Ultra-Fast Wavelength-Hopping Optical CDMA

1
Ultra-Fast Wavelength-Hopping Optical CDMA
Principal Investigator Eli Yablonovitch
Co-PIs Prof. Rick Wesel, Prof. Bahram Jalali,
Prof. Ming Wu Electrical Engineering Department,
University of California, Los Angeles CA 90095
Objectives
Approach

• To create an Optical Code Division Multiplexing
  system that
• That is more secure than a WDM optical
  communications system using conventional time
  domain codes.
• That suffers little or no capacity degradation
  compared to a WDM system.
• That is ultimately scalable to 100 simultaneous
  users running at 10Gbits/sec each.
• That is usable for both free-space optical as
  well as fibers
• That will be reasonably close in hardware cost
  compared to a WDM system.

We will encode individual bits in a
wavelength-time matrix, that is programmed by
provably secure algorithms, and that hops with
every bit period.



Accomplishments

• Have distinguished the advantages and
  dis-advantages between direct sequence spread
  spectrum
• and frequency hopping.
• Designed a system for secure wavelength hopping
  OCDMA.

2
Amplitude Modulation
t
carrier
freq
w
wDw
w-Dw
3
Coherent Communication (homodyne detection)
signal(t)
signal(t)cosw t
signal(t)úcosw tú2
local oscillator
carrier wave
Receiver
Transmitter
4
Direct Sequence Spread Spectrum
Transmitter
signal(t)
carrier wave
noisy carrier
cosw t code(t) signal(t)
code(t)
Receiver
cosw t2 code(t)2 signal(t)
local oscillator cosw t
local code generator
code(t)
5
d(t)
sds(t)
direct sequence encoding
c(t)
Figure 2a
c(t)
Tc
1
direct sequence PN code
0
t
d(t)
1
Tb
data
0
t
sds(t)
1
PN ? data
0
t
Figure 2b
6
Frequency Hopping Spread Spectrum
1. Time Division - TDMA 2. Wavelength Division -
WDMA 3. Code Division - CDMA
2 2 2 2 2 2 2 2 2 2
1
2
1
1 1 1 1 1 1 1 1 1 1 1 1
2
1
1
2
1
1
1
1
1
1
1
1
Wavelength
2
1
1
Wavelength
Wavelength
1
2
1
2
2
2
2
2
2
2
2
2
2
2
Time
Time
Time
TDM
WDM
OCDMA
7
CDMA or Spread-Spectrum

• Seemingly wasteful of bandwidth

Orthogonality Condition
Codes are orthogonal, N channels N
codes Channel capacity is unchanged!

• Secret
• Covert
• Jamming resistant

• Multi-path or speckle resistant
• Self-managed network - users
• pick codes at random

8

• Direct Sequence (homogeneous broadening)
• Frequency Hopping (inhomogeneous)
• first patented by Hedy Lamarr (actress) in
  1941
• Used by Secret Service (U.S.)
• Military radios
• Cellular telephones
• Subtle optimization competition between TDMA
  and CDMA
• About 50 of US cellphones use CDMA, including
  particularly the Sprint PCS network.
• World-Wide Generation 3.0 Cellphone standard will
  be CDMA.

9
Dispersion-Limited Signal Propagation Distance
TDM
CDM
WDM
h dispersion coefficient M of channels
(length of code)
10
The basic idea wavelength-time matrix
2 2 2 2 2 2 2 2 2 2
1
2
1
1 1 1 1 1 1 1 1 1 1 1 1
2
1
1
2
1
1
1
1
1
1
1
1
Wavelength
2
1
1
Wavelength
Wavelength
1
2
1
2
2
2
2
2
2
2
2
2
2
2
Time
Time
Time
TDM
WDM
OCDMA
Legend
? user1,
? user2
2
1
11
Generate hopping patterns

1. Bob chooses secret primes p and q and
computes n pq. 2. Bob chooses integer
e which is prime to (p-1)(q-1). 3. Bob computes
d with de mod (p-1)(q-1) ? 1. 4. Bob makes n
and e public, and keeps p, q, d secret. 5.
Alice encrypts m as c ? me mod n, and sends c to
Bob over a public channel. 6. Bob
decrypts by computing m ? cd mod n. 7. Both Bob
and Alice use m as a seed and feed it in to
Advanced Encryption Standard (AES) encoder to
generate a string of random numbers. 8.
That string is fed back into to AES encoder to
generate a 2nd string, etc., etc. 9. Both
Bob and Alice use the string of random
numbers to fill the wavelength-time matrix, using
modular arithmetic. 10. Bob and Alice
generate the hopping patterns according to
the wavelength-time matrix, using a
different modular arithmetic
RSA public key algorithm
Sk
Seed
AES encoder
Sk-1
12
Fill the wavelength-time matrix Random numbers
232 192 108 173 182 69 178 228 185 156 141 96 186
37 157 168 55 106 148 201 181 35 143 8 164 228
220 134 221 104 27 137 192 23 235 110 36 16 192
4 50 56 201 107 181 6 128 249 146 241 104 136 58
183 208 42 99 60 193 30 101 111 252 128
yk N mod (64-k), k 0, 1, 63
192 mod 63 3
1
7
32
4
1
59
44
6
40
24
20
43
14
5
16
28
31
23
18
39
17
42
49
8
46
36
48
58
38
15
22
61
45
Wavelength
Wavelength
19
41
56
60
12
37
57
10
0
0
50
27
51
55
47
13
33
2
3
2
35
30
26
11
53
3
9
21
34
25
62
54
52
63
29
Time
Time
232 mod 64 40
173 mod 61 51
The pattern never repeats
108 mod 62 46
Then randomly fill the next matrix using a
continuation of the random string.
13
Define users from wavelength-time matrix
User k numbers with N mod 8 k-1, k 1, 2, , 8
User1
User2
32
40

1
24
16
8
17
49
48
Wavelength
Wavelength
56
41
57
33
0
9
25
Time
Time
14
High level of security in the case with only one
user
15
Overall system design using electronic switches
Data 1
l1
Modulator
Space Division Switch small buffer
Data 2
l2
Fiber
Modulator
41
Transmitter
Data 3
l3
Modulator
Data 4
l4
Modulator
Data 1
l1
Detector
Space Division Switch small buffer
Data 2
l2
Detector
Receiver
14
Data 3
l3
Detector
Data 4
l4
Detector
Hopping
pattern
16
The first milepost demo of 4x2.5Gbps transmitter
155MHz
16X16 Switch
2.5Gbps
Data 2.5Gbps
116
161
User 1
l1
Modulator
Data
16X16 Switch
116
161
User 2
l2
Modulator
Fiber
Data
16X16 Switch
41
116
161
User 3
l3
Modulator
16X16 Switch
Data
116
161
User 4
l4
Modulator
Hopping
116
161
de-Serializer
Serializer
pattern
17
The first milepost demo of 4x2.5Gbps receiver
155MHz
Data
16X16 Switch
l1
116
161
Detector
User 1
Data
16X16 Switch
116
161
Detector
User 2
l2
Fiber
14
l3
Data
16X16 Switch
116
161
Detector
User 3
16X16 Switch
Data
l4
116
161
Detector
User 4
Hopping
pattern
116
161
de-Serializer
Serializer
18
Switching Fabrics
In general, the implementation of an NXN switch
need NlogN 2X2 switches. For an NXN rearrangeable
permutation switch, the number of 2X2 swithes is
at least log(N!), which is approximately equal to
NlogN?Nlog(2?N)/2. For N16, log(N!) 44.2.
Network implementing 16X16 using 56 2X2 switches.
19
Overall system design using LiNbO3 optical
switches
l1
l0
14
41
Modulator
OE
EO
Data 1
16X16 LiNbO3 Space Division Switch
Fiber
l2
l0
14
41
Modulator
OE
EO
Data 2
41
l3
l0
14
41
Modulator
OE
EO
Data 3
l4
l0
14
41
Modulator
OE
EO
Data 4
l1
Data 1
14
41
OE
EO
OE
16X16 LiNbO3 Space Division Switch
Data 2
l2
14
41
OE
EO
OE
14
l3
Data 3
14
41
OE
EO
OE
l4
Data 4
14
41
OE
EO
OE
Hopping
pattern
Bit interleaving time division multiplexer
Bit time division demultiplexer
14
41
20
Availability of components 2X2 switch
VSC8302.5Gbits/sec Dual 2x2Crosspoint Switch

• Features
• Up to 2.5GHz Clock, 2.5Gb/s NRZ Data Bandwidth
• Output Jitter lt40ps Peak-to-Peak
• Output Skew lt50ps
• Single 3.3V Power Supply
• Industry Standard 44 Pin PQFP Packaging
• Switch configuration time lt 1ns

21
Availability of components de-Serializer and
Serializer
VSC8164 116 de-Serializer
Features 2.5Gb/s Operation 3.3V Single Supply
Operation
VSC8163 161 Serializer
Features 2.5Gb/s Operation 3.3V Single Supply
Operation
22
(No Transcript)
23

• Since the wavelength-hopping occurs in the time
  domain, the initial implementation requires only
  time-encoded WDM hardware.
• Four OCDMA channels at 10 Gbit/sec requires only
  4 WDM channels, that can be implemented in Coarse
  WDM hardware, time encoded by a Silicon chip.
• 100 simultaneous OCDMA users (out of 1000
  subscribers) can be implemented at the expense of
  more WDM hardware, and would require Dense WDM.
• Component count can be reduced, and spectral
  efficiency increased, by using chirped sources
  and time gating in Silicon to fill-in the
  spectral guard bands

Receiver Transmitter
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28

• Fast wavelength-hopping OCDMA
• is compatible with conventional WDM components,
• allowing early technology demonstrations.
• Rapid Summary of Mile-Posts
• demonstration of the wavelength?time concept
  using discrete conventional off-the-shelf WDM
  components.
• 2 users _at_ 2.5Gbit/sec is expected to lead
  rapidly to 4 users _at_ 10Gbit/sec, using
  conventional components.
• 100 simultaneous users, out of 1000 subscribers
  should be feasible, but would require a large
  number of Dense WDM components.
• Time chirped hardware would lead to more
  efficient use of components, and more efficient
  spectral packing of the optical channels, and the
  inter-channel spaces.
• Critical Milestones, (Go/No Go decision) at 15
  months
• 1. Deliver Fiber System of Two OCDMA users _at_
  2.5Gbit/sec.
• 2. Validate Si-Ge time gating chip design for gt4
  users at higher speed.

29
Progression of FWH-OCDMA capabilities as a
function of hardware progress Initial
Demonstrations Using Conventional WDM
components 1. Four OCDMA users _at_ 2.5Gbit/sec.
(15 month deliverable) 2. Ultimately Ten OCDMA
users _at_ 10Gbit/sec. All components are
off-the-shelf, except for fast time-gating logic
that implements the hopping code in Si-Ge logic
technology. Later Demonstrations Using Chirped
WDM hardware 1. In this later phase, we will
demonstrate an Optical-CDMA transmitter with four
wavelengths and four parallel electro-absorption
modulators, duplicating the coarse WDM result
four OCDMA users _at_ 10Gbit/sec. 2. Increase the
number of parallel optical channels, that will
require large numbers of modulators and
photo-detectors on-chip.
30
hop pattern generator
c(t)
sds(t)
wavelength select
d(t)
...
l1
l1 lN
l2
WDM DEMUX
WDM MUX
transmitted signal
...
lN
Multi-wavelength source
Figure 6
31
wavelength select
hop pattern generator
c(t)
...
3-dB splitter
li0 or li1
DEMUX
MUX
...
(A)
output data
li1 or li0
threshold device
DEMUX
MUX
...
...
wavelength select
hop pattern generator
c(t)
Figure 7
32
(No Transcript)