Improving scalability and data security
of incoherent OCDMA systems by employing
ultrafast all-optical signal processing techniques
Prof Ivan Glesk
Prof Ivan Glesk
Department of Electronic and Electrical Engineering
3rd International Workshop on OPS and OCDMA, October 23-24, 2008, Shanghai, China Lightwave Communications Research Laboratory
Princeton
University
Acknowledgement
Prof Paul R. Prucnal
Dr. Lei Xu
•
Introduction
•
Incoherent OCDMA
•
Enhancing performance with all-optical signal processing
–
MAI suppression
–
Improving security
–
Improving system power budget
–
increasing number of simultaneous users
–
Increasing scalability and spectral efficiency with M-ary encoding
•
Experimental results
•
Conclusions
Outline
2D Wavelength-Hopping Time-Spreading Codes
Incoherent OCDMA based on 2D prime codes
Incoherent OCDMA based on 2D prime codes
•
Codes generated by simple one-line algorithm:
w = weight: # of wavelengths
p = code length: # of chips
i = code number
for j = 0: w-1
code(i +1, j +1) = mod(j * i, p)
+
1
End
1
p
λ
1λ
wBit Interval
Time [ps]
Example: (4,11) 2D-WHTS
•
Analytical expression for upper bound on BER
•
•
The code scheme uses picosecond pulses at multiple wavelength to generate codes
Experimental set up to study laser coherence
Multiwavelength ps laser
Multiwavelength ps laser
)
cos(
2
1 22
1
I
I
I
δφ
I
I
tot=
+
+
Interferometric equation:
4I
02I
0“same” pulse interaction
Pulse is coherent with “itself”
“pulse to pulse” separation
NO interference is observed
Experimental results:
- 1.6 ps near transform-limited pulses efficiently utilize bandwidth
- Turn-key operation (< 40 fs timing jitter)
- High output power (15 dBm per wavelength)
OCDMA Transmitter transmitting 2D
OCDMA Transmitter transmitting 2D
-
-
WHTS codes
WHTS codes
MOD
RF Data In: 1 0 1
λDe M U X
M U X
D
D
D
D
2D Tunable Encoder
Encoded bit “1”
λ
1λ
pps multi -
λ
laser
Receiver diagram
Receiver diagram
2D Tunable Decoder
From Network
Electronic Data bit
“1” λ
De M U X
M U X
D
-1D
-1D
-1D
-1D = reconfigurable optical delay lines
PD
PD
D-1 = indicates “inverse” delay in reference to the delays D in the Transmitter
Decoded signal: Data bit “1”
Cross-correlation signal from multiple users
τ
chipτ
PD2D-code
OC
OC
-
-
48 OCDMA Princeton
48 OCDMA Princeton
’
’
s Testbed
s Testbed
•
OC- 48 data rate
•
4 Transmitters
•
1 Tunable Receiver/Decoder
2D (4,101) WHTS codes
Control Interface
Performance
Performance
–
–
with 4 simultaneous users
with 4 simultaneous users
-20 -18 -16 -14 -12 -10 -8 -6 -4
1E-12 1E-11 1E-10 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3
Bit Error Rate (log)
Power (dBm)
Transmitter 3
Transmitters 1,3
Transmitters 1,2,3
Transmitters 1,2,3,4
PD: 15 GHz photodiode
Transmitter 1 Transmitter 2 Transmitter 3 Transmitter 4
Performance of each user
Performance of each user
From Network
Electronic Data Out λ
De M
U X
M U X
D
-1D
-1D
-1D
-1PD
PD ~
τ
PD2D-code
Electronic Data Out
λ
De M U X
M U X
D
D
D
D
All-optical gating
From Network
2D-code
Eliminating MAI with 2ps time gating
Eliminating MAI with 2ps time gating
PD
PD
Novel OCDMA receiver design
Novel OCDMA receiver design
Conventional OCDMA receiver
~
τ
PD<
τ
chipTOAD
TOAD
-
-
Terahertz Optical Asymmetric Demultiplexer
Terahertz Optical Asymmetric Demultiplexer
Glesk et. al., "Demonstration of All-Optical Demultiplexing of TDM Data at 250 Gb/s," Electronics Letters 30, 339 (1994)
Switching Window
sw =
(2Δ
x)/c
t 0
Δ
n
ΔΦ
~
Δ
n
Data IN
Data OUT
Clock IN
Δ
x
50/50
LP
P
NLE
AD
1/9Clock
C B A
B
I = 120 mA
0 10 20 30 40
12 13 14 15 16 17
Input
Output SW=f(Δx)
1.8 ps
•
Picosecond all-optical gating
•
Low control pulse energy ~500fJ
•
High SNR (BER < 10
-9)
•
Was integrated
•
6 dB gain
Switching Window is a Function of SOA Displacement
Switching Window is a Function of SOA Displacement
0 5 10 15 20 25 30 35 40 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Time [ps]
TOAD Switching Windows (SW)
Calculated Measured
0 5 10 15 20 25 30 35 40 100 110 120 130 140 150 160 170 180 190 Time [ps]
Glesk et. al., "Polarization Insensitive Terabit Optical Demultiplexers," Proc. SPIE 2481, p. 13 (1995). Invited paper
- 5 0 5 1 0 1 5 2 0 2 5 3 0
0 1 0 2 0 3 0 4 0 5 0 6 0 7 0
T D L S e ttin g s ( m m )
E x p e r im e n t a l M e a s u r e m e n t s
Δ
x
SOA Data In Midpoint Clock DeMux Out Stretching TOAD{
( ) ( ) 2 ( ) ( ) cos( ( ) ( ))}
4 G1 t + G2 t ± G1 t G2 t
φ
1 t −φ
2 t=
1( )
( )
t SW (t) It I
In
TOAD
TOAD
-
-
based receiver demonstration
based receiver demonstration
TOAD
TOAD
–
–
based OCDMA Receiver
based OCDMA Receiver
OCDMA Receiver
OCDMA Receiver
–
–
NO Time Gating
NO Time Gating
Achieved
Performance improvement with all
Performance improvement with all
-
-
optical time gating
optical time gating
Number of simultaneous users
Probability of Error
(4,101) 2D-Wavelength-Hopping Time-Spreading prime codes
10-9
2ps time gating
2ps time gating
0 10 20 30 40 50 60 70 80 90 100
No time gating
No time gating
Special features:
•
Multi-level Security for users
•
Implemented
One-time Pad
in optical domain to secure data
In collaboration with Lockheed Martin
-
OCDMA based
- bus network architecture
Secure communication platform
Secure communication platform
for avionics applications
for avionics applications
4 × 1 c o mb in er
RF Code swapping
EDFA Tx-A Opt Enc
Tx-B Opt Enc
Tx-Dc Opt Enc
RF In Data A
RF In Data B
RF In Data D
Server p s mult iw av e len g th laser 3 3
Tx-Dd Opt Enc
3 MOD MOD 2 x 2 2 x 2
λ1,λ2
,λ3
1% taps
optical bus
99:1 99:1
99:1 RF Out
Data A
PD OptDec Da
Rx-A RF Out
Data B PD OptDec Db Rx-B RF Out Data C PD OptDec Dc Rx-C Eavesdrop
RF Code swapping
RF Out Data D OptDec Dc OptDec Dd 2 x 2 Rx-D PD
Concept of One
Concept of One
-
-
time Pad
time Pad
Electronic XOR
Problems:
–
RF signature radiation which may be
vulnerable to side channel attacks
–
Electronic speed limited to a few GHz
Optical layer XOR
Advantages:
–
Encoded data never exists in electronic
form
–
No RF signature is generated
Data
Data KeyKey XOR OutXOR Out
0
0
0
0
1
1
1
0
1
1
1
0
Traditional electronic XOR
Traditional electronic XOR
Novel optical approach
Novel optical approach
Dual-code transmitter
OptEnc “D1 ”
OptEnc “D2 ”
Fast 2×2 Switch
Fast 2×2 Switch
RF Data In
Σ
OptDec “D1 ”
OptDec “D2 ”
d1 d1 d1 d1
d2 d2 d2 d2
RF Data Out
PD
Switch Control: “1” bar “0” cross
Decryption
Optical implementation of One
Optical implementation of One
-
-
time Pad
time Pad
Optical bus
Dual-code receiver
Fast 2×2 Switch Key
Encryption
Data modulation
Optical
Experimental demonstration
Experimental demonstration
d2 : 1 3 5
Used codes
D D 1 × 4 D D 1 × 4 RF Key in λ2 λ0 λ1 λ2 λ0 λ1OptEnc “d1 ”
OptEnc “d2 ”
Control:
“1” bar “0” cross
RF Data In
Fast 2×2 Switch D D D D Fast 2×2 Switch PLC PLC PLC PLC PLC PLC Optical Encoders
λ1 λ0 λ2
λ1 λ0 λ2
OptEnc “C1 ” output
800 ps
OptEnc “C2 ” output
800 ps
d1 : 1 2 3
Optical
Out
Rx-C
D D 1 × 4 1 × 4 TFF2 TFF0 TFF1 D D 1 × 4 1 × 4 TFF2 TFF0 TFF1 Fast 2×2 Switchkey = 27 -1 PRBS sequence
PD 2x1
Optical Decoder “C1 ”
PLC
λ1 λ0 λ2
Optical Decoder “C2 ”
D D 1 × 4 1 × 4 TFF2 TFF0 TFF1 PD
Optical Decoder “C1 ”
Rx-Eve
Out
RF
RF
Code C
2Encrypted Encrypted Input Input
correlation:
Decoded
bit “1”
Switch Control:“1” bar “0” cross
Operation of secure
Operation of secure
channel
channel
and eavesdropper
and eavesdropper
Output of Rx-EVE
Output of Rx-D
With Key ON, an
When D does no code swap: One-time Pad OFF
–
EVE can eavesdrop with good performance (red diamonds)
When D does code swap: One-time Pad ON
–
EVE cannot eavesdrop, data cannot be received,
no BER can be obtained
With code swapping no BER
could be obtained on
eavesdropper channel
-26 -24 -22 -20 -18 -16 10-14
10-12 10-10 10-8 10-6 10-4 10-2
BER (log)
Rx D (27-1 swap) TxD only
Rx D (27-1 swap), all Txs Rx EVE (no swap) TxD only
Path to miniaturization
Tx-D D D 1 × 4 D D 1 × 4 Swapping Pattern λ2 λ0 λ1 λ2 λ0 λ1 “Dc” OptEnc “Dd” OptEnc Control: “1” bar “0” cross Data Fast 2×2 Switch D D 1 × 4 D D 1 × 4 Fast 2×2 Switch PLC PLC PLC PLC PLC PLC Optical Encoder Tx-D Tx-D 99:1 FC/UPC FC/UPC FC/UPC Tap D Tx-D D D 1 × 4 D D 1 × 4 Swapping Pattern λ2 λ0 λ1 λ2 λ0 λ1 “Dc” OptEnc “Dd” OptEnc Control: “1” bar “0” cross Data Fast 2×2 Switch D D 1 × 4 D D 1 × 4 Fast 2×2 Switch PLC PLC PLC PLC PLC PLC PLC PLC PLC PLC PLC PLC Optical Encoder Tx-D Tx-D 99:1 FC/UPC FC/UPC FC/UPC Tap D Rx-D D D 1 × 4 1 × 4 TFF2 TFF0 TFF1 D D 1 × 4 1 × 4 TFF2 TFF0 TFF1 Fast
2×2 Switch Swapping Pattern PD 2x1 “Dc” OptDec “Dd” OptDec Control: “1” bar “0” cross PLC FC/UPC FC/UPC FC/UPC Tap Tap Rx-D D D 1 × 4 1 × 4 TFF2 TFF0 TFF1 D D 1 × 4 1 × 4 TFF2 TFF0 TFF1 Fast
2×2 Switch Swapping Pattern PD 2x1 “Dc” OptDec “Dd” OptDec Control: “1” bar “0” cross PLC PLC FC/UPC FC/UPC FC/UPC Tap Tap
Thin film filters (TFFs) in combination
with fiber delay lines
Holographic Bragg Reflectors
(HBRs)
Optical Decoder
Path to miniaturization through integration
Top View of Integrated Encoder
HBR
HBR
-
-
based Dual code Encoder/
based Dual code Encoder/
Decoder
Decoder
Single device can process two 2D codes simultaneously including wavelength selection
0
λ1 λ2 λ3
λ1 λ2 λ3
500 ps
code0 (1,2,3)
code1 (1,3,5)
Princeton
M
M
-
-
ary concept
ary concept
•
Motivation:
–
Higher spectral efficiency
M-ary sends multiple bits of information per one symbol transmitted
•
How?
–
Uses pulse positioning (PPM)
–
Needs all-optical method for symbols decoding at high data rates
•
We implemented M-ary with 4 levels
–
Hardware can operate at a lower rate
•
We implemented M-ary modulation with 4 levels
– Increases number chips in code sequence by converting 10Gbps data to M-ary at 5G (each M-ary symbol contains 2 bits of information)
τ0
τ0 + 100ps τ
1
τ1 + 50ps
Switch 1 Switch 2 Electronic board From SC source 100ps 150ps 50ps 0ps delay (1 1) (1 0) (0 1) (0 0)
(C1 C2)
11 10 01 00 Data 2D D S3 S2 S1 S0 symbol τ0
τ0 + 100ps τ
1
τ1 + 50ps
Switch 1 Switch 2 Electronic board From SC source 100ps 150ps 50ps 0ps delay (1 1) (1 0) (0 1) (0 0)
(C1 C2)
11 10 01 00 Data 2D D S3 S2 S1 S0 symbol
PPM M-ary encoding architecture and symbol correspondence
C-S. Brès et al, “Novel M-ary Architecture for Optical CDMA using Pulse Position Modulation,” IEEE LEOS Sydney, Australia, 2005, paper ThBB1
2T ps (frame period) T ps
(bit period)
2 bits IN at x Gb/s 1 pulse OUT
Electrical domain Optical domain 0 0
0 1
1 0
1 1
00 01 10 11 2T ps (frame period) T ps (bit period)
1 pulse OUT at x/2 Gb/s
Electrical domain Optical domain 0 0
0 1
1 0
1 1
00
10
11
Concept of M
BERT
All
All
-
-
optical M
optical M
-
-
ary Decoder
ary Decoder
TOAD 1 TOAD 2 TOAD 3 1 x 4 Control pulse
00
01
10
11
200ps 100ps 1 x 2 1 x 4 5 Gsym/s 10 Gb/s PD 1 x 4 1 x 2 1 x 2TOAD receivers
Experimental illustration of parallel TOAD gating S0 S1 S3 S2
S1
S3
S2
0 1
1 0
1 1
50 ps
Split, delay, combine Delay and position In last half of frame
Delay and position In first half of frame
From TOAD 1 From TOAD 2 From TOAD 3 OUTPUT INPUT 100 ps Autocorrelation Peaks 11 10 01 00 S2 S3 S1 S0 S1 S3
0 1
1 0
50 ps 100 ps S1 11 10 01 00 S1 S0
•
Pulse
is present in one and
only one PPM slot:
– Position pulse back within the 100ps bit according to the M- ary code received
– “00” symbol can be used for clock recovery
Experimental Results
Experimental Results
•
Encoded M-ary data:
– 1 pulse per symbol ( 00, 01, 10 or 11) within 200ps
•
Decoded M-ary data:
– Original 10Gb/s data recovered
– Pulses equally spaced every 100ps
Decoded M-ary data
200ps [01] [11] [11] [10][11] [00] [00] [00]
200ps
0 1 1 1 1 1 1 0 1 1 0 0 0 0 0 0 Encoded M-ary data
•
Eye patterns:
–
Random superposition of the 4 PPM slots:
4 eyes within 200ps, 50ps apart
–
Recovered pattern: 10Gb/s eye
Eye before decoding Eye after decoding
Symbol interval
100ps
Bit interval
[01] [00] [10] [11] 50ps
Additional All
Additional All
-
-
optical signal processing is added
optical signal processing is added
Before OCDMA decoder After OCDMA decoder After TOAD time gating
Experimental BER for Tx3 for 1 to 8 user case Comparison of theory and experiment
Autocorrelation
Cross-correlation
Autocorrelation
Total Received Power (dBm)
1E-11 1E-10 1E-09 1E-08 1E-07 1E-06 1E-05 1E-04 1E-03
-20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0
BER (log)
1 users 4 users 6 users 8 users
10-20 10-15 10-10 10-5
Number of simultaneous users K
Probability of error (log)
0 20 40 60 80 100
50 70 90 110 130 150 10-13 10-12 10-11 10-10 10-9 10-8 10-7
Number of Simultaneous Users K
P ro b a b ility o f e rro r ( lo g )
Scaling using M
Scaling using M
-
-
ary 8 levels
ary 8 levels
110 Users at 10-9
M-ary 8 levels
%
8
.
112
75
13
10
110
=
×
×
=
SE
GHz
GHz
users
SE
1369
=
φ
y
cardinalit
Number of
simultaneous users
@ 10
-9BER = 110
Add/Drop Code Multiplexers for OCDMA Ring Network
Add/Drop Code Multiplexers for OCDMA Ring Network
1. Ring with code add/drop, code
lives only between add / drop
points
•
Avoids interception of data by downstream nodes
•
Code can be reused in separate parts of ring
•
Enables scaling size of ring by code re-use
2. Full interconnection possible without switching
Drop Add Add Drop Add Drop
Drop
Add
Add
Drop Drop
Add
Add
Drop
Drop Add
Code 1 Code 1
Topology of Self-Healing OCDMA Node with Distributed Processing
SW1
Data link in
Data link out Backup link
out
Backup link in
OCDMA Add/Drop
Module SW2
Received Data
Transmitted Data
Monitor/ protection
circuit
Each node monitors the integrity of fiber optic link in both directions.
Protection switches SW1 & SW2 reroute the data if a node/link failure is detected.
Add/Drop Code in OCDMA Ring Network
Demonstration of code removal from OCDMA ring
Demonstration of code removal from OCDMA ring
From the network
Drop
Decoder
All-Optical
Time Gate
Code
Restore
Passing through traffic
Brès et al., IEEE Photon Tech Lett VOL. 17, No. 5, MAY 2005
Implementation of code drop module
Code drop module
Performance - 2 & 4 users
Dropped code for the node
4 simultaneous users
Conclusion
Conclusion
•
OCDMA system improvements through all-optical signal processing
–
2ps time gating was used
• it significantly increased number of simultaneous users
• improved system’s power budget