Transcript Modulation formats for digital fiber transmission

Modulation formats for digital
fiber transmission
Eric Tell 050329
Outline
•
•
•
•
•
Fiber performance limitations
WDM
Optical vs. radio communication
Optical modulators
Modulation formats
– Amplitude shift keying
– Duo-binary signalling
– Optical single sideband signalling
• Simulation/experimental results
• Summary
Fiber performance limitations
• Fiber Loss
• Chromatic dispersion
– different refractive index for different
wavelengths
• Fiber non-linearities
Chromatic dispersion
• Distance limit ~1/(bit rate)²
– Example: Single mode fiber @1550nm
• chromatic dispersion: 17ps/km-nm
• dispersion limited distance: ~100km @10Gbit/s
• comparable to loss limit
• EDFA => increased loss-limited distance
– Chromatic dispersion becomes the limiting factor in
single mode long-haul fibers!
• We want to decrease the bandwidth for a given
datarate!
Wave Division Multiplexing
•
•
Decreased channel spacing leads to
interchannel interference and makes it
difficult to compensate for fiber
nonlinearities
Narrower subchannels would be nice...
WDM (cont'd)
• In a high capacity link the whole EDFA
spectrum is filled with subchannels
• The bandwidth of each subchannel is
proportional to its bit rate
• Total fiber capacity is given by the spectral
efficency: (bitrate per channel)/(channel
spacing)
WDM (cont'd)
• In a practical case using NRZ a spectral
efficiency of 40% can be reached
Power spectral density of NRZ
WDM (cont'd
• More GB/s per channel does not increase
total bandwith, however
– It results in fewer channels to manage
– Increased channel spacing decreases
some non-linear distortions
• BUT to reach higher spectral efficiency a
format with narrower spectrum for a given
bandwidth is needed (while at the same
time not increasing other impairments)
How can this be achieved?
• M-ary Amplitude Shift Keying (ASK)
• Duo-binary signaling
• Optical Single Sideband (OSSB)
Comparison to radio systems
• Much of the same theory can be applied,
except
– Carrier frequency is different
• 1550 nm => 194 Thz
– The available components are different
• no coherent detection (no PLLs)
– The channel is different
Component imperfections
• Modulators are nonlinear
– difficult to achieve pure AM
• PIN photo detectors responds to optical
power rather than electrical field amplitude
(“square envelope”)
• Dispersion introduces a frequency
dependent phase shift
• “intensity-modulated” approaches are
used
Optical Modulators
• Direct modulation
– directly modulate the drive current of a semiconductor
laser
• Absorbtion modulation
– Modulate the absorption spectrum of reverse-biased
diod placed in front of the laser
– Faster and more linear than direct modulation (60 GHz)
• The Mach-Zender (MZ) modulator
– modulation my adding phase shifted signals
Optical modulators (cont'd)
• Direct modulators and absorption modulators
directly modulates the optical power, but will
also generate phase modulation
• The MZ modulator is more flexible and can
generate different kinds if modulation other
than NRZ/RZ/ASK
The MZ modulator
waveguide
contacts
V1(t)
LiNbO3
Ein/2
Ein
Eout
γEin/2
V2(t)
Eout 



Ein j1 (t )
E
e
 e j 2 (t )  in e jv1 (t ) / V  e jv2 ( t ) / V
2
2

MZ modulator transfer function

Eout (t ) 1 jv1 (t ) /V
(v1 (t ), v2 (t )) 
 e
 e jv2 (t ) /V
Ein
2

With γ=1 this can be rewritten as:
(v1 (t ), v2 (t ))  cos(

2V

(v1 (t )  v2 (t )) e j ( v1 (t )v2 (t )) / 2V
Amplitude modulation

Phase modulation (chirp)
With v1(t)=-v2(t) we remove the phase modulation and get:

Eout  Ein cos( v1 (t ))
V
Pout  Pin cos (
2

V
v1 (t ))
MZ modulator biasing
v2 (t )  Vdc  v1 (t )
(v1 (t ),Vdc )  cos(
“Normal bias”:

2V

(2v1(t )  Vdc ) e jVdc / 2V

Vdc  V / 2
  v1 (t ) 1  1  

 (a(t ),V / 2)  cos 
  
1  v1 (t )
4 
2  V

  V
“Bias at extinction”:
Vdc  V
  v1 (t ) 1 



 (a(t ),V )  cos 
    sin  v1 (t )    v1 (t )
2 
V
 V

  V
MZ modulators - observations
• These modulators are only linear in a small
region
– A problem for other than RZ/NRZ signaling
• There must normally be an unmodulated
carrier in order to use non-coherent
detection
M-ASK
• Less bandwidth
levels bandwidth • More power needed for a given
BER
2
±B
• non-linearities become limiting in
4
±B/2
long-haul DWDM systems
8
±B/3
• More complicated (analog and
16
±B/4
digital) electrical circuits
• Possibly useful in multi-mode
32
±B/5
dispersion limited systems e.g. 10
64
±B/6
Gbit/s Ethernet
Duo-binary signaling
• Introduce correlation between consecutive
symbols
• A special case of partial response signaling:
Duo-binary signaling
• Add consecutive
symbols => three
signal levels
-1,1,1,-1
-2,0,2,0
MZ modulator
AM-PSK Duo-binary
• Problem: Normally impractical to handle
three levels
• Solution: Use 0,E,-E
– The detector will detect two levels 0 and E²
– By precoding these two levels will correspond to
0 and 1
– a.k.a Amplitude Modulated Phase Shift Keying
(AM-PSK) duo-binary signaling
0,0,1,0,1
AM-PSK duo-binary system
1,-1,-1,1,1
1
map
xor
1,1,0,1,0
Precoder
0,1,1,0,0
0,0,-2,0,2
0,0,-E,0,E
MZ modulator
biased at extinction
0,0,E2,0,E2
|x|2
Photo detector
(fiber)
Optical Single Sideband (OSSB)
• Observation: The frequency spectrum is
symmetrical
• Implication: Half of it can be filtered out to
save bandwidth => Single Sideband
Transmission!
• Used e.g. in TV
Subcarrier OSSB
• In conventional subcarrier modulation the
subcarrier appears on both sides of the
optical carrier
• Dispersion causes a phase shift between the
two signals, which depends on the distance
• At certain points the entire signal is canceled
out!
Subcarrier OSSB (cont'd)
(decided to skip the equations: Optical fiber communications IVB, eq.16.30-16.36)
Creating an SSB signal
• Two ways
– Use a filter (half the energy is lost)
– Use the Hilbert transform
• known as a Hartley modulator
Hartley modulator
SSB signal:
sSSB (t )  as (t ) cos(2f ct )  aˆ (t ) sin(2f ct )
Baseband signal:
sSSB (t )  as (t )  jaˆ (t )
Optical SSB modulator
“Approximation” of SSB signal:
Hilbert
transform
a(t)
Optical carrier
g (t )  a(t )e jaˆ (t )
MZ
Amplitude
modulator
â(t)
Phase
modulator
OSSB signal
Simulation results: ASK/duo-binary
Dispersion induced receiver sensitivity degradation for Gbit/s signalling
More practical issues…
• ASK
– Nees more power =>
non-linearities limiting
• Duo-binary
– Needs extra filtering
– Optical dispersion
compensation could be
an alternative
– 225 km @10Gbit/s
1550 nm has been
reached
Experimental results: OSSB
Experimental receiver sensitivity degradation vs. fiber length @ 10Gbit/s, BER=10-9
DWDM
• “Normal” NRZ
– 40% spectral efficiency over 150 km
• Duo-binary AM-PSK
– 100% over 100 km
• OSSB
– 66% over 300 km
Summary
• Distance between repeaters is limited by
either of
– Fiber loss
– Chromatic dispersion
– Fiber non-linearities
• With the advent of EDFA chromatic
dispersion has become the limiting factor in
long-haul systems
Summary (cont’d)
• We want to limit the bandwidth in order too
– Reduce the effects of chromatic dispersion
– Reach higher spectral efficiency in DWDM systems
• Two potential methods:
– Duo-binary signaling
– Optical single sideband
• Both methods could potentially halve the bandwidth
• None of the methods are currently used in commercial
systems, but there are some promising experimental results