Transcript Optical Wireless Communication using Digital Pulse Interval

1
Free Space Optical
Communications
Professor Z GHASSEMLOOY
Associate Dean for Research
Optical Communications Research Group,
School of Computing, Engineering and Information Sciences
The University of Northumbria
Newcastle, U.K.
http://soe.unn.ac.uk/ocr/
Iran 2008
2
Northumbria University at Newcastle, UK
2
Iran 2008
3
Outline






Introduction
Why the need for optical wireless?
FSO
FSO - Issues
Some results
Final remarks
3
Iran 2008
OCRG - Research Areas
Optical Communications
Wired
Optical Fibre
Communications
• Chromatic dispersion
compensation using
optical signal processing
• Pulse Modulations
• Optical buffers
• Optical CDMA
Wireless
Photonic
Switching
Indoor
• Pulse Modulations
• Equalisation
• Error control coding
• Artificial neural network &
Wavelet based receivers
• Fast switches
• All optical routers
Free-Space
Optics
(FSO)
 Subcarrier modulation
 Spatial diversity
 Artificial neural
network/Wavelet
based receivers
4
HK Poly-Univ. 2007
OCRG - People
Staff
• Prof. Z Ghassemlooy
• J Allen
• R Binns
• K Busawon
• Wai Pang Ng
Visiting Academics
• Prof. Jean Pierre, Barbot
France
• Prof. I. Darwazeh
UCL
• Prof. Heinz Döring
Hochschule Mittweida Univ.
of Applied Scie. (Germany)
• Dr. E. Leitgeb
Graz Univ. of Techn. (Austria)
PhD
•
•
•
•
•
•
•
•
•
•
•
M. Amiri
M. F. Chiang:
S. K. Hashemi
R. Kharel
W. Loedhammacakra
V. Nwanafio
E. K. Ogah
W. O. Popoola
S. Rajbhandari (With IMLab)
Shalaby
S. Y Lebbe
MSc and BEng
•
•
•
•
A Burton
• D Bell
G Aggarwal • M Ljaz
O Anozie
• W Leong
(BEng)
S Satkunam (BEng)
6
Photonics - Applications
• Photonics in communications: expanding and scaling
Long-Haul
Metropolitan
Home access
Board -> Inter-Chip -> Intra-Chip
• Photonics: diffusing into other application sectors
Health
(“bio-photonics”)
Environment
sensing
Security
imaging
Iran 2008
RF
Radio on
Fibre
Traditional
Radio
Lightwave
Source
RF & Optical Communications Integration
Traditional
Optics
Optical
Wireless
Fibre
Free Space
Transmission Channel
Free Space Optical
(FSO)
Communications
The Problem?
AND THAT IS ?
….. BANDWIDTH when and where required.
Over the last 20 years deployment of optical fibre cables in the backbone
and metro networks have made huge bandwidth readily available to
within one mile of businesses/home in most places.
But, HUGE BANDWIDTH IS STILL NOT AVAILABLE TO THE END
USERS.
9
10
Optical Wireless Communication
Abundance of unregulated bandwidth - 200 THz in the 700-1500
nm range
No multipath fading - Intensity modulation and direct detection
What
does
It
Offer
?
High data rate – In particular line of sight (in and out
doors)
Improved wavelength reuse capability
Flexibility in installation
Secure transmission
Flexibility - Deployment in a wide variety of network architectures.
Installation on roof to roof, window to window, window to roof or
wall to wall.
10
Iran 2008
11
Optical Wireless Communication
Multipath induced dispersion (non-line of sight, indoor) - Limiting data rate
D
r
a
w
b
a
c
k
s
SNR can vary significantly with the distance and the ambient noise
(Note SNR  Pr2)
Limited transmitted power - Eye safety (indoor)
High transmitted power - Outdoor
Receiver sensitivity
May be high cost - Compared with RF
Large area photo-detectors - Limits the bandwidth
Limited range: Indoor: ambient noise is the dominant (20-30 dB larger
than the signal level . Outdoor: Fog and other factors
11
Iran 2008
12
Access Network bottleneck
(Source: NTT)
12
12
Iran 2008
13
Access Network Technology
xDSL
 Copper based (limited bandwidth)- Phone and data combine
 Availability, quality and data rate depend on proximity to service provider’s
C.O.
Radio link
 Spectrum congestion (license needed to reduce interference)
 Security worries (Encryption?)
 Lower bandwidth than optical bandwidth
 At higher frequency where very high data rate are possible, atmospheric
attenuation(rain)/absorption(Oxygen gas) limits link to ~1km
Cable
 Shared network resulting in quality and security issues.
 Low data rate during peak times
FTTx
 Expensive
 Right of way required - time consuming
 Might contain copper still etc
13
Iran 2008
14
Optical Wireless Communications
 Using optical radiation to communicate
between two points through unguided
channels
 Types
- Indoor
- Outdoor (Free Space Optics)
14
Iran 2008
FSO - Basics
SIGNAL
PROCESSING
Cloud
Rain
Smoke
Gases
Temperature variations
Fog and aerosol
PHOTO
DETECTOR
DRIVER
CIRCUIT






Transmission of optical radiation through the atmosphere obeys the BeerLamberts’s law:
2
d2
L / 10
Pr  Pt  2

10
d1 ( D  L) 2
Dominant term at
99.9% availability
α : Attenuation coefficient dB/km – Not controllable and is roughly independent
of wavelength in heavy attenuation conditions.
d1 and d2: Transmit and receive aperture diameters (m)
D: Beam divergence (mrad)(1/e for Gaussian beams; FWHA for flat top beams),
This equation fundamentally ties FSO to the atmospheric weather conditions
Link Range L
15
FSO Link
 Transmitter
 Lasers 780,850,980,1550nm, also 10 microns
 Beam control optics
o Multiple transmit apertures to reduce scintillation problems
o Tracking systems to allow narrow beams and reduced geometric losses
 Receiver




Collection lens
Solar radiation filters (often several)
Photodetector - Large area and low capacitance (PIN/APD)
Amplifier and receiver
o Wide dynamic range requirement due to very high clear air link margin
o Automatic gain and transmitter power control
17
Optical Components – Light Source
Operating
Wavelength
(nm)
Laser type
Remark
~850
VCSEL
Cheap, very available, no active
cooling, reliable up to ~10Gbps,
~1300/~1550
Fabry-Perot/DFB
Long life, compatible with EDFA, up to
40Gbps
50–65 times as much power compared
with 780-850 nm
~10,000
Quantum cascade Expensive, very fast and highly
laser (QCL)
sensitive
Ideal for indoor (no penetration through
window)
For indoor applications LEDs are also used
Eye safety -17 Class 1M
Iran 2008
18
Optical Components – Detectors
Material/Structure
Wavelength
(nm)
Responsivity
Typical
(A/W)
sensitivity
Gain
Silicon PIN
300 – 1100
0.5
-34dBm@
155Mbps
1
InGaAs PIN
1000 – 1700
0.9
-46dBm@
155Mbps
1
Silicon APD
400 – 1000
77
-52dBm@
155Mbps
150
InGaAs APD
1000 – 1700
9
Quantum –well and
Quatum-dot
(QWIP&QWIP)
10
~10,000
Germanium only detectors are generally not used in FSO because of their high dark current.
18
Iran 2008
Existing System Specifications
 Range: 1-10 km (depend on the data rates)
 Power consumption up to 60 W










15 W @ data rate up to 100 mbps and  =780nm, short range
25 W @ date rate up to 150 Mbps and  = 980nm
60 W @ data rate up to 622 Mbps and  = 780nm
40 W @ data rate up to 1.5 Gbps and  = 780nm
Transmitted power: 14 – 20 dBm
Receiver: PIN (lower data rate), APD (>150 mbps)
Beam width: 4-8 mRad
Interface: coaxial cable, MM Fibre, SM Fibre
Safety Classifications: Class 1 M (IEC)
Weight: up to 10 kg
19
20
Power Spectra of Ambient Light Sources
Normalised power/unit wavelength
1.2
Pave)amb-light >> Pave)signal (Typically 30 dB with no optical filtering)
Sun
1
Incandescent
0.8
1st window IR
0.6
2nd window IR
Fluorescent
0.4
x 10
0.2
0
1.5
1.4
1.3
1.2
Wavelength (m)
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
20
Iran 2008
21
FSO - Characteristics





Narrow low power transmit beam- inherent security
Narrow field-of-view receiver
Similar bandwidth/data rate as optical fibre
No multi-path induced distortion in LOS
Efficient optical noise rejection and a high optical signal
gain
 Suitable to point-to-point communications only (out-door
and in-door)
 Can support mobile users using steering and tracking
capabilities
 Used in the following protocols:
- Ethernet, Fast Ethernet, Gigabit Ethernet, FDDI, ATM
- Optical Carriers (OC)-3, 12, 24, and 48.
 Cheap (cost about $4/Mbps/Month according to fSONA)
21
Iran 2008
22
Cost Comparison
Source:
22
Iran 2008
Existing Systems
 Auto tracking systems - 622 Mbps [Canobeam]
 TereScop - 1.5 Mbps to 1.25 Gbps (500m – 5km)
 Cable Free - 622 Mbps to 1.25 Gbps (High power class 3B
Laser at 100 mW)
 Microcell and cell-site backbone – GSM, GPRS, 3G and EDGE traffic
o No Frequency license
o No Link Engineering
o Management via SNMP, RS232
o or GSM connection
 Last mile
o 155 Mbps STM-1 links
o 622 Mbps ATM link for Banks etc
24
When Did It All Start?
800BC
150BC
1791/92
- Fire beacons (ancient Greeks and Romans)
- Smoke signals (American Indians)
- Semaphore (French)
1880
- Alexander Graham Bell demonstrated the photophone – 1st
FSO (THE GENESIS)
(www.scienceclarified.com)
1960s
- Invention of laser and optical fibre
1970s
- FSO mainly used in secure military applications
1990s to date - Increased research & commercial use due to successful trials
24
Iran 2008
25
FSO - Applications
In addition to bringing huge bandwidth to businesses /homes FSO also finds
applications in :
Hospitals
Others:
 Inter-satellite communication
 Disaster recovery
 Fibre communication back-up Multi-campus university
 Video conferencing
 Links in difficult terrains
 Temporary links
e.g. conferences
FSO challenges…
Cellular communication back-haul
25
Iran 2008
Hybrid FSO/RF Wireless Networks

RF wireless networks
- Broadcast RF networks are not scaleable
- RF cannot provide very high data rates
- RF is not physically secure
- High probability of detection/intercept
- Not badly affected by fog and snow, affected by
rain

A Hybrid FSO/RF Link
- High availability (>99.99%)
- Much higher throughput than RF alone
- For greatest flexibility need unlicensed RF band
27
LOS - Hybrid Systems
Video-conference for Tele-medicine CIMIC-purpose and disaster recovery
27
Iran 2008
28
FSO - Challenges
Major challenges are due to the effects of:
CLOUD,
GASES,
SIGNAL
PROCESSING
SMOKE,
PHOTO
DETECTOR
DRIVER
CIRCUIT
RAIN,
TEMPERATURE VARIATIONS
FOG & AEROSOL
To achieve optimal link performance,
system design involves
tradeoffs of the different parameters.
POINT A
POINT B
28
Iran 2008
29
FSO Challenges - Rain
 = 0.5 – 3 mm
Effects
Photon absorption
Options
Remarks
Increase
transmit Effect not significant
optical power
29
Iran 2008
FSO Challenges - Physical Obstructions
Pointing Stability and Swaying Buildings
Effects
Loss of signal
 Multipath induced
Distortions
 Low power due to
beam divergence and
spreading
 Short term loss of
signal

Solutions
Spatial diversity
 Mesh architectures: using
diverse routes
 Ring topology: User’s n/w
become nodes at least one
hop away from the ring
 Fixed tracking (short
buildings)
 Active tracking (tall buildings)

30
Remarks

May be used for
urban areas,
campus etc.
Low data rate
 Uses feedback

FSO Challenges – Aerosols Gases &
Smoke
Effects
 Mie scattering
 Photon absorption
 Rayleigh scattering
Solutions
 Increase transmit
power
 Diversity techniques
31
Remarks
 Effect not severe
32
FSO Challenges - Fog
 = 0.01 - 0.05 mm
In heavy fog conditions, attenuation is
almost constant with wavelength over the
780–1600 nm region.
In fact, there are no benefits until one gets
to millimeter-wave wavelengths.
Effects
Options
Increase transmit
optical power
 Hybrid FSO/RF

Mie scattering
 Photon absorption

Remarks
Thick fog limits link
range to ~500m
 Safety requirements
limit maximum optical
power

32
Iran 2008
33
FSO Challenges - Fog
Weather
condition
Precipitation
Amount
(mm/hr)
Visibility
Dense fog
Thick fog
dB
Loss/km
Typical Deployment Range
(Laser link ~20dB margin)
0m
50 m
-271.65
122 m
200 m
-59.57
490 m
500 m
-20.99
1087 m
Moderate fog
Snow
Light fog
Snow
Cloudburs
t
100
770 m
1 km
-12.65
-9.26
1565 m
1493 m
Thin fog
Snow
Heavy rain
25
1.9 km
2 km
-4.22
-3.96
3238 m
3369 m
Haze
Snow
Medium
rain
12.5
2.8 km
4 km
-2.58
-1.62
4331 m
5566 m
Light haze
Snow
Light rain
2.5
5.9 km
10 km
-0.96
-0.44
7146 m
9670 m
Clear
Snow
Drizzle
0.25
18.1 km
20 km
-0.24
-0.22
11468 m
11743 m
23 km
50 km
-0.19
-0.06
12112 m
13771 m
Very clear
(H.Willebrand & B.S. Ghuman, 2002.)
33
Iran 2008
FSO Challenges - Beam Divergence
 Beam width
 Typically, for FSO transceiver is relatively wide: 2–10-mrad
divergence, (equivalent to a beam spread of 2–10 m at 1 km), as is
generally the case in non-tracking applications.
 Compensation is required for any platform motion
 By having a beam width and total FOV that is larger than either
transceiver’s anticipated platform motion.
 For automatic pointing and tracking,
 Beam width can be narrowed significantly (typically, 0.05–1.0 mrad
of divergence (equivalent to a beam spread of 5 cm to 1 m at 1 km)
- further improving link margin to combat adverse weather conditions.
- However, the cost for the additional tracking feature can be significant.
35
FSO Challenges - Others
 Background radiation
 LOS requirement
 Laser safety
Iran 2008
 Free Space Optics
 Characteristics
 Challenges
 Turbulence
- Subcarrier intensity multiplexing
- Diversity schemes

Results and discussions

Wavelet ANN Receiver

Final remarks
FSO Challenges - Turbulence
Effects
Irradiance fluctuation
(scintillation)
 Image dancing
 Phase fluctuation
 Beam spreading
 Polarisation
fluctuation

Options
Diversity techniques
 Forward error control
control
 Robust modulation
techniques
 Adaptive optics
 Coherent detection not
used due to Phase
fluctuation

37
Remarks
Significant for long
link range (>1km)
Turbulence and thick
fog do not occur
together
 In IM/DD, it results in
deep irradiance
fades that could last
up to ~1-100 μs

38
FSO Challenges - Turbulence
Cause: Atmospheric inhomogeneity / random temperature variation along beam
path.
The atmosphere behaves like prism
of different sizes and refractive indices
Phase and irradiance
fluctuation
Depends on:
• Zones of differing density act as lenses,
scattering light away from its intended path.
• Thus, multipath.
Result in deep
signal fades that
lasts for ~1-100 μs
 Altitude/Pressure, Wind speed,
 Temperature and relative beam size.
 Can change by more than an order of magnitude during the course of a day, being the
worst, or most scintillated, during midday (highest temperature).
 However, at ranges < 1 km, most FSO systems have enough dynamic range or margin to
compensate for scintillation effects.
Iran 2008
39
Turbulence – Channel Models
Irradiance PDF:
pI (I ) 
 (ln( I / I 0 )   l 2 / 2) 2 
1
exp 

2  l I


2 l 2
1
I 0
Model
Comments
Log Normal
Simple; tractable
Weak regime only
I-K
Weak to strong
turbulence regime
K
Strong regime only
Rayleigh/Negative
Exponential
Gamma-Gamma
Saturation regime only
Based on the modulation process the received
irradiance is
x
y
I  I I
Irradiance PDF by Andrews et al (2001):
 ) / 2
(
2()(
p( I ) 
I
()()
 
) 1
2
   (2 I )
2
 
 
0.49l
  1
  exp 
12 / 5 7 / 6 
  (1  1.11l )  
1
2
 
 
0.51l

  1
  exp 
12 / 5 5 / 6 
(
1

0
.
69

)
l
 
 
1
All regimes
I 0
Ix:
due to large scale effects;
obeys Gamma distribution
Iy:
due to small scale effects;
obeys Gamma distribution
Kn(.): modified Bessel function
of the 2nd kind of order n
σl2 : Log irradiance variance
(turbulence strength indicator)
To mitigate turbulence effect we, employ subcarrier modulation
Iran 2008
with spatial diversity
40
Turbulence Effect on OOK
No Intensity Fading
No Pulse Bit “0”
Threshold level
Pulse Bit “1”
A
A/2
With Intensity Fading
A
All commercially available systems use OOK with fixed threshold
which results in sub-optimal performance in turbulence regimes
40
Iran 2008
Turbulence Effect on OOK
Using optimal maximum a posteriori (MAP) symbol-by-symbol detection with
equiprobable OOK data:

dˆ (t )  arg maxd P(ir / d (t ))
2
2

  ((ir  RI )  ir ) 

exp


 
2
2

0




  ln( I / I )   2 / 2

0
l
exp 
2
2


l


2
1
2 l 2
1
.
I


dI


0.5
Noise variance
0.5*10-2
0.45
10-2
3*10-2
0.4
Threshold level, i
th
5*10-2
0.35
OOK based FSO requires
adaptive threshold to perform
optimally….
0.3
0.25
0.2
….but subcarrier intensity
modulated FSO does not
0.15
0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Log Intensity Standard Deviation
0.8
0.9
1
41
SIM – System Block Diagram
DC bias
m(t)
d(t)
Data in Serial/parallel
converter
.
.
Subcarrier
modulator
.
.
m(t)+bo
Summing
circuit
Optical
transmitter
Atmospheric
channel
ir
d’(t)
.
.
Parallel/serial
Data out
converter
Spatial
diversity
combiner
Subcarrier
demodulator
42
Photodetector
array
Subcarrier Intensity Modulation
 No need for adaptive threshold
 To reduce scintillation effects on SIM
 Convolutional coding with hard-decision Viterbi decoding (J. P. KIm
et al 1997)
 Turbo code with the maximum-likelihood decoding (T. Ohtsuki, 2002)
 Low density parity check (for burst-error medium):
- Outperform the Turbo-product codes.
- LDPC coded SIM in atmospheric turbulence is reported to achieve a
coding gain >20 dB compared with similarly coded OOK (I. B. Djordjevic, et
al 2007)
 SIM with space-time block code with coherent and differential
detection (H. Yamamoto, et al 2003)
 However, error control coding introduces huge processing
delays and efficiency degradation (E. J. Lee et al, 2004)
43
SIM – Our Contributions
Multiple-input-multiple-output (MIMO) (an array of transmitters/
photodetectors) to mitigate scintillation effect in a IM/DD FSO link
 overcomes temporary link blockage (birds and misalignment) when
combined with a wide laser beamwidth, therefore no need for an active
tracking
 provides independent aperture averaging with multiple separate
aperture system, than in a single aperture where the aperture size has
to be far greater than the irradiance spatial coherence distance (few
centimetres)
 provides gain and bit-error performance
 Efficient coherent modulation techniques (BPSK etc.) - bulk of the
signal processing is done in RF that suffers less from scintillation
 In dense fog, MIMO performance drops, therefore alternative
configuration such as hybrid FSO/RF should be considered
 Average transmit power increases with the number of subcarriers,
thus may suffers from signal clipping
 Inter-modulation distortion
45
Subcarrier Modulation - Transmitter
A1
A2
Input
data
d (t )
Serial to
Parallel
Converter
.
.
.
.
.
.
AM
g(t)
PSK modulator
at coswc1t
g(t)
PSK modulator
at coswc2t
m(t ) 
M
 A j g (t ) cos(wcj t   j )
j 1
Σ m(t)
Σ
Laser
driver
Atmopsheric
channel
DC bias
b0
g(t)
PSK modulator
at coswcMt
Modulation index is constrained to avoid over modulation
  [ Rh0 Pt ,00 Nc' ]1
45
Iran 2008
46
Subcarrier Modulation - Transmitter
2
1
0
-1
  [ Rh0 Pt ,00 Nc' ]1
M
m(t )   A j g (t ) cos(wcj t   j )
j 1
-2
5-subcarriers
Output
power
-3
P
-4
-5
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Pmax
2
m(t)
1
b0
Drive current
Iran 2008
SIM - Receiver
SNRele
Pr 
( IRA ) 2

2 2
Nc
PSK Demodulator
x
g(-t)
Sampler
coswc1t
 h P 1  d (t ) cos(2f t n(t )
i 1
i
t ,i
i
i
Photodetector
ir
PSK Demodulator
at coswc2t
.
.
.
Photo-current
PSK Demodulator
at coswcMt
ir (t )  R I (1  m(t ))  n(t )
R = Responsivity, I = Average power,  =
Modulation index, m(t) = Subcarrier signal
di(t) = Data
47
Parallel
to Serial
Converter
dˆ (t )
Output
data
48
Subcarrier Modulation
 Performs optimally without adaptive threshold as in OOK
 Use of efficient coherent modulation techniques (PSK, QAM etc.)
- bulk of the signal processing is done in RF where matured devices like stable,
low phase noise oscillators and selective filters are readily available.
 System capacity/throughput can be increased
 Outperforms OOK in atmospheric turbulence
 Eliminates the use of equalisers in dispersive channels
 Similar schemes already in use on existing networks
But..
 The average transmit power increases as the number of
subcarrier increases or suffers from signal clipping.
 Intermodulation distortion due to multiple subcarrier impairs
its performance
48
Iran 2008
SIM - Spatial Diversity
 Single-input-multiple-output
 Multiple-input-multiple-output (MIMO)
49
SIM - Spatial Diversity
Combiner
F
S
O
i1 (t )
C
H
A
N
N
E
L
i2 (t )
iN (t )
Assuming identical PIN photodetector on each
links, the photocurrent on each link is:
a1
a2
.
.
.
.
a
N

iT (t )
M
R 
iri (t ) 
I i 1   A j g (t ) cos(wcj t   j
N 
j
PSK
dˆ (t )

)   ni (t )


Subcarrier
Demodulator
ai is the scaling
factor
Diversity Combining Techniques
Maximum Ratio
Combining (MRC)
[Complex but optimum]
ai
 ii
Equal Gain
Combining (EGC)
Selection Combining
(SELC). No need for phase
a1  a2  ...  a N
iT (t )  max( i1 (t ), i2 (t )...i N (t ))
50
information
SIM Spatial Diversity – Assumptions
Made
 Spacing between detectors > the transverse correlation
size ρo of the laser radiation, because ρo = a few cm in
atmospheric turbulence
 Beamwidth at the receiver end is sufficiently broad to cover
the entire field of view of all N detectors.
 Scintillation being a random phenomenon that changes
with time makes the received signal intensity time variant
with coherence time o of the order of milliseconds.
 Symbol duration T << o , thus received irradiance is time
invariant over one symbol duration.
51
52
Subcarrier Modulation - Spatial Diversity
One detector
Two detectors
Three detectors
A typical reduction in intensity fluctuation with spatial diversity
Eric Korevaar et. al
Iran 2008
 Free Space Optics
 Characteristics
 Challenges
 Turbulence
- Subcarrier intensity multiplexing
- Diversity schemes

Results and discussions

Wavelet ANN Receiver

Final remarks
Error Performance – No Spatial Diversity
Normalised SNR at BER of 10-6 against the number of subcarriers for various
turbulence levels for BPSK
Normalised SNR @ BER = 10-6 (dB)
20
15
Increasing the number of
subcarrier/users, results
In increased SNR
10
5
0
Log intensity
variance
0.1
0.2
0.5
0.7
-5
-10
1
2
3
4
5
6
7
Number of subcarrier
8
9
10
SNR gain compared
with OOK
55
Error Performance – No Spatial Diversity
BPSK BER against SNR for M-ary-PSK for log intensity variance = 0.52
DPSK
BPSK
16-PSK
8-PSK
-2
10
10
BER
BPSK based subcarrier
modulation is the most
power efficient
Log intensity
-4
variance = 0.52
-6
10
BER 
-8
10


2 
Q SNRe log 2 M sin( / M ) p( I )dI
log 2 M 0
-10
10
20
30
25
SNR
35
40
(dB)
Iran 2008
56
Spatial Diversity Gain
Spatial diversity gain with EGC against Turbulence regime
2 Photodetectors
3 Photodetectors
70
Saturation
Diveristy Gain (dB)
60
50
40
Moderate
30
20
10
Weak
Turbulence Regime
Iran 2008
Spatial Diversity Gain for EGC and SeLC
25
Link margin (dB)
Log Intensity
Variance
0.22
20
0.52
0.72
1
15
Link margin for SelC is lower
than EGC by ~1 to ~6 dB
10
5
0
Dominated by received irradiance,
reduced by factor N on each link.
-5
-10
EGC
Sel.C
BER = 10-6
1
Pe ( SelC) 
2
3
N
2
N
4
5
6
No of Receivers
n
7
[ w 1  erf ( x )


i 1
N 1
i
i
8
.e
9
10
(  K 0 2 exp( 2 xi 2l  l 2 ))
Zeros of the n order w n = Weight factor of the nth order
xi n = Hermite
i
Hermite polynomial
polynomial
th
i 1
i 1
]
K 0  RI0 A 2  2 N
Spatial Diversity Gain for EGC and MRC
30
BER = 10-6
Log Intensity
variance

1  /2
 
0
25
Spatial Diversity Gain (dB)
Pe( EGC) 
1

20
0
1

m
2 u  u )
 wi Q( K1e ( x
i
)
1
MRC
EGC

15
Pe ( MRC ) 
2
0.5



 
 ( I ) dI
Q

/
I
P
MRC
I

0
10
5
0


K12
2
exp  
Z
P ( Z ) d dZ
 2 sin 2 ( )
 Z



0.22
1
2
3
Most diversity gain
region
4
6
5
No of Receivers
7
8
1

 /2
 S ( )
N
d ,
0
10
9
The optimal but complex MRC diversity is marginally superior
to the practical EGC
58
Multiple-Input-Multiple-Output
Combiner
It1
It2
d(t)
BPSK
ModuLator
and
.
.
.
Laser
driver
ItH
F
S
O
i1 (t )
C
H
A
N
N
E
L
i2 (t )
iN (t )
a1

a2
.
.
.
.
a
iT
BPSK
Subcarrier
Demodulator
dˆ (t )
N
By linearly combining the photocurrents using MRC, the individual SNRe on each
link
SNRele i

RA
 
2
 2 N H
59


I

ij 
j 1

H
2
MIMO Performance
-3
10
At BER of 10-6:
1X5MIMO
1X8MIMO
4X4MIMO
2X2MIMO
1X4MIMO
-4
10

2 x 2-MIMO requires additional ~0.5
dB of SNR compared with 4photodetector single transmittermultiple photodetector system.

4 x 4-MIMO requires ~3 dB and ~0.8
dB lower SNR compared with
single transmitter with 4 and 8photodetectors , respectively.
-5
BER
10
-6
10
-7
10
-8
10
-9
10
log intensity variance= 0.52
12
14
16
18
20
22
2
(dB)
SNR (R*E[I]) / No
1
Pe 

/ 2
 S ()
N
d,
24
26
S () 
2


K2

w j exp  
exp[
2
(
x
2



)]

j
u
u 
2
 j 1
 2 sin 

1
m
K2 
0
60
RI 0 A
2 N 2 H
 Free Space Optics
 Characteristics
 Challenges
 Turbulence
- Subcarrier intensity multiplexing
- Diversity schemes

 Results and discussions
Wavelet ANN Receiver

Final remarks
62
Transmission System - Receiver
Models
Data in
TX
Channel
+
Noise
Data out…
Slicer
MMSE
Data out
Slicer
Equaliser
MF
Data out
Slicer
NN
CWT
Wavelet - NN
Iran 2008
63
PPM System – NN Equalization
n(t)
M
0100
M
0010
PPM
Encoder
PPM
Decoder
Xj
Optical
Transmitter
Decision
Device
X(t)
Z(t)
h(t)
Yj
Neural
Network
∑
Optical
Receiver
Zj
Zj-1
Zj
Matched
Filter
.
Zj-n
Ts = M/LRb
.
 A feedforward back propagation neural network .
 ANN is trained using a training sequence at the operating SNR.
 Trained AAN is used for equalization
Iran 2008
64
Impulse Response of Equalized Channel
Impulse response of unequalized
channel
impulse response of equalized
channel
• Pulse are spread to adjust pulse .
• Equalized response in a delta
function which is equivalent to a
impulse response of the ideal
channel
• ISI depends on pulse spread
Iran 2008
65
Results (1)
Slot error rate performance of 8- PPM in diffuse channel with Drms of 5ns at 50
Mbps
Adaptive linear equalizer with
least mean square (LMS)
algorithm is used.

The
performance of ANN
equalizer is almost identical to
the linear equalizer.
Iran 2008
66
Results (2)
Slot error rate performance of 8- PPM in diffuse channel with Drms of 5ns at 100
Mbps
Unequalized performance at
higher data rate is unacceptable at
all SNR range

Linear and neural equalization
give almost identical performance.

Iran 2008
67
Results (3) - Wavelet-AI Receiver
Wavelet
SNR Vs. the RMS delay spread/bit duration
Iran 2008
68
Wavelet-AI Receiver - Advantages and
Disadvantages
 Complexity
- many parameters & computations.
 High sampling rates
- technology limited.
 Speed
- long simulation times on average machines.
 Similar performance to other equalisation techniques.
 Data rate independent
- data rate changes do not affect structure (just re-train).
 Relatively easy to implement with other pulse
modulation techniques.
Iran 2008
Visible Light Optical Wireless System with
OFDM
Visible-light communication system
Distribution of illuminance
Distribution of horizantal illuminance [lx]
Number of LEDs
60 x 60 (4 set)
1400
1200
Illuminance[lx]
Down
link
Up
link
1000
800
600
400
200
5
4
5
3
4
3
2
2
1
y[m]
1
0
0
x[m]
FSO Network – Two Universities in
Newcastle
71
Agilent Photonic Research Lab
Agilent Photonic
Research Lab
Optical Fibre
Research Collaboration
A-Block
Free space optical
Du-plex communication
link
(Northumbria
and Newcastle Universities)
at a data rate of 155 Mbps
Iran 2008
Collaborators
• Graz Technical University, Austria
• Houston University, USA
• University College London, UK
• Hong-Kong Polytechnic University
• Tarbiat Modares University, Iran
• Newcastle University, UK
• Ankara University, Turkey
• Agilent, UK
• Cable Free, UK
• Technological University of Malaysia
• Others
•
73
Final Remarks
 Could the promise of optical wireless live up to reality?
 Yes!!
 But
 Optical wireless must complement radio, not compete
 Industry must be bold in research and development
 Lower component cost, and single technology based
deviced
 Integration with existing systems
 Lover receiver sensitivity
 Of course more research and development at all
levels
Iran 2008
74
Summary

Access bottleneck has been discussed

FSO introduced as a complementary technology

Atmospheric challenges of FSO highlighted

Subcarrier intensity modulated FSO (with and
without spatial diversity) discussed

Wavelet ANN based receivers

74
Iran 2008
75
Acknowledgements
 To many colleagues (national and international)
and in particular to all my MSc and PhD students
(past and present) and post-doctoral research
fellows
Iran 2008