An All Optical OTDM Router Based On
SMZ Switch
Razali Ngah, and Zabih Ghassemlooy
Optical Communication Research Group
School of Engineering & Technology
Northumbria University, United Kingdom
http: soe.unn.ac.uk/ocr/
1
Contents









Aim and objectives
Introduction
Optical time division multiplexing (OTDM)
Ultrafast optical time-domain technology - Issues
All optical switches
All OTDM router
Simulations and results
Conclusions + further work
Publications
2
Aim and Objectives
Aim: To develop a novel synchronization
technique using all optical switches for
ultra high speed OTDM networks
Objectives:
1. To study the requirement of ultra high speed
OTDM packet switching
2. To investigate all optical demultiplexing
techniques and devices
3. To develop a novel synchronization technique
using all optical switch
3
Introduction
 Multiplexing:- To extend a transmission
capacity
 Electrical
 Optical

Drawbacks with Electrical:

Speed limitation beyond 40 Gb/s (80 Gb/s future) of:



Electo-optics/opto-electronics devices
High power and low noise amplifiers
Bandwidth bottleneck due to optical-electronic-optical conversion
Solution: All optical transmission, multiplexing, switching, processing, etc.
4
Multiplexing : Optical

Wavelength division multiplexing (WDM)

Optical time division multiplexing (OTDM)

Hybrid WDM-OTDM

5
OTDM

The total capacity of single-channel OTDM network = DWDM

Overcomes non-linear effects associated with WDM:
(i) Self Phase Modulation (SPM) – The signal intensity of a given channel
modulates its own refractive index, and therefore its phase
(ii) Cross Phase Modulation (XPM) – In multi-channel systems, other interfering
channels also modulate the refractive index of the desired channel and therefore its phase
(iii) Four Wave Mixing (FWM) – Intermodulation products between the WDM
channels, as the nonlinearity is quadratic with electric field


Less complex end node equipment (single-channel Vs. multi-channels)
Can operate at both:


1500 nm
1300 nm
6
OTDM : Principle of Operation

Multiplexing is sequential, and could be carried out in:


A bit-by-bit basis (bit interleaving)
A packet-by-packet basis (packet interleaving)
Data (10 Gb/s)
Data (10 Gb/s)
Rx
Span
N*10 Gb/s
10 GHz
Rx
Network
node
Light
source
N
Rx
Drop
Clock
Add
Clock
recovery
Transmitter
Receiver
Fibre delay line
Modulators
Amplifier
OTDM MUX
OTDM DEMUX
Fibre
7
OTDM : Multiplexing of Clock Signal
Clock
Address
Payload
Guard band
(Sync.)
 Space division multiplexing: separate transmission fibre
 time varying differential delay & high cost
 Wavelength division multiplexing: different wavelength
 only practical for predetermined path
 Orthogonal polarization: orthogonally polarized clock pulse
 polarization mode dispersion and other non linear effects
 Intensity division multiplexing: higher intensity for clock pulse
 difficult to maintain in long distance transmission
 Time division multiplexing: self-synchronization - clock is located at the
beginning of the packet)
8
Ultrafast optical time-domain technology :
Issues

Synchronization (all optical clock recovery)


Contention resolution


Clock recovery: using all optical switch combined with
optical feedback
Type: Optical buffering, deflection routing &
wavelength conversion
Routing strategies

Switch-level routing and contention resolution
9
All Optical Switches


Key components required in all optical signal
processing for ultrahigh speed OTDM networks
Applications:




Optical cross-connects: provisioning of lightpaths
Protection switching : rerouting a data stream in the
event of system or network failure
Optical Add/Drop multiplexing: insert or extract optical
channels to or from the optical transmission system
Optical signal monitoring
10
All Optical Switches – contd.
Non-linear Optical Loop Mirror (NOLM)
Terahertz Optical Asymmetric Demultiplexer
(TOAD)
x
Long fibre
loop
Control
coupler
Control
pulse
SLA
Fibre
loop
CW
CCW
Control
Pulse c
Coupler
Data in
Port 1

Port 2
Data
In s
PC Data
out
Requires high control pulse energy and long
fiber loop

CW
CCW
Data
out
Coupler
PC
Asymmetrical switching window profile due
to the counter-propagating nature of the
data signals
11
All Optical Switches – contd.
Symmetric Mach-Zehnder (SMZ)


Symmetrical switching window profile
Integratable structure
12
All Optical Switches – contd.
Comparative study of all optical switches [Prucnal’01]
Device
Switching
Time
Repetition
Rate
(GHz)
Noise Figure
(dB)
Ease of
Integration?
Practicality
SMZ
< 1 ps
100+ GHz
6
YES
HIGH
TOAD
< 1 ps
100+ GHz
6
YES
MEDIUM
NOLM
0.8 ps
100+ GHz
0
NO
LOW
UNI
< 1 ps
100+ GHz
6
NO
MEDIUM
13
SMZ Switch : Principle
(i) No control pulses
Control Pulse
Input Port 1
SOA1
OTDM Signal Pulses
Control Pulse
Input Port 2
Output
Port 2
3 dBCoupler
SOA2
(ii) With control pulses
Control Pulse
(switch-on)
SOA1
Output
Port 1
OTDM Signal Pulses
Optical
filter
Tdelay
Control Pulse
(switch-off)
3 dBCoupler
SOA2
14
SMZ : Switching Window

1
W (t )  G1 (t )  G2 (t ) 2 G1 (t )G2 (t ) . cos (t )
4
  0.5LEF ln( G1 / G2 )
G1 and G2 are the gains profile of the data signal at the output of the SOA1 and SOA2, ΔФ is
the phase difference between the data signals, and LEF is linewidth enhancement factor
Gain Profile of Gc1(__) and Gc2(--)
20
SMZ switching window
25
18
20
16
SMZ gain
15
Gain
14
10
12
5
0
40
10
45
50
55
60
Time (ps)
65
70
75
8
6
4
2
40
45
50
55
60
65
Time (ps)
70
75
80
85
90
15
SMZ : Switching Window (simulation)
TABLE I. SIMULATION PARAMETERS
Parameter
Value
SOA
. LengthLSOA
0.3 mm
. Active area,
3.0x10-13 m2
. Transparent carrier density, No 1.0x1024 m-3
. Confinement factor, 
0.15
. Differential gain, g
2.78x1020 m2
. Linewidth enhancement,
4.0
. Recombination coefficient A
1.43x108 1/s
. Recombination coefficient B
1.0x10-16 m3/s
. Recombination coefficient C
3.0x10-41 m6/s
. Initial carrier density
2.8x1024 m-3
. Total number of segments
50
Data and control pulses
. Wavelength of control & data 1550 nm
. Pulse FWHM
2 ps
. Control pulse peak power
1.2 W
. Data pulse peak power
2.5 µW
16
SMZ : Switching Window (comparison)
SMZ Switching Window
20
20
15
15
SMZ Gain
SMZ gain
SMZ switching window (Cross)
10
10
5
5
0
45
50
55
60
Time (ps)
65
Theoretical
70
75
2.025
2.03
2.035
2.04
Time (s)
2.045
2.05
2.055
2.06
-9
x 10
Simulation
17
SMZ : Switching Window (experimental)
Experimental switching window profile of the SMZ [Toliver’00 Opt. Comm]
18
SMZ : On-Off Ratio

The ratio of the output power in the on-state to the output power in
the off-state
Target signal
Crosstalk
Input signal of the SMZ
Transmitted output of the SMZ
19
SMZ : On-Off Ratio – contd.
16.0
1.20
14.0
1.00
On-off ratio (dB)
0.60
8.0
6.0
0.40
4.0
0.20
18
On-off ratio (dB)
0.80
10.0
Normalised transmission power
12.0
20
16
14
12
10
8
6
4
2
0
10
40
80
100
160
Bit rate (Gb/s)
2.0
0.0
0.00
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
5.5
6
6.5
7
On-off ratio at different data rate
Linewidth enhancement factor
On-off ratio and normalised transmission power
Against linewidth enhancement factor
20
SMZ : BER Performance
Receiver parameters
___________________________________
Parameter
Value
Pre-amplifier
Mode
Noise Figure
Gain
Gain controlled
4 dB
25 dB
PIN detector
Responsivity
1 A/W
Thermal noise
10 pA/Hz1/2
Cutoff frequency
7.0x109 Hz
__________________________________________
21
SMZ : BER Performance – contd.
0
10
back-to-back 10Gb/s
SMZ 4x10Gb/s
SMZ 8x10 Gb/s
SMZ 16x10 Gb/s
-2
10
-4
10
-6
10
-8
BER
10
-10
10
-12
10
-14
10
-16
10
-18
10
-20
10
-44
-42
-40
-38
Received power (dBm)
-36
-34
BER against the average received power for (a) back-to-back without demultiplexer,
(b) 40 – 10 Gb/s demultiplexer, (c) 80 – 10 Gb/s demultiplexer and (d) 160 – 10 Gb/s
demultiplexer
22
SMZ : BER Performance – contd.
Comparison with experimental results
Ngah’04
Tekin’02
IWC4
Diez’00
Elec. Lett
Hess’98
PTL
Jahn’95
Elec. lett
Back-to-back
(10 Gb/s)
Sensitivity
-38 dBm
-35 dBm
-35 dBm
-34 dBm
-37 dBm
40-10 Gb/s
demux.
Power penalty
1.2 dB
NA
NA
0 dB
2.5 dB
80-10 Gb/s
demux.
Power penalty
1.4 dB
1 dB
1.2 dB
4 dB
NA
160-10 Gb/s
demux.
Power penalty
1.5 dB
3.5 dB
2.8 dB
NA
NA
23
1x2 All OTDM Router
( a)
( c)
SMZ1
(clock
extract)
(e)
SMZ2
(read
address)
Port 1
(f)
SMZ3
(route
payload )
( b)
Port2
(d)
(a) OTDM Signal
(b) Extracted Clock
(c) Address + Payload
(d) Address
(e) Payload
(f) Payload
24
OTDM Router : Synchronization
Self-synchronization:
 low hardware costs and control control complexity
 require a single pulse in the first bit position of the packet
Control Pulse
(switch-off)
OFDL
SOA1
OTDM
Signal
Optical
filter
Output
Port 1
Control Pulse
(switch-on)
OFDL
3 dBCoupler
SOA2
PC
Clock, Address and payloads have the same intensity, polarization, width and wavelength
25
OTDM Router : Synchronization (simulation)
26
OTDM Router : Simulation Results
Crosstalk
OTDM packet signal
Extracted clock from the OTDM packet
27
OTDM Router : Simulation Results –contd.
40.0
On-Off Ratio (dB)
35.0
30.0
25.0
20.0
15.0
10.0
5.0
0.0
200
400
600
800
1000
1200
Bits Period (ps)
The on-off ratio against the bit period
28
OTDM Router : Simulation Results – contd.
Clock extraction and
demultiplexing for OTDM
packet signal
Crosstalk
Demultiplexed payload at the transmitted port
29
OTDM Router : Simulation
30
OTDM Router : Simulation Results
OTDM input packet
Clk
Add
Payload
31
OTDM Router : Simulation Results – contd.
Extracted clock signal at the reflected output of SMZ1
32
OTDM Router : Simulation Results – contd.
Add
Payload
Data packet at the transmitted output of SMZ1
33
OTDM Router : Simulation Results – contd.
Address bit at the reflected output of SMZ2
34
OTDM Router : Simulation Results – contd.
Payload at the transmitted output of SMZ2
35
OTDM Router : Simulation Results – contd.
Payload at the port 1 of SMZ3
36
Performance Issues
(1) Relative Intensity Noise (RIN)

Relative timing jitter between the control and the signal pulses
induces intensity fluctuations of the demultiplexed signals
37
Relative Intensity Noise (RIN)


The output signal can be described by:
w(t ) 
T
x
(t ) p (t   ) dt

where Tx(t) is the switching window profile and p(t) is the input data profile

The expected of the output signal energy is given as:
E ( ) 

 w(t ) p (t   )dt
t


pt(t) probability density function of the relative signal pulse arrival time:
1
pt (t ) 
e
2 t RMS
1 t
 
2  t RMS
2


where tRMS is the root mean square jitter
38
Relative Intensity Noise (RIN) – contd.

The variance of the output signal, depending on the relative arrive time is:
Var ( ) 

2
2
w
(
t
)
p
(
t


)
dt

E
( )
t



Assuming that the mean arrival time of the target channel is at the centre
of the switching window, RIN induced by the timing jitter of the output
signal can be expressed as:
Var ( )
RIN ( )  2
E ( )

The total RIN for the router is three times the value of single SMZ
39
Performance Issues – contd.
(2) Channel Crosstalk (CXT)

Due to demultiplexing of adjacent non-target channels to the output port
when the switching profile overlaps into adjacent signal pulses
40
Channel Crosstalk (CXT) – contd.

CXT is defined by the ratio of the transmitted
power of one non-target channel to that of a
target channel



Et is the output signal energy due to the
target channel
Ent is the output signal energy due to the
nontarget channel
The total crosstalk for the router
Ent
CXT  10 log
Et
Et 
tc TD / 2
T
x
(t ) p(t  t c )dt
tc TD / 2
Ent 
tc T TD / 2
T
x
(t ) p(t  t c )dt
tc TD / 2
CXT  (1  CXT )3 1
41
BER Analysis

Assuming 100% energy switching ratio for SMZ and the probability of mark
and space are equal, the mean photocurrents for mark Im and space Is
are:
__
__
I m  I sig [1  CXTn ]
I s  I sig[CXTn ]
___
I sig  RinoutGLPsig
where R is the responsivity of the photodetector, ηin and ηout are the input and output
coupling efficiencies of the optical amplifier, respectively; G is the optical amplifier
internal gain, L is optical loss between amplifier and receiver, and Psig is the preamplified average signal power for a mark (excluding crosstalk)

The variance of receiver noise for mark and space:
2
rec , x
_

4
KT
2
2
2
k
  s   th  2q( I x  I ASE ) Be  
 i a  Be
 RL

___
42
BER Analysis – cont.

The noise variance of optical amplifier

The noise variance of RIN



2
RIN , m
2
The expression for calculating BER is given as:
where

and
The average photo-current equivalent of ASE
___
Q
___
Im  Is
2
4I x I ASE Be I ASE
Be (2Bo  Be )


Bo
Bo2
2
2
 I m RIN T Be  I sig RIN ROUTER

2
amp, x
2RIN , s  I s RIN T Be
I ASE  Nsp (G  1)outqBo L
BER 
1 exp( 0.5Q 2 )
Q
2
2
 Total
The total variance
2
2
2
2
2
 Total
  rx2 ,m   rx2 ,s   amp






,m
amp, s
RIN ,m
RIN , s
43
BER: Theoretical Results
Block diagram of a router with a receiver
Incoming
OTDM
Signal
Pin
Receiver
1x2 Router
t = st
Pk
SMZ
1
SMZ
2
Photodetector
SMZ
3
BER
Filter
Optical
Amp.
Clock
Address
Optical path
Electrical path
System Parameters
Parameter
Value
in
-2
dB
out
-2
dB
out
Gain
(overall)
25 dB
L
-2
dB
R
1
A/W
RL
50

Tk
293
K
Nsp
2
RINT
10-15
Hz-1
Bo
400
GHz
Ia2
100
pA2
/Hz
RINR
OUTER
RMSji
CXT
tter
n
-21
dB
1 ps
-25
dB
Be
0.7
Rb
44
RIN and CXT : Results
0
-8
OTDM router
SMZ demultiplexer
FWHM = 2ps
-10
-5
-10
-12
SMZ crosstalk (dB)
Relative intensity noise (dB)
-15
-14
-16
-18
-20
-25
-30
-20
-35
-22
-40
-24
-26
OTDM router
SMZ demultiplexer
FWHM = 2ps
-45
0
2
4
6
8
10
12
Control signals separation (ps)
14
16
18
RIN against control pulse separation for a
single SMZ and a router
20
-50
0
2
4
6
8
10
12
Control signals separation (ps)
14
16
18
20
CXT against control pulse separation for a
single SMZ and a router
45
BER : Results
10Gb/s baseline
10Gb/s with router
-2
10
-4
10
Bit error rate
-6
10
-8
10
-10
10
-12
10
-44
-42
-40
-38
-36
-34
-32
-30
Average received optical power (dBm)
-28
-26
-24
-22
BER against average received power for baseline and with an optical router
46
BER : Simulation Results
0
10
10Gb/s back-to-back
10Gb/s with router
-2
10
-4
10
-6
10
-8
BER
10
-10
10
-12
10
-14
10
-16
10
-18
10
BER increases with the number of SMZ
stages due to the accumulation of ASE
noise in the SOAs hence, resulting the
RIN increases.
-20
10
-44
-42
-40
-38
-36
-34
-32
Received power (dBm)
-30
-28
-26
-24
-22
BER against average received power
for baseline and with an optical router
47
Conclusions




All optical demultiplexer and 1x2 router based on
SMZ has been implemented in a simulation
environment using VPI.
BER analysis has been performed.
The application of low noise SOA will reduce the
power penalty.
SMZ switch becomes a key component for ultra
high speed OTDM networks.
48
Publications
Conference
(1)
R. Ngah, Z. Ghassemlooy, G. Swift, T. Ahmad and P. Ball, “Simulation of an all Optical Time Division Multiplexing Router
Employing TOADs”, 3rd Annual Postgraduate Symposium on the Convergence of Telecommunications, Networking &
Broadcasting, Liverpool, 17-18 June 2002, pp. 415-420.
(2) R. Ngah, Z. Ghassemlooy, and G. Swift, “Simulation of an all Optical Time Division Multiplexing Router Employing Symmetric
Mach-Zehnder (SMZ),” 7th IEEE High Frequency Postgraduate Student Colloquium, London, 8-9 Sept. 2002, pp. 133-139.
(3) R. Ngah, Z. Ghassemlooy, and G. Swift, “40 Gb/s All Optical Router Using Terahertz Optical Asymmetric Demutiplexer (TOADs)”
International Conference on Robotics, Vision, Information and Signal Proceeding, Penang Malaysia, 22-24 Jan 2003, pp. 179183.
(4) R. Ngah, Z. Ghassemlooy, and G. Swift, “Simulation of 1 X 2 OTDM router employing Symmetric Mach-Zehnder (SMZ)”
Postgraduate Research Conference in Electronic, Photonics, Communication & Networks, and Computing Science, Exeter, 14-16
April, pp 105-106.
(5) R. Ngah, Z. Ghassemlooy, and G. Swift, “Comparison of Interferometric all-optical switches for router applications in OTDM
systems” 4th Annual Postgraduate Symposium on Convergence of Telecommunications, Networking and Broadcasting,
Liverpool, 16-17 June 2003, pp. 81-85.
(6) A. Als, R. Ngah, Z. Ghassemlooy, and G. Swift, “Simulation of all-optical recirculating fiber loop buffer employing a SMZ switch”
7th World Multiconference on Systemics, Cybernetics, and Informatics, Florida, 27-30 July 2003, pp 1-5.
(7) R. Ngah, and Z. Ghassemlooy, “BER performance of an OTDM demultiplexer based on SMZ switch” Postgraduate Research
Conference in Electronic, Photonics, Communication & Networks, and Computing Science, Hetfordshire, 5-7 April 2004, pp 228 –
229.
(8) R. Ngah, and Z. Ghassemlooy, “Bit Error Rate Performance of All Optical Router Based on SMZ Switches,” First IFIP International
Conference on Wireless and Optical Communications Networks (WOCN 2004), Oman, 7 – 9 June 2004, Accepted for
publications.
(9) R. Ngah, and Z. Ghassemlooy, “The Performance of an OTDM Demultiplexer Based on SMZ Switch,” IEE Seminar on Future
Challenges and Opportunities for DWDM and CWDM in the Photonic Networks, University of Warwick, 11 June 2004, Accepted
for publications.
(10) R. Ngah, and Z. Ghassemlooy, “Simulation of Simultaneous All Optical Clock Extraction and Demultiplexing for OTDM Packet
Signal Using SMZ Switches,” 9th European Conference on Networks & Optical Communications (NOC 2004), Eindhoven, 29 June
– 1 July 2004, Accepted for publications.
(11) R. Ngah, and Z. Ghassemlooy, “Noise and Crosstalk Analysis of SMZ Switches,” International Symposium on Communication
Systems, Networks and Digital Signal Processing, University of Newcastle, 20 - 22 Juuly 2004, Accepted for publications.
49
Publications – contd.
Journal
(1) R. Ngah, and Z. Ghassemlooy, “Simulation of 1x2 OTDM
Router Employing Symmetric Mach-Zehnder Switches”
Accepted for publications in IEE Proceeding Circuits,
Devices & Systems.
50
Acknowledgement
 Thanks to the University of Teknologi Malaysia for
sponsoring the research.
51
THANK YOU
52