PhD Machine learning for optimal component design

This Ph.D. project will explore the latest advances within machine learning to enable pre- and –post compensation techniques for distortion-free communication over the ultra-wideband (O+E+S+C+L) transmission window. The project will investigate how to learn the optimum signaling and pulse-shaping schemes from the transmit-receive data. Therefore, the application and efficient training for auto-encoders will be investigated. The project will cover both algorithm development as well as experimental implementations.

This PhD position will investigate the research topics below using both theoretical and experimental methods, but not limited to these topics:

• Develop novel machine learning based transmitter and receiver schemes for ultra-wideband optical communication systems

• Investigate the impact of non-ideal transmission schemes on the performance of the learned signaling schemes

• Apply machine learning algorithms to improve the knowledge on the transfer function applied by the nonlinear fiber channel

• Investigate the limitations of the machine learning based transmitter-receiver architectures