Supervisors: Prof. D. Zibar (DTU), Prof. F. Da Ros (DTU)
The project goal is to develop a machine learning based approach to realize transmitter pre-distortion and receiver post-distortion architectures that can enable distortion-free transmission through multi-core fibres. Your tasks will include research into novel machine learning algorithm for mitigation of impairments associated with using multi-core fibres as transmission medium. The project will cover both algorithm development as well as experimental implementations. The position is focused on developing machine learning frameworks, in terms of constellation and pulse-shaping, as well as equalization for mitigating transmission impairments associated with data transmission though multi-core fibres. The focus will be on energy-efficient solutions. More specifically, the focus will be on minimizing the interference between the cores using energy-efficient signal processing based solutions. The overall goal will be to maximize the information rate and the transmission distance given the energy-consumption constraint. Specifically, you will focus on the following areas: 1) gradient based learning for signal symbol and pulse shaping 2) model-free reinforcement learning signal pre-distortion strategies, 3) extracting the relevant information from correlation matrices of multi-core fibres 4) maintenance of the GitHub repository for the developed code and 5) organizing and managing joint experiments with the collaboration groups.
Planned secondment(s):
- Iscte: Prof. T. Alves M12-16 (5 months duration) “Simulation and experimental framework for MCF communication systems”
- UST: Prof. G. Rademacher M24-27 (4 months duration) “Development of coding framework”