Neuromorphic reservoir for nonlinear optical signal equalization

Guillermo von Hünefeld; Binoy Chacko; Gregor Ronniger; Mahdi Kaveh; Isaac Sackey; Mahtab Aghaeipour; Peter Bienstman; Colja Schubert; Ronald Freund

doi:10.1117/12.3002627

11 March 2024 Neuromorphic reservoir for nonlinear optical signal equalization

Guillermo von Hünefeld, Binoy Chacko, Gregor Ronniger, Mahdi Kaveh, Isaac Sackey, Mahtab Aghaeipour, Peter Bienstman, Colja Schubert, Ronald Freund

Author Affiliations +

Proceedings Volume 12880, Physics and Simulation of Optoelectronic Devices XXXII; 128800H (2024) https://doi.org/10.1117/12.3002627
Event: SPIE OPTO, 2024, San Francisco, California, United States

Abstract

The upcoming optical telecommunication networks face a significant challenge due to a massive increase in internet traffic. To handle this, higher-capacity transmission schemes are being implemented. To increase the optical signal-to-noise ratios (OSNR), higher launch powers are used, which are limited by nonlinear distortions caused by the Kerr effect in the transmission fibers. Currently, expensive and power-hungry digital signal processing (DSP) solutions are used to tackle this problem. Our proposal offers an alternative solution using a neural network based on a photonic reservoir to address the nonlinear distortions in transmission links. This approach is potentially more cost-effective and consumes less energy. The photonic reservoir design is based on a four-port architecture incorporating multimode interferometers (MMIs), Mach-Zehnder-Interferometers (MZIs), and semiconductor optical amplifiers (SOAs). Inside the reservoir, the optical signals from past and current transmissions are mixed, providing the network with a memory-like capability. The training process focuses solely on driving the MZI and SOA arrays, resulting in accurate outcomes with reduced training time and energy consumption. We numerically demonstrate the mitigation of nonlinearities in high-order transmission links using a photonic reservoir. By comparing various configurations of the neural network (NN), we highlight the specific advantages of each implementation. Looking ahead, we aim to implement this approach using a photonic integrated circuit (PIC) to further enhance its practicality and efficiency.

Conference Presentation

Citation Download Citation

Guillermo von Hünefeld, Binoy Chacko, Gregor Ronniger, Mahdi Kaveh, Isaac Sackey, Mahtab Aghaeipour, Peter Bienstman, Colja Schubert, and Ronald Freund "Neuromorphic reservoir for nonlinear optical signal equalization", Proc. SPIE 12880, Physics and Simulation of Optoelectronic Devices XXXII, 128800H (11 March 2024); https://doi.org/10.1117/12.3002627

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available