End-to-End Learning of Geometrical Shaping Maximizing Generalized Mutual Information
Paper in proceeding, 2020
GMI-based end-to-end learning is shown to be highly nonconvex. We apply gradient descent initialized with Gray-labeled APSK constellations directly to the constellation coordinates. State-of-the-art constellations in 2D and 4D are found providing reach increases up to 26% w.r.t. to QAM.
Author
Kadir Gümüş
Eindhoven University of Technology
Alex Alvarado
Eindhoven University of Technology
Bin Chen
Eindhoven University of Technology
Christian Häger
Chalmers, Electrical Engineering, Communication, Antennas and Optical Networks
Erik Agrell
Chalmers, Electrical Engineering, Communication, Antennas and Optical Networks
2020 Optical Fiber Communications Conference and Exhibition, OFC 2020 - Proceedings
21622701 (eISSN)
Vol. Part F174-OFC 2020 W3D.49781943580712 (ISBN)
Optical Fiber Communication Conference (OFC) 2020
San Diego, CA, USA,
San Diego, CA, USA,
Coding for terabit-per-second fiber-optical communications (TERA)
European Commission (EC) (EC/H2020/749798), 2017-01-01 -- 2019-12-31.
Signal shaping in optical communications—Beyond the Gaussian channel
Swedish Research Council (VR) (2017-03702), 2018-01-01 -- 2021-12-31.
Areas of Advance
Information and Communication Technology
Subject Categories
Telecommunications
Communication Systems
Signal Processing
DOI
10.1364/OFC.2020.W3D.4