The main aim of the project is to comprehensively model and mitigate the two main limiting factors of FSO at the physical layer – atmospheric turbulence and pointing-errors — making use of advanced AI tools to overcome the high complexity associated with these open research problems.
In order to model the impact of atmospheric turbulence on outdoor deployments, the OptWire project will rely on extensive experimental campaigns within controlled laboratory environments: i) a wind-tunnel scenario, provided by CESAM/UA and ii) a custom-design atmospheric chamber to be manufactured by PIC Advanced, SA. After the collection of very large amounts of data for a wide range of turbulence parameters, AI-based model extraction will be performed through big data analysis using machine learning algorithms. This AI-optimized channel prediction model will then be critical for the design of enhanced mitigation measures, such as adaptive modulation and coding, beam shaping and multiple-input-multiple-output transmission diversity. On the other hand, combatting the impact of pointing errors will be pursued through the use of high-precision optical and mechanical beam steering devices, whose operation will be optimized through modern AI-based approaches, such as evolutionary algorithms and artificial neural networks. In addition, the optimization of signal modulation and spectral efficiency will be performed through end-to-end learning of the FSO link, resorting to the use of auto-encoders to automatically adapt the signal properties to the varying conditions of the link.
|Start Date: 01-12-2021|
|End Date: 30-11-2024|
|Team: Fernando Pedro Pereira Guiomar, Paulo Miguel Nepomuceno Pereira Monteiro, Petia Georgieva, Antonio Luis Jesus Teixeira, Maria do Carmo Raposo de Medeiros, Marco André Tavares Fernandes, Bruno Tavares Brandão, Manuel dos Santos Neves, Beatriz Manata de Oliveira|
|Groups: Optical Communication Systems and Networking – Av, Applied Mathematics - Av, Optical Communication Systems and Networking – Co|
|Partners: CESAM, Universidade de Aveiro, PIC Advanced, SA|
|Local Coordinator: Fernando Pedro Pereira Guiomar|