Share Email Print

Optical Engineering

Raman crosstalk cancellation in wavelength-division multiplexing by using neural networks
Author(s): Ramon M. Rodriguez-Dagnino; Jose A. Faz-Gomez; Ignacio Larraga-Orozco
Format Member Price Non-Member Price
PDF $20.00 $25.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

In wavelength-division multiplexing (WDM) optical systems, as the number of users is increased, the crosstalk becomes more severe due to imperfections in the passive optical filters (gratings) and nonlinear effects due to the power increasing in the optical fiber. We consider only stimulated Raman scattering (SRS) as the nonlinear degradation in this multichannel system. We propose an adaptive neural network algorithm based on back-propagation networks to mitigate these crosstalk problems. We also show that the back-propagation neural network algorithm has a better performance than the least mean squares (LMS) algorithm.

Paper Details

Date Published: 1 June 2003
PDF: 7 pages
Opt. Eng. 42(6) doi: 10.1117/1.1569493
Published in: Optical Engineering Volume 42, Issue 6
Show Author Affiliations
Ramon M. Rodriguez-Dagnino, Monterrey Institute of Technology (ITESM) (Mexico)
Jose A. Faz-Gomez, Axtel (United States)
Ignacio Larraga-Orozco, Alestra (United States)

© SPIE. Terms of Use
Back to Top