Make your plans to attend
Vaccination required
>
Conference 12028 > Paper 12028-6
Paper 12028-6

Fiber nonlinearity compensation using photonic neural network

Abstract

Digital compensation of nonlinear transmission impairments is one of the most practical ways to reduce cost per bit post installment. Artificial neural networks are effective in compensating nonlinear impairments and practical as they only require training data rather than detailed system information. However, they require significant resources and power consumption when implemented in ASIC. Moreover, the requirements increase sharply with increasing baud-rates which is one of the surest ways to reduce transponder costs. Photonic neural networks on the other hand have a much shallower dependence on the baud-rate and are shown to be capable of implementing neural-network based nonlinearity compensation.

Presenter

NEC Labs. America, Inc. (United States)
Fatih Yaman received his Ph.D. in Optical Engineering from The Institute of Optics at University of Rochester in 2006. After postdoctoral research at CREOL, he joined NEC Laboratories of America in 2010 as a research staff member. His current research interests include performance optimization of nonlinear transmission systems, advanced modulation formats and coding schemes, AI-based communication system, and enhanced digital signal processing techniques for long-haul transmission, and optical amplification.
Presenter/Author
NEC Labs. America, Inc. (United States)
Author
Shinsuke Fujisawa
NEC Labs. America, Inc. (United States)
Author
Thomas Ferreira de Lima
NEC Labs. America, Inc. (United States)
Author
NEC Labs. America, Inc. (United States)
Author
NEC Corp. (Japan)
Author
Takanori Inoue
NEC Corp. (Japan)
Author
Yoshihisa Inada
NEC Corp. (Japan)