You are on page 1of 41

Transition from

Electronics to
Photonics in
Communication
Network and
Phased ArrayProf. (Dr.)N Das
Radar Emeritus Professor,
NIST,Berhamporepo
Unprecedented demand for network capacity due to
exponential rise of Internet users and new generation services
(VoD , Image Processing, IPTV etc)
This has led to replacement of Coax cables with Fibers for
tremendous inherent capacity. 1st generation OPT network
SONET/SDH provided high speed optical interface to IP
routers. But the switching and network intelligent functions were
handled by electronics.

NETWORK

Data Link
I S
Network
O
P Data Link N
E IP over SONET
Physical T
But the switching burden in 1st generation OPT network has
been laid entirely on electronics.
In every switching node there is a need of O/E or E/O
conversion with buffering electronically too. But electronics cant
keep up when data rate is in order of Gbps. Besides electronic
equipments are not transparent to data rate and protocol.
Technology was needed to switch optical signals without
conversion to electrical form.
Thus a 2nd generation network concept emerged where routing,
switching and intelligent functions of the network could be
migrated to an optical layer.
The client layers make use of light paths provided by the optical
layer.
A light path is a connection between two network nodes that is
set up by assigning a dedicated wavelength to it on its link in its path.
Light paths are simply replacement for hardwired fiber connections
between SONET terminals.
The individual wavelengths are likely to carry data at fairly high
bit rates. (10 Gbps) and this entire BW is provided to the upper
layer by a light path User Applications
Datagrams
VC VC

MPLS
SONET/ SDH
Connection
SONET/ SDH Fiber Channel
Layer IP Layer
Layer

Light Paths

OPTICAL LAYER

(Layered view of 2nd Generation Optical Network Layer supporting varied Client Layers on its
top)

It turns out that switching and routing of high capacity connections is


more economical at the optical layer than in electrical layer.
Optical layer is evolving to provide additional functionality beyond
provision of circuit-switched light paths i.e light paths on demand
and provide redundancy intelligent optical ring and mesh
network.
Benefits of optical layer in an All-Optical network is fully exploited
with technology for Optical
, . , Switches (OLTs), Wavelength cross
1
1
2
2
3
3
4
4

connects and A/D multiplexer involving wavelength based


switching and routing.
1
11 , 21, 31 , 41 2 12 , 21, 31 , 42
3
4
De-multiplexer Multiplexer

12 , 22, 32 , 42 11 , 22, 32 , 41

(Static Wavelength Cross Connect, routing signals from input port to output port based on )

`
Dynamic WXCs can be constructed by combining optical switches
with Multiplexers and De-multiplexers. The switches are used inside
wavelength cross connects to re-configure them to support new light
paths. In this application, the switches are replacements for manual
fiber path panels but with significant added software for end-to-end
network management. It requires 1-10 msec switching time (number
of ports > 1000).

Switch

Fabric
Optical switch requirement for OXCs include
Scalability (Ability to build switches with large port counts that
perform adequately)

High reliability , low loss , good uniformity of optical signals


independent of path length

Less Cross-talk between desired and other paths.

Most of the cross-connects presently use electrical core for


switching (O/E/O). ATM switches and IP routers can be used
to switch data using the individual channels within a WDM link,
but this implies that 100s of switch interfaces must be used to
terminate a single link with large number of channels. O/E/O
switching is not transparent.
The core of an O/O/O cross-connect is an optical switch that
is transparent and the cross-connect therefore is ready for
future data rate upgrades.
Cost, size and complexity is also reduced.
But the network management functions ( performance
monitoring and fault isolation) could not be implemented
entirely in the optical domain. (limitations of optical technology
in bit processing and lack of memory)
O/O/O cross-connect dont allow 3R signal regeneration. So
problem in long-haul transmission.
Opaque cross-connect is a compromise between O/E/O and
O/O/O approaches. Mostly optical at the switching fabric but
still rely on a limited subset of surrounding electronics to
monitor system integrity. Switch core is optical and is not b.w
limited and allows wavelength conversion, signal
regeneration, QoS monitoring all within the cross-connect
switch but is not transparent.
Optical Add/Drop Multiplexers reside in network nodes to
insert or extract optical channels (wavelengths) to or from the
optical transmission stream.
Switches that function as OADMs are wavelength selecting
switches. Using an OADM channels in a WDM signal can be
added or dropped without any electronic processing.
In addition optical switches are used as Protection switching
(1-10 msec, port count 1X2 or 2X2) , extremely reliable
detection the origin and the nature of failure, to notify to other
nodes.
Also used as optical signal monitoring (OSM a network
mgmt. operation). A tapped portion of the aggregated WDM
signal received and separated into individual wavelengths and
monitored channel by channel. (for signal power, wavelength
accuracy and optical power cross-talk). Because the tapped
optical signal is very low in power the optical switch employed
has to have high extinction ratio, low insertion loss and good
uniformity.
With new data routes establishment network provisioning
requires switch to carry out re-configuration request over time
interval of the order of a few minutes. High capacity
reconfigurable switches can respond automatically and
quickly to service request for increasing network flexibility and
thus b.w and profitability.

Optical Packet Switched networks would require larger


switches with switching time of the order of nano-seconds
( 60 Byte pkt at 10Gbps requires 48 nano-seconds).

Switches for external modulation (turn On/Off data in front of


laser source) requires switching time as a small fraction of bit
duration (10Gbps 100 ps) of about 10 ps.
Most solutions for all-optical switching are still under study.
Research on technologies are going on.
Optical Switch Fabrics
Apart from switching time other important parameters of a
switch are
Insertion loss (uniformity for all i/p and o/p connections)
Cross-Talk
Extinction ratio (On-Off switches- should be large)
Polarization dependant loss (Low and same for both states of
polarization) PDL
Scalability
Temperature resistance
Energy usage
Reliability
Existing Optical Switching Technologies
Opto-Mechanical Switches (use prisms, mirrors and directional couplers) exhibit low
insertion loss and low PDL, low Cross-Talk and low fabrication cost. Switching speed few
msec
But lacks scalability, long term reliability (mechanical components). Switch configuration
limited to 1X2 and 2X2 port sizes.
Mainly used in fiber protection and very low port count A/D applications.
MEMS (Micro-Electro-Mechanical-System)
o Use tiny reflective surfaces to redirect the light (mirror control)
o 2D-MEMS (On-Off mirror; simple but
lossy). Path length grows linearly with
growing optical loss. 3.7 dB (8X8),
5.5 dB (16X16), 7 dB (32X32).
High port count >1000 not practical
due to optical loss.
o 3D-MEMS (dedicated Movable mirror
for each i/p & o/p, tilting freely about
two axes). Port count >1000 OXC
switch application, complexity
All Optical MEMS have the following potential benefits
o Scalability
o Low-loss
o Short switching time
o Low power consumption
o Low cross-talk
o Low polarization effect

Other applications for MEMS include


o Wavelength Add/Drop multiplexing
o Optical service monitoring
o Optical protection switching

With challenges in
o Mirror fabrication
o Opto-mechanical packaging
o Mirror control algorithm
o Implementation
Electro Optics Switch
A 2X2 Electro Optics Switch uses a DC whose coupling ratio is
changed by varying the RI (index) of the material (LiNbO3) in
the coupling region.

This switch is capable of changing its state very rapidly (<ns


determined by the capacitance of electrode configuration)
Highly reliable
Larger switches through integration of 2X2 switches on single
substrate
Drawbacks are
High insertion loss
Possible PDL
More expensive than mechanical switches
Thermo Optics Switches (variation of RI of the material with
temperature)

Interferometric (MZI)
Splits the signal into two distinct beams twice, heating one
arm of the interferometer causes the RI (index) to change
and hence the variation of the optical path of that arm is
experienced. Phase difference between the light beams
cause constructive or destructive interference and thus o/p
port is selected.
Digital Optical Switches are integrated optical devices
made of sillica on silicon. The switch is composed of two
interacting waveguide arms through which light propagates.
The phase error between the beams at the two arms
determine the o/p port. An electrode through controlled
electronics provide the heating.

Merits of the above are


Small size as any other thermo optical switch
Allows the integration of variable optical attenuators and AWGs on the
same chip with the same technology
Drawbacks are
High driving power characteristic
Cross-talk may be un-acceptable
Insertion loss may be un-acceptable
Limited integration density (Large die area High power dissipation
requiring forced air cooling )
Liquid-Crystal Optical Switches
Based on the change of polarization state of incident light by a
liquid crystal on application of an electric field over the liquid
crystal. The change of the polarization in combination with
polarization selective beam splitters allows optical space switching.
A 1X2 liquid-crystal optical switch structure is shown below.
Liquid-crystal switches have no moving parts, very reliable and
satisfactory optical performance but is temperature sensitive.

Bubble Switches (Operation based on heating and cooling of


substrate) and Acoustooptic switches (Acoustooptic effect) are
polarization insensitive switches.
If the incoming signal is multi-wavelength, it is even possible to
switch several different wavelengths simultaneously, as it is possible
to have acoustic waves in the material with different frequencies at
the same time. The switching speed of the acoustooptic switches is
limited by the speed of sound and is in the order of microseconds.

Acousto optic switch


SOA (Semiconductor Optical Amplifier) can be used as On/Off
switch by varying the bias voltage. Amplification in On state and
absorption in Off state makes this device achieve very high
extinction ratio.
Larger switches can be fabricated by integrating SOAs with passive
couplers
But its an expensive component and polarization dependant.

AWG is a device where several copies of the same signal but


shifted in phase by different amounts are added together. It can be
used as static OXC in the form of (nX1) wavelength multiplexer or
(1Xn) wavelength de-multiplexer.
It has lower loss and flatter pass-band and is easer to realize on an
integrated-optic structure. The i/p & o/p waveguides, the multiport
couplers and the arrayed waveguides are all fabricated on a single
substrate silicon. 32 channel AWGs are commercially available.
Modular form of AWGs, more versatile for wavelength routing with
suitable architectures has been reported.
Comparison of Optical switching technologies in tabular form :
Insertion Loss Cross Talk PDL Switching
(dB) (dB) (dB) Time
Optomechanical 0.5 -55 0.07 4 ms
8X8

MEMS 0.2-3.7 -50 0.4 12 ms


8X8
Electro-optic 9 -30 5 ns
8X8

Thermo Optic 8 0.5 3 ms


8X8

Liquid Crystals 1.4 -50 0.2 5 ms


2X2

Bubble 2.5-7.5 -50 0.3 10 ms


2X2

Acousto Optic 6 -35 3 s


1XN
Large (> 2X2) Optical Switch Considerations:
Number of small switches (2X2 or 1X2) cascaded Cost, Packaging,
Splicing & ease of fabrication.
Loss uniformity (for different combinations of i/p and o/p ports)
Number of cross-overs single layer integrated optics connections by
means of waveguides. Crossover creates power-loss and cross-talk.
Blocking characteristic Switching Architectures like cross bar.
Two possible architectures for Larger Switch fabrication presented here.
n-Stage Plannar Architecture ( A good compromise between
the Crossbar and Benes Architecture) which is re-arrangably non-
blocking and requires n(n-1)/2 switches. The shortest path length is n/2
and the longest path length is n. It has no crossovers but the main
drawbacks are
1) wide sense non-blocking
2) Loss non-uniformity
Optical Pkt Switching
With optical pkt..switching, pkt streams can be multiplexed together
statistically, making more efficient use of capacity and providing increased
flexibility over pure WDM (granularity at the level of a fraction of a
wavelength).
If pkt switching (header + DATA) is realized, it can
o Allocate WDM channels on demand (microsecond)
o Share network resource efficiently
o Support burst traffic efficiently
o Offer high speed data rate/format transparency and configurability
In a synchronous or slotted network, pkts arriving at the i/p ports
must be aligned in phase with a local clock reference. Maintaining
synchronization is not a simple task in optical domain. Assuming an
Internet environment, segmentation of IP datagrams to fixed length
pkts. At the edge of the network and reassembling at the other
edge is a problem at very high speed. In Asynchronous or un-
slotted network pkts are not of the same size. They are not aligned
and switch action could take place at any time (not predictable).
This leads to contention and increased pkt loss ratio. But has
obvious advantages with increased robustness & flexibility, lower
cost and ease of setup. A good traffic performance is attained and a
complicated pkt. alignment unit is avoided.
Use of Wavelength Domain for contention resolution counteracts the
disadvantages.
In a slotted network FDLs are used as optical buffers and a pkt can
take different paths unequal length within the switch fabric. The
switching matrix (controlled by routing information from header)
perform the switching and buffering functions.
Electronic processing (detecting and recognizing) of pkt. headers at
Gbps speed is still difficult to implement.
Subcarrier multiplexing where the header and payload are
multiplexed on the same wavelength (to avoid fiber dispersion) but
the pay load data are encoded at the base-band but the header bits
are encoded on a properly chosen subcarrier frequency at a lower
bit rate. This enables header retrieval without use of an optical filter.
But if the payload data rate will increase, then the base-band will
expand and might overlap with subcarrier frequency.
Contention Resolution in optical pkt. Switching
When two pkts must be forwarded to the same o/p channel at the
same time CONTENTION occurs and affects the overall network
performance.
Limitations in optical domain with no capability of bit level
processing and optical buffer. There is no efficient way to store
information in the optical domain indefinitely.
Buffering : Buffer contending pkts. And exploiting time domain
In traditional electronic pkt . switching, pkts are stored in the switchs
RAM until the switch is ready to forward them. Electronic RAM is
cheap and fast where as optical RAM doesnt exist. FDLs are the
only way to buffer a pkt. In the optical domain (contending pkts. Are
sent to travel over an additional fiber length and delayed)
FDLs are bulky and expensive, cant store pkt. Indefinitely. Also
once a pkt. enters a FDL, it cant be retrieved before iit emerges on
the other side i.e FDLs lack Random Access Capability.
Additional quality degradation through travel over extra fiber length.
The number of FDLs required to achieve a certain pkt. Loss rate
increases with traffic load.
The lengths of the FDLs are dictated by pkt. Durations.
The need for buffering is greatly reduced in synchronous network
operation.
Use of Wavelength dimension with TOWCs minimize number of
FDLs

If n wavelengths can be multiplexed on a single FDL, each FDL


has a capacity of n pkts. TOWCs can be used to assign pkts. To
unused wavelengths in the FDL buffers.
Deflection Routing exploits space domain for contention resolution
If two or more pkts. need to use the same o/p length to achieve the
minimum distance routing, then only one will be routed along the desired
link while others are forwarded on paths of greater distance. The deflected
pkts. may follow long paths to their destinations incurring long delays and
in addition sequence of pkts may be disturbed. (no queuing delay of buffer
but propagation delay)
A potential problem with the above is introduction of routing loops which
can be avoided by using a hop counter for each deflected pkt. and
development of deflection algorithm.
The most important advantage of this method is that it doesnt require huge
effort to be implemented neither hardware wise nor in terms of control
algorithms. The effectiveness of this technology critically depends on the
network topology. Mesh topology with higher number of interconnections
greatly benefit from deflection routing.
Wavelength Conversion Additional wavelength dimensions is
unique in optics can be utilized for contention resolution.
If two pkts of the same wavelength are addressing the same switch outlet,
one of them can be converted to another wavelength using a TOWC.
Only if the wavelengths run out is it necessary to resort to optical buffering.
So this this technique reduces the inefficiency in using FDLs particularly in
asynchronous networks.
A Proposed Pkt Switched Architecture (All Optical)
All Optical Wavelength converter with less PDL seems to be
the most crucial component in optical switching and
networking.
There are four approaches to achieve wavelength conversion
Opto-electronic approach (1R, 2R, 3R signal regeneration)
Optical grating (makes use of an optical device whose
characteristics change with intensity of the i/p signal) Only
intensity modulated signals can be converted.
Interferometric technique (MZI using CPM cross phase
modulation)
Wave Mixing (Four wave mixing phenomenon due to fiber
non-linearity)
The last three are all optical wavelength converter. All optical
wavelength converters are still in the research laboratories,
awaiting significant cost reductions and performance
improvements before they can become practical.
Switched Photonic Delay Lines for True Time Delay
Antenna Beam Steering
PAAs have features that are highly desirable for many commercial
applications such as cellular communication, sattelite
communication, air-traffic radars and other mobile platform antenna
systems.
In military applications TR antenna is used; typically fed
electronically from the processing station.
High loss, heavy and degrade SWAP efficiencies of the link.
Photonic Technology not only incorporate low loss,
lightweight, flexible cabling and immunity to EMI (optical fiber
cabling) but, offers two major advantages which is not there
in conventional electronic approaches.
A) Broad bandwidth
B) Antenna remoting over distances
In a photonic approach, remote control / PAA can be accomplished
using low-loss optical fibers to transmit the signal from the
controller side to the antenna side.
PDLs are key components to develop wide bandwidth, compact,
lightweight and small size PAA controllers.
More important as broad-band PAA systems start moving from sea
borne to air borne applications.
Conventional Electronic PAA Control Schemes
2
Incremental Phase Shift d sin
RF
It is however frequency sensitive : if RF excitation is changed and
accordingly is not changed the beam will move (Beam Squinting)
f
sin
f
True Time Delay Steering: To prevent beam squinting while
maintaining large instantaneous B.W (as in ferrite based MSA Arrays)
modulo 2 phase shifts are replaced by time-delay steering
The incremental time-delay d
sin
c
The total delay path length that has to be provided amounts to

[c ( T )] L sin m
Where m is maximum scan angle for aperture L
A typical schematic diagram of an electronic time-delay network shown below

Signal is routed via electronic switches through the N delay paths


whose length and thus the time of propagation increases
successively by a power of 2.
Total T can take any value from 0 to (2 N
1) in increments of

T (b01 20 b11 21 ....... bN 1 2 N )


Where the b coefficients can take values 0 or 1 (non-delay or
delay paths).
Beam Shaping can also be provided by appropriate
arrangement of feed signals ; to minimize pattern ripples, side
lobe levels, change Null positions or control of o/p power levels.
Broad beam acquisition of targets
Narrow beam high precision tracking subsequently
PAA can track many targets simultaneously on a time sharing
basis. (Air Traffic Control Radar)
CRITICAL FEATURES OF PDL

Of critical importance to high precision military applications is


design of delay lines for wide-band RF systems with Long
Delays with Low Intrinsic Loss.

SM low-loss optical fiber based technology is used for long


delay lines. Large time-bandwidth product is attainable with
glass fibers.
However impishness, environmental robustness, manufacturing
scalability and time-delay precision are important particularly for air-
bone application
Lithographically generated chip-scale PLC technologies offer the
precision and scalability but limited delay due to large propagation
loss in optical wave guides. (1dB/Km) compared to SM fiber loss
(0.3dB/Km). This leads to more insertion loss. Use of optical
amplifiers SOA to compensate loss would entail SNR and B.W
penalty.
The ability to realize therefore fiber like losses in a compact chip-
scale integrated photonic platform that can be readily integrated with
passive non-linear optical and active optoelectronic component is
one of the real need of the development of compact PDL system.
These capabilities dont exist today primarily due to lack of an
adequate integrated optical waveguide technology.
Switched and Non-switched photonic delay lines have been
proposed. The switching depends upon the presently available
optical switches (Opto Electronic)
Non fiber based techniques have also been proposed for
implementation of time delay. (Beam forming using free space
Interferometric architecture) based on polarization switching by 2d
spatial light modulators. Free space propagation based delay lines
using polarizing beam splitters and prisms have also been
proposed. 2d SLMs act as optical polarization switching element.
In a project report of Hellenic Naval Academy NY. optiflex
technology along with SOA switching circuits have been
proposed. A technique of accurately forming the fiber based
delays that can provide the low insertion loss & the long time-
delay is the OptiFlex circuit technology.
Optiflex based PDL is unique technology that can adequetly
address the stringent time-delay requirements of PDLs. The
concept of OptiFlex technology was invented and developed to
form arbitrary complex fiber optic-fabrics capable of transporting
of optical signals between termination points in optical network
equipment with negligible loss.
The optiFlex technology may be used to lay multiple fibers on a
piece of plastic sheet, providing low loss distribution, compact,
flexible overlays. The process is very accurately controlled via
automated equipment and is used to design construct and
terminate a flexible optical fiber fabric with sequential layering of
plannar processes to form an optical back-plane.
We see that the fiber delays can be implemented in small
optiflex circuits. Optilex technology is transparent to the fiber
used (high dispersion or high band radius)
Bits# Delay (ns) length difference(m)
1 0.1 0.02
2 0.2 0.04
3 0.4 0.08
4 0.8 0.16
5 1.6 0.32
6 3.2 0.64
7 6.4 1.28
All active switching, optical splitting/combining, filtering and
reference delays are performed on a PLC. Multi-fiber pigtails bring
the signal to/from fiber delays based on optiflex technology. SOAs
can provide switching speed of nano seconds with minimum
insertion loss and maximum on/off isolation.
A PDL architecture based on SOA switching and optiflex circuit.
In fact the future generation technology is PHOTONICS replacing
electronics though at present it co-exists with electronics at least
in the area of Communication, Networking and Radar technology

You might also like