’a r o p l b - مواقع اعضاء هيئة التدريس | ksu...

COSTO’2008

Presentation: Habib Fathallah

King Saud University

Université Laval

SYSTÈME DE SUPERVISION ET

D’ADMINISTRATION DES RÉSEAUX

OPTIQUES PASSIFS À LARGE BANDE

1ère Journées Internationales

de Communications Optiques et Systèmes Tout-Optique

Tunis, 6 décembre 2008 COSTO’2008

COSTO’2008

Presentation Outline

Introduction: FTTH PON evolution/ Optical performance monitoring

PON monitoring challenges Limits of OTDR / Existing alternative approaches

Our Optical Coding (OC) System approach 1 D , 2 D, and hybrid 1D/2D encoders

Our OC system vs Standard OCDMA

Performance analysis and results Power budget /Geographical distribution and Source effects

System capacity/Signal to Noise Ratio/Probability of False Alarm

Mirror Cavity based New Coding

OC system for WDM & TDM/WDM PONs

Summary and future work

COSTO’2008

Photonic Access Evolution

Point to Point Active Remote

TDM PON

+ Follows Telco wiring Practices

+ Bit Rate & Protocol Independent

********************************************

- CO Fiber Management

- Fiber Availability

+ Simple deployment model

+ Flexibility of #Users vs Line Rate

+ Bandwidth Upgradeability

*******************************************

- Shared Bandwidth/ Active OSP

- Sub-Optimal Scalability

+ Simplified CO Fiber Management

+ Passive OSP Plant Solution

+ Maturing Technology

*******************************************

- Shared Bandwidth (DS & US)

- Real time Software Intensive

- Complex upgrade/Link Budget

WDM Fiber Access

Wavelength

Filter

+ of P2P & TDM PON + Security, + Reach distance ******************************************* -Still to come

COSTO’2008

Remote Node

Photonic Access Evolution (2)

CO Wireless BH

ONT

Business/

Multi-Tenant Unit ONU

Residential

ONT Indoor/Outdoor

OLT

Service Terminal / Packet Processing

OLT: Optical Line Terminal

ONT: Optical Network Termination

Splitter

TDM over WDM PON:

COSTO’2008

Customers

Standard/products

R&D: Near future

16

32

64

128

16 32

128

256

512

1024

64

GPON

EPON

BPON

APON

TDM or TDM/WDM or CDMA/…


COSTO’2008

FTTN: Fiber-To-The-Node (usually MDF )

FTTC:

Fiber-To-The-Cabinet – usually Street Cabinet (“SC”)

Fiber-To-The-Curb

FTTP: Fiber-To-The-Premise

FTTB:

Fiber-To-The-Building

also: Fiber-To-The-Basement

FTTM: Fiber-To-The-MDU (MDU = Multi-Dwelling-Unit)

FTTH: Fiber-To-The-Home


FTTX:

COSTO’2008

Optical Performance Monitoring: Why

We monitor to:

Administer

Maintain provision

Guarantee a service quality

OPM is the physical layer monitoring in order to:

determine the quality of the signal: power, dispersion, l shift, …

root the cause(s)

Principle of OPM

Good quality signal all network components working well

As Capacity and Complexity increases

Importance of OPM increases.

COSTO’2008

Transparency (Bit

rate, Modulation

format, Protocol)

Reliability

Signal

degradation

discovery

capability

sensitivity

BasicService

quality

Accuracy

Low power

consumption

Commerecial/Customer

Simplicity Low cost!

Value-Added Fault

localization

capability

Scalability

New service

requests

Requirements for Monitoring


We care about …

COSTO’2008

For fault management: Identification,

Localization, …

As a feedback for active modules:

Amplifiers and attenuators for gain balancing

Nodes, switches, routers, attenuators

Dynamic dispersion compensators

As an alarm to predict network failures:

For traffic routing before or after failure occurring.

D. C. Kilper, etal , “Optical performance monitoring,” J. Lightwave Technology., vol. 22, no. 1, 2004.


We exploit the monitoring information …

COSTO’2008











COSTO’2008

Places damaged by wildlife in FTTH infrastructure…

NTT-EAST

Underground

Aerial

Ⅰ：Birds

Ⅱ：Rodents

Ⅲ：Insects

Ⅱ：Rats

Ⅲ：Termites

Ⅰ：Woodpeckers

Ⅲ：Moth larvae

Ⅲ：Ants

Ⅰ: Woodpeckers

Ⅱ:Squirrels, Rats, Japanese flying squirrels

Ⅲ:Moth larvae

Ⅲ：Cockroaches

Ⅰ：Crows

Ⅲ：Cicadas

Wildlife damaged both underground cables and aerial cables.

OPM for FTTX: NTT-Japan case

COSTO’2008

Fault events caused by wildlife in FTTH …

0

5

10

15

20

25

30

1999 2000 2001 2002 2003 2004 2005

Faulty e

vents

0

100

200

300

400

500

600

FT

TH

subscri

bers

(m

illio

ns)

Optical cables

Copper cables

FTTH subscribers

OPM for FTTX: NTT-Japan case

COSTO’2008

13

OPM in FTTX PON Context: Motivation

In-Service and in-time full knowledge of the state of the network

is required because this allows :

ameliorate fault prevention reducing the maintenance cost

Increase the network reliability and reduce the network down-time

Increase the customers satisfaction and reduce their complaints

Contract, guarantee and respect service level agreements with

customers

Speed up the provisioning of new customers and new services

Speed up the installation of new networks and reduce its cost

COSTO’2008

Key issues:

Cost (CAPEX & OPEX)

Capacity

Complexity

Scalability

CAPEX: Capital Expenditure

OPEX: Operational Expenditure

OPM in FTTX PON Context: Cost

COSTO’2008

14 May 2012 15

IN

OUT

OTDR

Fiber under test

Fiber-end

reflection

Rayleigh scattering

OTDR: Measures the back

reflected + backscattered light

The OTDR is adequate for POINT-TO-POINT networks

Challenges of FTTX Monitoring: Limits of OTDR

COSTO’2008

14 May 2012 16

OTDR: Measures the cumulative back reflected + backscattered light

Information about individual branches is lost

OTDR inappropriate for POINT-TO-MULTIPOINT Networks

IN

OUT

OTDR

PS

1625 nm 1675 nm

U band

Challenges of FTTX Monitoring: Limits of OTDR

COSTO’2008

14 May 2012 17

Bragg grating based identifiers with:

Capacity of existing approaches is very limited

IN

OUT

Tunable OTDR

PS

Tunable light

1625 nm 1675 nm

U band

tunable OTDR laser, or

broadband source and tunable filter, or

multi-wavelength laser and tunable filter

Challenges of FTTX Monitoring: Existing approaches

COSTO’2008











COSTO’2008

14 May 2012 19

Encoding the reflected pulses using passive devices

Time coding allows sharing of the same monitoring source

Out-of band coding allows In-Service operation (U band)

Feeder

OLT

OCDM

Monitoring

OC1

OC2

OCN

RNCO

WS

PS

00100001001

10100100000

10000000101

1675 nm1625 nm

U band

FTTX Monitoring: Our Optical Coding Approach

COSTO’2008

14 May 2012 20

FTTX Monitoring: Our Optical Coding Approach (2)

H. Fathallah, and L. A. Rusch, “Network management solution for PS/PON, WDM/PON and hybrid PS/WDM/

PON using DS-OCDM,” OFC 2007, OThE2.

The desired DDF status (healthy/faulty) can be monitored by

observing the auto-correlation peak

Feeder OC2

COOC1

OCDM

Monitoring

OLT RNDDF

WS

U band short-pulse

10001000100000

0100010004000100010

10001000100000

COSTO’2008

14 May 2012 21

Optical Coding Monitoring: 1-D Encoders

1D

PS PS

1DFBG

1D Encoded Sequence

001000100010000

Short Pulse

Autocorrelation Peak

time

Out-of-phase

spikes

With the same

center wavelength

1D Decoded Signal

COSTO’2008

14 May 2012 22

Optical Coding Monitoring: 2-D Encoders

Optical

Pulses

010001000001000001

FBG

001000100010001000

TDLBPFOptical

Pulse

wa

ve

len

gth

wa

ve

len

gth

time time

FBG

λ4

λ1

λ2 λ1λ2λ4FBG

2D

COSTO’2008

14 May 2012 23

Monitoring vs. standard OCDMA applications

Codes transmit monitoring (or state) information instead of data

No data modulation at the encoders

No transmission of streams of data

No need for high speed transmission

OCDM monitoring

Exploits mature and inexpensive components

Avoids high speed electronics

Suffers less from interference

Suffers less from near-far problem

Optical coding: Monitoring vs standard OCDMA

COSTO’2008











COSTO’2008

14 May 2012 25

Performance of OC Monitoring: Power Budget

Signal

Power

Distance

RNDDF

Downstream

Upstream

CM RN CM

feeder

CO

Decoder

CM DDF

RN

feeder

Decoder

1-D Signal

Power

Distance

RNDDF

Downstream

Upstream

CM RN CM

feeder

CO

CM DDF

RN

feeder

Decoder

2-D

2-D reduces the coding-decoding loss to Zero

COSTO’2008

14 May 2012 26

Performance of OC Monitoring: Coding Loss

Mohammad M. Rad 26

1D-FBG

2D-FBG

-20

-10

0

10

20

30

40

50

2 4 6 8 10 12 14 16

Code Weight (w)

Auto

-Cor

rela

tion

Peak

Los

s [d

B]1D- and 2D-PS

En/De Loss

Code weight = 4

COSTO’2008

14 May 2012 27

Performance of OC Monitoring: Beat Noise

Mohammad M. Rad 27

time

Desired

pulses

Interference

pulses

Detected

signal(·)2

Be

Int TH DSNds Big Nii i i i i i

( ) ( ) ( )BN d d BN d Int BN Int Inti i i

Cross-Terms: Beat Noise

(3)- (·)(·) desired-to-desired

(4)- (·)(·) desired-to-Interference

(5)- (·)(·) Interference-to-Interference

Square Terms

(1)- (· )2 Desired Signal (DC)

(2)- (· )2 Interference noise (DC)

COSTO’2008

14 May 2012 28

Example: (Bo « Be)

Bo=10 MHz (DFB Laser),

Be=1GHz (1/pulse duration)

Autocorrelation Peak (DC)

Beat Noise

Beat Noise:

Be-Be BoBo

2 2BN(D-to-I) BN(I-to-I)

2( )BN D to D

Interference Intensity (DC)


COSTO’2008

14 May 2012 29

Be-Be BoBo

2(d-to-int)

2(int-to-int)

BN

BN

2(D-to-D)BN

o

e

B

B

Bo»Be

Autocorrelation

Broadband Sources (BBS) Bo=1 THz, Be=1GHz

Bandwidth consuming!!

1

Be-Be

ω1-ω2>>Be

ω1-ω2 ω2-ω1

Autocorrelation

2D Coding (wavelength & time)

Expensive!!

2


COSTO’2008

14 May 2012 30

time

0 0 0 1 0 0 0 0 1 0 0 0 01 10

1

23

2 233

1 1

Wav

elen

gth

44432

1

4

3

2

1

4

Are chosen to make the

autocorrelation peakDetection Window

No beating between desired pulses

Interference related beating reduces

L. Tansevski, L. A. Rusch, “Impact of the beat noise on the performance

optical CDMA systems,” IEEE Comm. Letters, vol. 4, no. 8, 2000


COSTO’2008

14 May 2012 31

1D severely suffers from Beat Noise

2D Coding Scheme is not Cost Effective

We apply Hybrid 1D/2D OCDM

Feeder OC2

CO

OCDM

Monitoring

OLTDDF

RN

2D

1 32 4

FBG Array

Decoded sequence

1D

Splitter

Reflected sequences

Performance of OC Monitoring: Hybrid Coding

COSTO’2008

32

OC1

OC2

OC3

Desired

Undesired

Undesired

interferes

Autocorrelation Peak

(2F-1)TC

does not interfere

(F-1)TC

Tc: observation window

region subject to interferece (lCD)

Correlation Distance (CD) Only CMs closer than CD can interfere with the desired CM sequence

lCD = c FTc/2CD= Half of the physical length of a code

F: Code Length, Tc: Pulse Width, c: Light Speed in Fiber

Physical distance respect to the CO affects interference statistics

Performance of OC Monitoring: Network Geography

Interference Statistics

COSTO’2008

33

a

a

Case [a] Case [b] Case [c]

Case [d]Case [e]

Area = fixed

/a p

2desired CM

undesired CM

remote node

Performance of OC Monitoring: Geographical distribution

COSTO’2008

34

We assume a linear detection process, i.e., powers add together.

SIR = Average of Interference

Average of Autocorrelation Peak2

Geographical distribution & code choice

Signal-to-interference-ratio (SIR)

Performance OC Monitoring: Geographical distributions

COSTO’2008

35

[a]

[b]

[c]

[d]

[e]

10

20

30

40

50

60

10

20

30

40

50

60

0 200 400 600 800 1000 0 200 400 600 800 1000

[a]

[b]

[c]

[d]

[e]

Network Size K Network Size K

SIR

[dB

]

SIR

[dB

]

[a]

[b]

[c]

[d]

[e]

1D coding0.25 km2 coverage

2D coding0.25 km2 coverage

[a]

[b]

[c]

[d]

[e]

[a]

[b]

[c]

[d]

[e]

10

20

30

40

50

60

10

20

30

40

50

60

0 200 400 600 800 10000 200 400 600 800 1000

Network Size K Network Size K

SIR

[dB

]

SIR

[dB

]

1D coding5 km2 coverage

2D coding5 km2 coverage

Less Dense

Population Coding Gain

Performance of OC Monitoring: Geographical distributions

COSTO’2008

36

[a]

[b]

[c]

[d]

[e]

50 100 150 200 250

[a]

[b]

[c]

[d]

[e]

10

15

20

25

30

6dB

Uniform Radial gives the middle SIR…

Performance of OC Monitoring: Geographical distributions

COSTO’2008

Optical Coding Monitoring;

A more realistic consideration

decoder

SamplingRs=T-1

2

High Gain APD

G

Electrical FilterComparator

DDF

Status

Received signal

37

2

PD RIN SN DN TNi t G r t i t i t i t i t

Signal dependent interference dependent

Dark+Thermal Desired Signal

+ Interference RIN+Beat+Shot

COSTO’2008

PON Monitoring: Signal-to-Noise-Plus-Interference Ratio

Mohammad M. Rad 38

2

22

sig

int n

SNIR

Desired signal: Auto-

correlation Peak

Interference

RIN+Beat Noise

Shot+Dark+Thermal

22 sig nSNR

1 1 1 SNIR SIR SNR

COSTO’2008

PON Monitoring SNIR : Transmitted Power

Uniform Radial

Δl = 564 m

lf = 20 km

Bo = 1 THz (BBS)

Bo = 10 MHz (laser)

Tc = 1 ns

K = 128

2Δl

[e]

10-5

10-4

10-3

10-2

Transmitted Power Ps [watt]

SN

IR &

SN

R [d

B]

-5

0

5

10

15

20

25

2D-Coh-FBG

1D-BBS-FBG

1D-BBS-PSC

1D-Coh-FBG

SNIRSNR

Mohammad M. Rad 39

Single Mode Fiber (4dBm)

COSTO’2008

False-Alarm Probability

Mohammad M. Rad 40

Detection Probability PD = The probability we correctly declare a healthy fiber

Neyman-Pearson Test:

A plot of PD vs. PFA (ideal case: PD = 1, PFA = 0)

PFA = The probability we falsely declare a broken fiber

False-Alarm Probability

COSTO’2008

OC Monitoring: Neyman-Pearson Test

PFA

PD=PFA

Worst Csae

K=512

K<256

K=1024

2D-Coh-FBG, Bo=10MHz

PD

PD

PFA

K=256

Increasing

Network Size

K=128

PD=PFA

Worst Csae

SINR=4.8dB

SINR=-3dB

K<64SINR>12dB

10-5

10-4

10-3

10-2

10-1

100

10-4

10-3

10-2

10-1

100

10-5

1D-BBS-FBG, Bo=1THz

10-5

10-4

10-3

10-2

10-1

100

10-5

10-4

10-3

10-2

10-1

100

SINR=-2dB

SINR=4.7dB

SINR>10dB

Δl = 564 m

lf = 20 km

Tc = 1 ns

Ps = 4 dBm

Uniform Radial

[e]

2Δl

Mohammad M. Rad 41

1D we have PFA ~ 0 and PD ~ 1 for K = 64

2D we have PFA ~ 0 and PD ~ 1 for K = 256

COSTO’2008











COSTO’2008

Periodic Coding: Cavity based Encoder

Code 1(p1 =1) : 1111···

Feeder CM2

CO

U band short-pulse

OLT

1000110···

0010100010···

CM1

OCDM Monitoring

(a)

PSRN

Code 2(p2 =4) : 10001000 ···

(b)

IN

OUT R1 R2#i i sl p T

OUTF

(d)(c)

WS ONT R2<1IN

OUTB il 0R2=1

ONT

OUTF 0il

IN

OUTB

Two grating cavity: simple, low cost, the length fixes the code, etc.

COSTO’2008

Periodic

Coding

R1

0.2 0.4 0.8 1

w = 3

w = 8

0

w = 4

0.8

0.9

1.0

(a) Total code power 1 2 4

Real PC

Ideal PC38%

16.7%

0.1

0.2

0.3

0.4

(b) Ideal vs. Real code pattern

5 6

Real PC

(c) Ideal vs. Real PC normalized autocorrelation for the code: [0, 9, 18, 27, 36, 45]

Ideal PC

Real PC

0.38

0

.2

.4

.6

.8

Aut

ocor

rela

tion

Pow

er

Ideal PC

0

.2

.4

.6

.8

0.167

(d) Ideal vs. Real PC normalized cross-correlation between [0, 9, 18, 27, 36, 45]

and [0, 17, 34, 51, 68, 85]

0

.2

1

.4

.6

.8

0

.2

1

.4

.6

.8

Aut

ocor

rela

tion

Pow

er

Cro

ss-C

orre

latio

n P

ower

Cro

ss-C

orre

latio

n P

ower

w = 2

COSTO’2008

14 May 2012 45

2 ls per customer: 1 for data + 1 for monitoring

Monitoring consumes 50% of the spectral capacity

OUT

Tunable OTDR

WD

M

C band for data

L band for monitoring

IN

Tunable light

C band L band

OC Monitoring of WDM PON

COSTO’2008

14 May 2012 46

Saves 50% of the spectrum resource

Uses standard U band for live monitoring even for WDM-PON

il

XW

DM

1

:KRemote Node

Feeder

CO

Enc11l OC1

OC2

KlOCK

PS

OC Monitoring of WDM PON

COSTO’2008

14 May 2012 47

Still no effective solution is known!

Notes:

• Electronic embedded monitoring techniques in the ONT side are note

addressed here.

• We focus on solutions 100% under the CO control.

il

XW

DM

1:K

Remote Node

Feeder

CO

1l

Kl

PS

OC Monitoring of TDM over WDM PON

COSTO’2008

14 May 2012 48

Only one wavelength monitors the whole TDM/WDM PON

il

XW

DM

1

:K

Remote Node

Feeder

CO

Enc1

1l

OC1

OC2

KlOCK

PS

PS

OC Monitoring of TDM over WDM PON

COSTO’2008

14 May 2012 49

’a r o p l b - مواقع اعضاء هيئة التدريس | ksu...

Documents