OFC/NFOEC: Measurement of Equivalent Zero-Dispersion Wavelength Distribution for…

OFC/NFOEC 2013, Anaheim CA (NW4E.4)

Mark Filer and Sorin Tibuleac

ADVA Optical Networking, Atlanta GA

[email protected]

Measurement of Equivalent Zero-Dispersion Wavelength Distribution for Capacity Increase in FWM-limited Networks

mailto:[email protected]

© 2013 ADVA Optical Networking. All rights reserved. Confidential.OFC/NFOEC 2013, Anaheim CA (NW4E.4)

f1 f2 f3

f113

f123,213

f112 f223

f132,312

f221

f231,321

f332 f331

f1 f2 f3

f113

f123,213

f112 f223

f132,312

f221

f231,321

f332 f331

f1

• Generally describes any third order process in which the interaction of three fields lead to the generation of a fourth

What is Four-Wave Mixing?

𝑓 𝑖𝑗𝑘= 𝑓 𝑖+ 𝑓 𝑗− 𝑓 𝑘

f2

f3

f321 f312 f221

• Several approaches to minimizing FWM-induced crosstalk• Reduced channel count• Limited transmission distance• Raman amps + low launch power• L-band• S-USCA


Zero-Dispersion Wavelength

• Generation of FWM components: phase-matching condition• Knowledge of fiber CD properties critical for optimal channel allocation

• Fiber zero-dispersion wavelength (ZDW) characteristics• ZDW is non-uniform along fiber length• Larger ZDW variation in older DSFs which are still used today

• Challenge: standard method for in-field dispersion measurement is Modulation Phase Shift (MPS) method• MPS yields average ZDW over entire link; however…• Near-end ZDW features more important than far-end

• Solutions: previous methods complex, equipment-intensiveWe propose a new method of determining equivalent ZDW1) Simplified in-field measurement of degenerate FWM efficiency2) Derivation of equivalent ZDW distribution using measured data as input

to search-and-optimization numerical algorithm


In-Field Measurement Technique

• Probe with (swept) pump induces degenerate FWM• Use OSA or OPM to obtain FWM efficiency vs frequency• Repeat for multiple fprobe values

• Take average of results

fpumpfprobe fFWM

f

λ0 𝑓 𝐹𝑊𝑀=2∙ 𝑓 𝑝𝑢𝑚𝑝 − 𝑓 𝑝𝑟𝑜𝑏𝑒

λ0


In-Field Measurement Technique (2)

• Concept demonstrated on 112km DSF comprised of 6 spools• Known ZDW’s (from MPS method)• Four probe wavelengths (1541, 1545, 1561, 1565 nm)

λ0,1

λ0,2

λ0,3

λ0,4


Equivalent ZDW distribution

• Next step: utilize inverse method for determining equivalent ZDW distribution from measured data• Multiple ZDW distributions possible for same solution• Physical distribution of ZDW not as critical as accurate modeling of FWM

efficiency of span• Simple example:


λ0,1 L1

λ0,2 L2

λ0,n Ln

Search-and-Optimization Algorithm

• Search-and-optimization routine: Genetic Algorithm• Merit function: experimental FWM efficiency spectrum• Optimization parameters: fiber segments, ZDWs

initial population

merit function evaluation

rank stop?

solution

new population

select xover mutate

size m FWM efficiencygenerations exceeded

max time exceeded

or stall

ks∙m kx∙m km∙m

Terminology

• candidate solution: ZDW set {λ0,1 , λ0,2 ,…, λ0,n}• population: a set of candidate solutions (length m)• generation: one population lifetime • selection: candidates retained (top ks)

• crossover: combinations of best candidates (kx)

• mutation: random changes to a candidate (km)

{λ0,1 , λ0,2 ,…, λ0,n}n


Example for 112km span DSF

• Measurement performed on 112km span DSF comprised of 6 shorter spools each with different ZDW

• Analytical FWM model* used to calculate FWM efficiency spectrum• Calculated and measured FWM efficiencies compared in a least-squares

sense merit function

• Genetic Algorithm parameters:• Fiber segments, = 20• Population size, = 500• Initial population, = 1550.43nm • Max generations = 225• Selection, = 10%• Crossover, = 55%• Mutation, = 35%

n m

{λ0,1 , λ0,2 ,…, λ0,n}

ks

kx

km

* J. Stay et al, proc. OFC/NFOEC, 2008, paper JWA14


Result for 112km span DSF

• GA merit function designed to emphasize matching of spectral “peaks” with less weighting given to regions of low FWM efficiency

* J. Stay et al, proc. OFC/NFOEC, 2008, paper JWA14

• Final solution set which yielded result above input to GA-based S-USCA optimization tool*


Application to S-USCA scheme

• S-USCA tool produces optimal channel plan given number of channels and channel range/spacing

• Increase in channel capacity realized when inputting equivalent ZDW distribution vs inputting MPS-measured (i.e., average) ZDW

58ch

72ch


S-USCA channel distributions compared

• Optimal allocation of channels for e.g., 70 of 96 channels differ greatly given knowledge of ZDW distribution

Opti

miz

ed f

or

avera

ge Z

DW

Opti

miz

ed f

or

dis

trib

ute

d Z

DW

channel number


Conclusions

• Demonstrated technique for characterizing equivalent ZDW distribution for maximized transmission capacity1. FWM efficiency of span measured2. Equivalent ZDW distribution extracted from inverse search-and-

optimization numerical algorithm

• Result used to generate optimized S-USCA for given network

• May be combined with additional mitigation procedures• Power pre-emphasis• Raman amplification

• Additional benefits realized when combined with:• Higher-order phase-modulated formats (mPSK vs OOK)• High-coding gain SD-FEC (BERs > 10-2)• electronic equalization• coherent technologies

OFC/NFOEC 2013, Anaheim CA (NW4E.4)

[email protected]

Thank you

mailto:[email protected]

Technology

OFC/NFOEC: Measurement of Equivalent Zero-Dispersion Wavelength Distribution for…