1

Optical Performance Monitoring Applications in Transparent Networks

Dan KilperAdvanced Photonics Research

Lucent Technologiesdkilper@lucent.com

C. R. Giles, W. Weingartner, A. Azarov, P. Vorreau, and J. Leuthold

WOCCApril 22, 2005Newark, NJ

2

Old Technology

Ultra-long Transport SystemsPoint-to-Point Transparency Current

ROADM ROADM

EndTerminal(opaque)

EndTerminal(opaque)

4000km at 10Gb/s>2000km at 40Gb/s

OA Repeaters

Mitigation of:NoiseDispersionGain variationNonlinearity

Advanced Technologies:Raman AmplificationDispersion Managed SolitonsDynamic Gain EqualizationDPSK, Advanced Modulation FormatsSEA

SF

LA

DEN

MIN

CHI

DAL

HOU

ATL

NY

WA

KC

3

ULH+ROADM/OXC MESH NETWORKULH+ROADM/OXC MESH NETWORK

N

E

S

W

λ1 λ2λ3λ4λ5 λ6λ7MOADMA B

ET

ET

ET

ET

C

D

λ4λ5

λ1 λ2λ3 λ7λ12

λ6

λ10λ11λ12

λ10λ11

Operational Complexity +Advanced Technology

Advanced MonitoringTransparent Reconfiguration• Intersecting lines must discover one another and exchange topology information.• Auto-provisioning must operate across the mesh network.• Faults are correlated across multiple systems.

• Greater flexibility requires better stability & control

4

Optical Network Performance Monitoring• First Generation: Total power monitoring. Amplifier gain adjustment, signal presence, link status verification.

• Second Generation: WDM channel presence / power and wavelength.Auto-provisioning and gain flattening.

• Third Generation: Channel optical SNR / Q-factor, active dispersion compensation. Fault isolation, dispersion compensation.

• Fourth Generation: Transparent network management. Channel performance verification after link concatenation.

• Fifth Generation: In-situ link parameter extraction from detailed channel signatures. Preplanning / preprovisioning assessment. Resource database creation.

TODAY

5

Eliminating Regenerators

• Must also consider fault management requirements• Cost of OADM/ULH technology (DGEF)/OPM < OEO

500 km 500 km 600 km OEOOEOOEOOEOOEO

OEOOEOOEOOEOOEO

OEOOEOOEOOEOOEO

OEOOEOOEOOEOOEO

1600+ km

OPMOPM

OADM

DGEF OEOOEO

OEOOEO

OEOOEOOEOOEOOEO

OEO

6

3G: DWDM Fault Management

OPM OPM OPM

OEOOEOOEOOEOOEO

OPM OPM OPM OPMOPM OPM

Report BER degradation

Read out OPM history:compare actual performance

with stored reference

Locate degradation:dispatch maintenance

• Advanced technologies/network complexities– Component alarms may be insufficient

• Need OPM that correlates with end terminal BER– OPM registers change when end terminal BER alarm triggers

• OPM granularity to suit carrier opex goals

OADM

OEOOEOOEOOEOOEO

OADM

7

Electronic Fault Management

• BER monitoring is sufficient– No errors in: No errors out– Noise does not propagate past regenerators

• Isolate faults to ~600 km

OSNR:BER

Distance

500 km 500 km 600 km OEOOEOOEOOEOOEO

OEOOEOOEOOEOOEO

OEOOEOOEOOEOOEO

OEOOEOOEOOEOOEO

8

Ultra-Long Haul Transmission

1800+ km

• Replace OEOs with OA repeaters: lose fault isolation• BER at OA repeaters has limited benefit• Noise propagates through repeaters

Error Free

Distance

OSNR:

OPM OPM OPM

OEOOEOOEOOEOOEO

OEOOEOOEOOEOOEO

ErrorFree

9

5 10 15 20 25 3010

15

20

25

30

35

OSN

R (d

B)

Node #

‘Good’

‘Impaired’

Fault Location

BER:10-12

10-9

Fault Metrics:10 dB BER drop10-9 BER threshold

Fault Isolation

• Need sensitivity to wide variety of impairments.• BER 10-9 gives ~ 4 orders of magnitude

advanced warning in FEC-based links.

10

• BER Measurement– Sensitive to end terminal impairments– Problem: BER in network better than end term.

• Noise loaded BER measurement– Sensitivity close to BER

• Other methods: OSNR, half-clock, pol. ext., histograms, tones, autocorrelation, …– Must show advantage over Q/BER approach

• Cost/sensitivity/impairment coverage– Target systems that cannot use Q-factor

Q Factor

OPM Fault Management Technologies

11

Q Factor• Signal to Noise Ratio Measurement

10G RZ Eye Diagram

Vth

NoiseSignalQ →

+−

=01

01σσµµ

µ0

σ1

σ0

µ1

( )

−+

−=

22 0

0

1

141

σµ

σµ thth

thV

erfcV

erfcVBER

12

Q Factor Monitoring Techniques(A) Variable threshold, dual decision – eye mapping(B) Variable threshold, use FEC/integrate data (C) Asynchronous histogram methods

• Sensitivity questions(D) Sampling techniques Asynch.

part

(C)

Vthr

(B) (D)(A)

Vthr

Vref

13

FEC Error Count Eye Mapping• Vary voltage threshold across center of eye• Use commercial 10 Gb/s receiver

1.1 1.2 1.3 1.4 1.5 1.6 1.7

-10

-9

-8

-7

-6

-5

-4

-3

-2

Log(

BER

)

Decision Level (Volts)

Low OSNR High OSNR

14

Q-factor vs. time• Determined measurement noise contributions

under different conditions• Error due to counting statistics, threshold voltage

accuracy, power fluctuations

0 2000 4000 6000 8000 1000013

14

15

16

17

18

19

20

FixedDecisionLevel

VariableDecisionLevel

Q-fa

ctor

(dB)

Time (sec)

Vary OSNR

15

Dispersion map issues

• Q factor varies with dispersion map• 10Gb/s: up to 1000 ps/nm

– OK for trend monitoring• 40 Gb/s: eye closed until end terminal

– Would need per-channel DCM/tunable DCM– Also obstacle to 40G optical networks

……

+300 ps/nm

Distance

node

∆Q sensitivity isweakly dependenton magnitude ofQ factor

-300 ps/nm

16

OSNR/Dispersion• Measure Q-Factor up

to –982 ps/nm accum. dispersion

• OSNR sensitivity only weakly dependent on dispersion

Use DCMs & SSMFto add dispersion

10 15 20 25 30 350

1

2

3

4

5

6

7

8

|Qm

eas -

Qbk

gd| (

dB)

OSNR (dB)

0 ps/nm -512 ps/nm -982 ps/nm

14 16 18 20 22 24 26 28 30 32 349

101112131415161718192021

0 ps/nm -512 ps/nm -982 ps/nm

Q-F

acto

r (dB

)

OSNR (dB)

Dispersion managed solitons: pulses retain shape throughout transmission!

• Always within receiver Q-factor range

17

Sensitivity varies with monitor location

• OSNR, non-linear impairments accumulate with distance

• Dispersion follows map

0 5 10 15 20 25 301214161820222426283032

BER:

10-9

10-14

Q2 (

dB)

Span #

Pre-compensation: 0 ps/nm -300 ps/nm -500 ps/nm -800 ps/nm

( ) 22201

ASEASEPBeat

P

σ++σDσ

-IDIQ =

( )21 fDD AP =

DA=Accum. Dispersionf = scaling factor(4 dB @ 800 ps/nm)

Dispersion Penalty

Calculate “optical” Q on line:Monitor independent:

OSNRDrop

18

Dispersion faults• Strongly dependent on map• Look for discontinuities along path• Use +/- bands to identify dispersion problems

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 3014

16

18

20

22

24

26

28

30

32Pre-compensation:-800 ps/nm

+624 ps/nm

-624 ps/nmQ2 (d

B)

Span #

Slope changes in Q trend

+/- 624 ps/nm for10-9 BER degradation

19

Performance Polling: Tunable Filter + OA

O/EO/E PPQQ-20 dB

Tap

Span loss ~ 20 dB

Filter OA

• Guarantee equal or better sensitivity than end terminal• Replace entire OEO terminal with single OE, channel

selector, and single channel OA• O/E provides BER, conventional PM, Q-factor, average

power, channel presence, wavelength drift

20

WDM vs. (O)TDMWDM: Access signals with OE throughout system

OE: Q

OEOOEOOEOOEOOEO

OEOOEO

OEOOEO

OEOOADM

OE: Q

OTDM: OE not available/feasible within network

OPM?

O3R

OEOOEOOEOOEOOEO

OTDM

OADM

OPM?

21

Quality of Service (QoS): per channel BER

QoS Monitoring in Transparent Networks

OEOPM

OEOPM

OEOPM

600 km 400 km 500 km

How do we monitor in an all-optical network?

OEOOEOOEOOEOOEO

OEOOEOOEOOEOOEO

OEOOEOOEOOEOOEO

OEOOEOOEOOEOOEO

O3ROPMO3ROPMO3ROPM

22

Regeneration Applications• Unambiguous indication of signal quality

– Correlation with common impairments• Do not need to isolate or measure impairments• No contingencies on relative impairment

contributions

• Absolute measure of signal quality– Usually only coarse measure

• Error free/not error free• Guarantee above threshold: 10-14 BER

• Satisfy operating requirements of system– System specific: input power, modulation

format, etc.

23

Optical Regeneration + Monitoring

SOA

Filter

CW

Data

Pdata

Pcv

1λ

2λ

2λ

For given input power: more or less power will arrive at the output depending on the input signal quality and the filter characteristics

24

Unambiguous Quality Indicator: Pout/Pin

-150 -100 -50 0 50 100 1500.0

0.2

0.4

0.6

0.8

1.0

Mon

itor S

igna

l (a.

u.)

Dispersion (ps/nm)12 14 16 18 20 22 24 26 28 30

0.6

0.8

1.0

OSNR (dB)

Dispersion

Unable to isolate noise or dispersionBut....

Monitoring signal decreases with decreasing signal quality

Noise

25

Summary

• Transparent optical networks generate a need for new system monitoring and management methods

• Focus on applications will drive technology development

• Fault management: Q-factor natural replacement for BER

• Regeneration applications: solutions tied to regeneration technologies & provide BER trend

26

Back-Up Slides

27

Is spectral OSNR useful?

1565 1566 1567 1568 1569 1570

-30

-20

-10

0R

elat

ive

Pow

er (d

B)

Wavelength (nm)

≠ OSNR

Noise Floor, 17 dB & 32 dB

10 Gb/s RZ on 50 GHz grid

Problems:• Tight channel spacing: overlapping spectra• Per-channel OSNR (OADM/OXC networks)• Filters modify spectra (OADM/OXC)• Poor coverage: MPI, pump RIN transfer, FWM

28

Is spectral OSNR useful?

Yes, under following constraints:

• OSNR-degradation is only impairment of interest or major impairment

• Channels are widely spaced in wavelength– Or spectral regions reserved for monitoring

• Used for amplifier monitoring (not channel monitoring)– Don’t follow channels through ROADM/OXC