers for the mitigation of fibre- and component-induced impairments. By analysing the symmetries of the optical channel and their interplay with the digital signal processing operations, they have demonstrated that it is possible to reduce the amount of data required for the training to 2, 4, and 6 times, depending on the level of augmentation used. Moreover, they argue that the complexity issue, associated with different types of neural networks-based equalizers, plays a very important role in the neural-network-based equalizer performance.
WP3 has been working on the optical performance monitoring and trying to apply it to modern optical networks supporting 5G and beyond. In this context, the overall telecom infrastructure will require high capacity and low latency connectivity.
WP4, managed to develop and assessed an algorithm to enable autonomous lightpath operation with a 100% accuracy in presence of nonlinear noise. UPC, responsible for this WP, aims to determine the minimum channel spacing to avoid optical filtering penalties. Finally, the characterization of the NLI noise has been used to train machine learning-based algorithms to detect optical tampering attacks and estimate the distance from the receiver.