Work Package 1 ESRs 1-5, Complexity reduction and data augmentation for ML&AI-based optical systems, led by Dr. Yaroslav Prylepskiy and Prof. Sergei Turitsyn of Aston University. This work package focuses on data augmentation and neural networks (NNs) simplification for the improvement of ML&AI models utilized in optical transmission systems. Industry partners will provide a large amount of real data that will be used to validate the investigated techniques of data augmentation. The large power required by ML algorithms is the main blocking point for the implementation of these techniques within commercial products. We will investigate and provide solutions to reduce the complexity of ML techniques.
Work Package 2 ESRs 1-4 ML&AI for component characterisation and optimisation. This WP focuses on the application of ML&AI techniques for design and optimization of optical components including lasers sources, frequency combs and optical amplifiers. ESRs will explore and develop new ML&AI algorithms for noise characterization of laser sources and inverse design of optical amplifiers and laser sources. MB components (wideband amplifier, lasers, etc.) need to provide similar characteristics over a large spectrum. This quite complex task has been showed that can be better carried out by using ML&AI.
Work Package 3 ESRs 2-5 AI-based control and management for ultra-wideband optical networks, led by Assoc. Prof Nicola Sambo of SSSA, Italy. This work package focuses on the design and the implementation of the control and management plane for MB optical networks. MB transmission will lead to different performance depending on the selected transmission band. Consequently, complex routing, modulation format and spectrum assignment algorithms are required. New strategies will be proposed to extract the highest benefit from the available transmission bands by employing ML&AI algorithms, such as genetic algorithms.
Work Package 4 ESRs 3-6 Distributed intelligence for network surveillance, led by Prof. Luis Velasco of UPC. Work package 4 focuses on the development of techniques for sharing data/knowledge among cooperating entities (including disaggregated scenarios) while in operation. Proposed is a learning life-cycle20 to facilitate ML&AI deployment in real operator networks. WP4 will closely collaborate with WP3 for the distributed intelligence to fit into the control & management architecture, and with WP1 to exploit data augmentation techniques.
Work Package 5 ESRs 1-6 Recruitment, management and implementation. Led by MENTOR Project Manager Ms. Karola Woods.
Work Package 6 ESRs 1-6 Impact, dissemination and outreach. Led by Project Manager Ms. Karola Woods of Aston University, with partners.