Note: Descriptions are shown in the official language in which they were submitted.
CA 02326800 2000-11-22
CANADA
PATENT APPLICATION
PIASETZKI & NENNIGER
File P10021JTN
Title:
AN OPTICAL SWITCH AND METHOD
OF SWITCHING OPTICAL SIGNALS
Inventor(s):
Stylianos Derventzis
Steve Hill
Blaine Hobson
Ali Langari
Mike Liwak
Robert B. Thayer
CA 02326800 2000-11-22
-1-
Title: AN OPTICAL SWITCH AND METHOD OF SWITCHING
OPTICAL SIGNALS
FIELD OF THE INVENTION
This invention relates generally to the field of signal
communication and more particularly to the field of optical signal-based
communication systems. Most particularly, this invention relates to optical
signal switching or cross connecting in optical-based information and data
communication systems.
BACKGROUND OF THE INVENTION
Optical signals are now used extensively in signal
communication systems to carry digital information. Through the use of
Dense Wavelength Division Multiplexing (DWDM) vast amounts of
information can be densely packed onto optical signals, which make the use
of such signals highly desirable. DWDM means that a large number of
individual wavelengths (at present about 40 to 80 over each of the C and L
bands) can be simultaneously used to carry data in a single fibre as
multiplexed signal components.
At present optical signal networks typically take the form of
large rings or hubs, which are cross connected to smaller local rings, which
may in turn be connected to even smaller rings within a very localized area
all of which is typically based on Synchronous Optical Network (SONET).
At each connection between the various rings the appropriate optical signals
must be directed or routed in the appropriate direction. At present these
connections are made by optical-electrical-optical (OEO) switches or cross
connects, which require that the optical signal be converted to an electrical
signal, routed, reconverted to an optical signal and then sent on its way.
Essentially what is required is an ability to route individual
information or data carrying wavelengths or signal components in particular
CA 02326800 2000-11-22
-2-
directions according to the intended destination of the information. The
same wavelength or channel may be used to carry many different pieces of
information having the same or different intended destinations. Most
desirably information carried by the specificwavelength or signal component
should be routed according to the information being carried and its intended
destination. At present, the routing of signals requires the OEO process in
which an optical signal is converted to an electrical signal, routed
electronically, and then converted back into an optical signal again for
delivery to the new destination which can create a bottleneck. This
equipment is also expensive.
Using DWDM means that a single fibre can carry multiple
wavelengths. Carrying multiple wavelengths increases the need for a
reliable cross connect and increases the needed capacity in the cross
connect. The use of DWDM technology means that such cross connects
must have enough capacity in the future to be able to connect together
hundreds of wavelengths to benefit from the greater signal carrying capacity
per fibre that DWDM provides. Because of the large number (likely more
than noted above in the future) of wavelengths on each fibre, and the large
number of fibres in a bundle, cross connecting represents an ever more
critical bottleneck in the transmission of data through optical networks and
transmission systems.
A number of strategies have recently been proposed to
overcome the current switching bottleneck. In one strategy, sophisticated
management software is used in an OEO switch to automate the setting up
and tearing down of wavelength connections. However, such systems,
while flexible, are not keeping up with the increases in bandwidth at the
speed required. Further even when it is possible to route more bandwidth,
the expense can be enormous. Another strategy that has been suggested
is to use an all optical connection which eliminates the electrical interface.
For example, an optical switch has been proposed which uses Micro-
Electro-Mechanical Systems (MEMS) such as arrays of tiltable tiny mirrors
CA 02326800 2000-11-22
-3-
that are tilted or translated to direct the optical signals passing into the
switch through an input plane to first one then to another output port across
the body of the switch on an output plane. MEMS are still cumbersome and
relatively slow to switch between output ports, on the order of 50 to 100
milliseconds. Also, there is some question of whether the tiny mirrors will
reliably function overtime, due to electro-mechanical failure such as
stiction.
As larger MEMS arrays are used system alignment becomes critical.
Another strategy recently suggested is to use the surface of
tiny bubbles to redirect light onto new paths for switching purposes.
Questions remain as to the stability of the bubbles' structures over
connection lifetimes. Scaling up this technology to meet the increasing
cross-connect demands is limited due to the strict 1 to 1 link and blocking
nature of the switch, similar to the limitations of planar MEMS.
While providing for a more optically based apparatus than the
conventional OEO systems, both of the MEMS and bubble reflecting
systems require that the signal be separated into individual wavelengths to
accomplish the switching and routing of any optical signal. This is
cumbersome because for each additional signal channel, another
connection is required and as the bandwidth expands the number of
connections inside the switch becomes enormous. For example, for a 1 to
1 connection with 4 input fibres and 4 output fibres carrying 60 signals per
fibre requires at least 240 independent controlled mirrors or bubbles.
What is desired is a high speed switch for allowing signal
components to be effectively and flexibly routed, which does not require the
use of MEMS or other structures requiring strictly a 1 to 1 connectivity,
which
does not require the OEO conversion and which offers rapid switching of
wavelengths to various output ports according to their intended destination.
What is also desired is an ability to add capacity to an existing installation
to increase the number of fibre connections andlor the number of
wavelengths that can be routed.
CA 02326800 2000-11-22
-4-
SUMMARY OF THE INVENTION
An all optical switch according to the present invention can be
provided which overcomes the limitations of the prior art and is flexible in
routing. According to the present invention there is provided an optical
switch having N switch ports for switching optical signals, said optical
switch
comprising:
N bidirectional signal processors for splitting and combining
optical signals, wherein an optical signal passing in one direction through
any one of said bidirectional signal processors is split into K parallel
signals
and wherein one or more optical signals passing through any one of said
bidirectional signal processors in the other direction are emitted as a single
optical signal, said one direction being oriented into said switch and said
other direction being oriented out of said switch;
K signal delivery matrices, each of said signal delivery
matrices having N matrix ports and broadcasting one of said K optical
signals from any one of said N matrix ports to all other of said N matrix
ports;
and
a plurality of bidirectional signal selectors, at least one located
between each of said N bidirectional signal processors and each of said N
matrix ports to manage the optical signals being broadcast through said
switch between said ports by selecting or deselecting one or more signal
components from each of said K optical signals.
According to a further aspect of the present invention each of
said K signal delivery matrices comprises:
a symmetrical signal splitter having three connections
associated with each of said N matrix ports wherein an input signal received
by any one connection is split into two equal and parallel signals one of each
of which passes out of said symmetrical signal splitter through the remaining
two connections, and
a bidirectional optical power amplifier interposed between each
of said symmetrical splitters for boosting a power of each of said split
signals
CA 02326800 2000-11-22
-5-
as said split signals pass through the optical power amplifier to get to the
next port sufficiently to substantially equal a power of an input signal
received by said symmetrical splitter.
According to a further aspect of the present invention each of
said K signal delivery matrices comprises:
N second bidirectional signal processors for splitting and
combining optical signals, wherein an optical signal passing in one direction
through said bidirectional signal processors is split into (N-1 ) parallel
signals
and wherein one or more optical signals passing through said bidirectional
signal processor in the other direction are emitted as a single signal, said
one direction being oriented into said matrix and said other direction being
oriented out of said matrix;
an optical connection for each of said (N-1 ) signals between
each of said second bidirectional signal processors and each other port; and
a power amplifier associated with said switch to amplify a
power of the optical signals being switched by a predetermined amount.
According to a still further aspect of the invention there are
provided in said switch first and second bidirectional signal selectors for
each of said K signals for each of said N ports and an optical signal
circulator connected between each pair of bidirectional signal selectors and
said N signal processors, said optical signal circulator having at least three
points of connection and circulating an optical signal received at one
connection point out of the next adjacent connection point on said circulator,
and wherein said signal delivery matrices further comprise a bidirectional
broadcast coupler having a first side and a second side, each side having
N connections, one each of said connections on said first side being
connected to one first signal selector of each of said pair of signal
selectors
and one each of the connections on the second side being connected to one
second signal selector of said pair of signal selectors, wherein an optical
signal passing from either side of said bidirectional broadcast coupler to the
other side of said bidirectional broadcast coupler is split into N parallel
CA 02326800 2000-11-22
-6-
signals each of which is passed to each of said N ports through a respective
signal selector.
According to a further aspect of the present invention there is
provided a method of switching optical signals through a switch having N
switch ports comprising the steps of:
a) receiving a signal at one of said N switch ports; and then
b) dividing said received signal into K informationally identical
signals; and then
c) selecting or deselecting signal components from one or
more of said K like signals; and then
d) providing said selected signal components to at least one
other of said N switch ports; and then
e) combining said selected signal components with other
selected signal components received at said one other of said N switch
ports; and then
f) emitting said combined selected signal components from
said one other of said N switch ports.
BRIEF DESCRIPTION OF THE DRAWINGS
Reference will now be made to various figures which show, by
way of example only, preferred embodiments of the invention and in which:
Figure 1 shows a general architecture for an optical switch
according to the present invention;
Figure 2 shows a first embodiment of the present invention
according to the general architecture of Figure 1;
Figure 3 shows a second embodiment of the present invention
according to the general architecture of Figure 1;
Figure 4 shows a third embodiment of the present invention
according to the general architecture of Figure 1;
Figure 5 shows a schematic of a signal selector of one type
suitable for the present invention;
CA 02326800 2000-11-22
-7-
Figure 6 shows a schematic of an optical power amplifier
suitable for the present invention;
Figure 7 shows an algorithm of a control system of a type
suitable for use according to the present invention; and
Figure 8 shows a system diagram for an optical signal
processor or splitterlcombiner suitable for use with the present invention.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
An optical signal processing architecture for a preferred form
of switch according to the present invention is shown as 10 in Figure 1. It
will be appreciated by those skilled in the art that the term "switch
architecture" as used in this disclosure means a configuration of
components which in combination provide the optical signal switching or
routing functions as set out more fully below. In this disclosure, the term
switch includes, but is not limited to, a device which is generally capable of
directing optical signals and optical signal components, such as individual
wavelengths or channels as needed and comprehends routing functions
such as pure switching as well as addldrop functions and the like. Further,
the term signal means an optical signal which may be for example a DWDM
optical signal which includes one or more individual signal components,
such as wavelengths or channels.
The optical signal processing architecture for the switch or
router of the present invention is comprised of a number of elements having
specific functions as set out more fully below. The preferred elements
include a number of switch ports 12, at least one optical signal processor 14
associated with each switch port 12 at one end and being associated with
a matrix port 15 at the other end, a plurality of signal delivery matrices 16
extending between all of the matrix ports 15 and a plurality of bidirectional
signal selectors 18, one of which may be located, for example, at each
matrix port 15. Each of these elements is described in more detail below.
CA 02326800 2000-11-22
_$_
The term port as used in this specification means any type of
connection which permits an optical signal carrier to be connected in such
a way so as to establish a signal path into or out of any given component.
The most preferred form of switch port 12 is one which permits an optical
signal carrier such as a fibre optic cable to be securely connected to the
switch to establish an optical signal path from the fibre optic cable into the
switch. The form of the matrix port 15 can vary and can be quite simple,
provided the optical signal reliably passes into the signal delivery matrix
16.
According to the present invention the number of matrix ports
12 can be varied to suit the individual requirements of the switching or
routing application. Thus, in a generalized architecture as shown in Figure
1 switch ports 12 one and two are shown, and the switch is shown to be
capable of having up to N ports. Thus, the switch of the present invention
may include as many ports as may be needed, but most commonly between
3 and 20 switch ports 12 will be sufficient. Thus, the switch 10 can be
considered to have N ports, where N is any whole number greater than one.
The next element in the switch architecture 10 of Figure 1 is
an optical signal processing element 14 which splits or combines optical
signals passing through it. In one direction (from the switch port 12 to the
matrix port 15) the signal processor 14 splits the signals into K
informationally identical copies and in an opposite direction (from the matrix
port 15 to the switch port 12) the signal processor 14 combines the K or less
optical signals into a single signal. The absolute value of K will vary
depending upon the switch requirements for multicasting, bidirectionality,
redundancy and desired bandwidth. In this specification in a preferred
embodiment K is set, by way of example only, at NI2. However, K could
also be any number up to and even exceeding N as desired. Thus whether
the signals are split or combined depends upon the direction the signal is
passed through the optical signal processor 14. Further, since optical
signals can pass through the signal processor in either direction, the signal
processor may be considered to be bidirectional, even though different
CA 02326800 2000-11-22
_g_
results occur when passing through in one direction as compared to the
other. Such splittinglcombining can be accomplished either simultaneously,
or sequentially depending upon the circumstances.
It will be noted that each switch port 12 has shown a single
associated optical signal processor or splitterlcombiner 14, but in some
embodiments of the invention more may be used. The purpose of the signal
processor 14 is to split an optical signal 20 into a plurality of signals 22
in
one direction and to combine a plurality of signals 22 into one signal 20 in
the opposite direction. In this sense the term split means to divide into a
plurality of informationally identical or copied signals and does not mean to
demultiplex. It will be appreciated that depending upon the type of splitter
used, properties other than the information content can be varied, such as
the power. As an example, if an input optical signal having a power of 1 was
passively divided into K identical signals each of the K signals would have
a power of 1/K (assuming no losses through the splitter element). It will be
understood that the present invention also comprehends active splitters 14,
which would also provide an optical power amplification to permit each of the
copied (informationally identical) signals to be at full power. Thus according
to the present invention each of the informationally identical divided or
copied signals is still a multiplexed signal having all of the signal
components of every other divided signal, but not necessarily at the same
power.
The signal processor 14 is most preferably oriented so as to
split or divide signals passing into the switch 10 and to combine separate
signals into a single signal exiting or on the way out of the switch 10. For
ease of understanding arrowheads 24 show a signal traveling into the switch
and arrowheads 26 show a signal passing out of the switch. Most preferably
therefore, the signal processor 14 is a bidirectional signal processor which
divides a signal 20 being passed into switch 10 into a number, such as K,
informationally identical copies or split signals 22 and combines signal
components into signals in the reverse direction.
CA 02326800 2000-11-22
-10-
The next element of the switch 10 according to the present
invention is the signal delivery matrix 16. The purpose of the signal delivery
matrix 16 is to broadcast a signal emanating from one matrix port 15 to all
other matrix ports 15. In this sense the term broadcast means directing a
signal from one to many. Thus, the signal delivery matrix 16 is connected
by an optical path which extends to all of the matrix ports 15. To permit a
signal from any one port to reach all the other ports, it is appreciated that
the
signal delivery matrix is preferred to function to deliver signals and signal
components in either direction across any connection. Thus, for example,
a signal may be passed from port 1 to port 2 or from port 2 to port 1. In this
sense the signal delivery matrix 16 is also bidirectional. Again, the signals
may be either simultaneously transmitted or sequentially transmitted.
In some embodiments of the present invention optical signals
will be transmitted in a way that maintains their signal properties, such as
power, regardless of which way the signals are passing through the signal
delivery matrix 16. In other embodiments as described below the
amplification of the optical signals can be done outside of the matrix 16 and
the gain of the amplification set to compensate precisely for any power
losses within the signal delivery matrix 16.
As shown in Figure 1, each of the input signals at the N switch
ports is split into K copies and according to the present invention there is
preferably a signal delivery matrix 16 for each of the K copied signals split
out from an input signal by the signal processor 14. According to the
present invention there are at least three configurations or architectures for
the signal delivery matrix 16, each of which is explained in more detail
below. However, the present invention is not limited to any of the specific
architectures and comprehends other combinations of elements which
provide the signal broadcast function as described herein.
Before referring to the overall switch architecture in any greater
detail, it is important to understand some additional aspects of the present
invention. One such aspect is the provision of a device which acts as a
CA 02326800 2000-11-22
-11-
signal selector 18. This element is provided to manage the optical signals
being broadcast through the signal delivery matrix. In simple terms the
signal selector acts as a bidirectional gate (again either simultaneously or
sequentially), which selectively selects signal components which are to be
passed through the selector 18. It will be appreciated by those skilled in the
art that the selection of one or more signal components from a multiplexed
signal means that other signal components are thereby deselected from any
further transmission. Selected in this sense means enough of the signal
component is transmitted through the selector to permit further manipulation
of the signal component. Deselected means enough of the signal
component is blocked, absorbed, deflected or dispersed so that further
manipulation of the signal component is prevented. It will be further
appreciated that many forms of signal selector are possible, including any
of active electro, magneto, acusto or thermo optical effects or any passive
means such as fibre, bragg grating, thin film filters, fused coupler filters
and
the like. The present description is of only one form which is presently
considered to be the most preferred.
As shown in Figure 1, at least one signal selector 18 is most
preferably associated with each of said K signal outputs of said bidirectional
signal processors 14 at each port. Thus, for a switch port having an input
signal divided into K informationally identical signals or copies, K signal
selectors are required. According to one preferred form of the present
invention the signal selectors are bidirectional. In this sense bidirectional
means that signals may pass through the signal selector into the signal
delivery matrix 16 and may also pass out of the signal delivery matrix 16
through a signal selector 18. This permits the signal selector 18 to select or
deselect signals either entering or exiting the signal delivery matrix 16
through the ports 15.
The number of signal selectors required is derived according
to a simple mathematical formula. Specifically, in this embodiment, the
number of signal selectors required is equal to or greater than the integer of
CA 02326800 2000-11-22
-12-
(N/2) multiplied by N, where N is equal to the number of switch ports 12.
Thus, when N equals 6, or for a switch having 6 switch ports then at least 6
times (6I2) or 18 signal selectors are required. As noted below more may
also be required, as in one embodiment of the present invention twice this
number of signal selectors are required.
As described above DWDM signals comprise a plurality of
individual wavelengths multiplexed together. Each separate wavelength
may be considered as a separate signal channel or signal component. The
signal selectors perform the function of selecting one or more signal
components to pass through into the signal delivery matrix, or to be
permitted to be emitted out of the signal delivery matrix.
A preferred form of signal selector 18 according to the present
invention is shown in Figure 5 and indicated generally as 40. Since the
signal is selected according to wavelengths, the device 40 may also be
referred to as a lambda selector. A preferred lambda selector 40 according
to the present invention is characterized by having two optical signal
sources, such as optical fibres 38, 39 connected to ports 41 and 42. In this
sense port means a coupler or connector which permits optical signals to
reliably pass into and out of the device 40. Further, source comprehends
any component which passes an optical signal into the device 40 such as
a fibre, a lens, a splitter or other device. Since the selector 40 is
preferably
bidirectional, it accepts input optical signals at either of the two ports 41,
42.
Any input optical signal passes along a predetermined signal path through
the device 40 as will now be described. As described in more detail below,
the bidirectionality of the lambda selector may occur simultaneously for the
same or different signal components, or, may be sequential in time
depending upon system configurations and needs.
The first element adjacent to both ports 41 and 42 is a
collimating lens 46. As will be appreciated by those skilled in the art the
collimating lens has the effect in one direction of converting a divergent
beam path into a parallel or planar beam path. According to the preferred
CA 02326800 2000-11-22
-13-
form of wavelength selector 40 the divergent beam is turned into a parallel
beam as the signal is passed further into the device 40. For a signal
passing in the opposite direction, the opposite effect occurs, namely a
parallel beam is turned into a convergent beam.
The next element is preferably a means for multiplexing and
demultiplexing such as a transmissive diffraction grating 48. It will be
understood that other forms of multiplexerldemultiplexer can also be used
such as transmissive gratings, arrayed waveguides, prisms and the like.
This element 48 separates polychromatic light into its spectral content
spatially by producing parallel beams of light each at a different wavelength.
In the opposite direction it combines parallel beams of light of different
wavelengths into polychromatic light. The separation into individual
wavelengths occurs as the signal passes further into the selector 40. As will
now be appreciated, the signal is now de-multiplexed after passing through
the diffraction grating 48.
The next element in the lambda selector is a focusing lens 50
to direct the de-multiplexed wavelengths into a specific location in space.
Again each of the lenses 50 acts in both directions.
The central element of the device 40 is an optical shutter array
52 of which there need only be one. The optical shutter is a means for
selecting and deselecting signal components or wavelengths. This element
comprises multiple optical shutters individually controllable, one located in
each point in space corresponding to the location that each individual
wavelength has been directed. This is shown schematically in Figure 5a
which is an end view showing a transmissive shutter window 54 surrounded
by a mounting 56. The individual beams are focused onto individual
windows 54. It will be appreciated that an operational relationship exists
between the focusing lens 50 and the shutter array 52. The relationship is
that the lens 50 controls, in the plane of the shutter array, the size, shape
and physical location of each wavelength, and the individual shutters of the
shutter array are located and sized to be directly in a path of each of said
CA 02326800 2000-11-22
-14-
individual wavelength beams. The optical shutter array 52 in the preferred
form has the ability to change opacity in rapid fashion for example, in
response to an electrical signal, to selectively permit individual wavelength
beams to pass through or to substantially block (i.e. deselect) the same.
According to a preferred form of shutter, the active element is a PLZT (a
combination of lead, lanthanum, zirconate and titanate) structure which is
capable of being excited by an electrical field which has the effect of
switching the opacity or light transmissiveness, believed to be in the order
of 20 x 10-9 seconds. This fast switching speed thus determines the speed
of the switching or routing of the signal components through the switch of
the present invention.
It will be appreciated by those skilled in the art that other
means for selecting and deselecting signals can be used. For example, the
shutter could be used to adjust the polarization of the signal, which can then
be selected on the basis of polarization. Other means of selecting or
deselecting can also be used, but what is desired is a means which is
rapidly operable to select and deselect signals.
It will now be appreciated that the interaction between the
selective switching of specific wavelengths of the optical signal by the
shutter array 52 and the use of the focusing lens 50 permits the selector 40
to select or deselect individual wavelengths. Thus, individual wavelengths
or signal components may be separated out from the multiplexed signal
according to switching or routing requirements. Signals which have been
deselected or blocked are not permitted to pass through the lambda
selector. Signals which are permitted to continue are then recombined and
passed to the other output port for further transmission.
According to the present invention it is preferred if both ports
41 and 42 are bidirectional, meaning that each port may be passing optical
signals through the amplifier in both directions simultaneously. As will be
appreciated by those skilled in the art, signals traveling in opposite
directions
will readily pass through one another without degrading the quality of any
CA 02326800 2000-11-22
-15-
opposite traveling optical signal. In some cases however, it may be
preferred to operate the device sequentially or to provide for separate
simultaneous amplification paths. This would be desirable in the event there
is any reflection of input signals or back directed Amplified Spontaneous
Emissions (ASE) which could add to and confuse output signals being
passed through the signal selector in the opposite direction. Thus, the
present invention comprehends a switch which may have signals andlor
signal components passing through elements in both directions
simultaneously, and also comprehends a switch which has signals and/or
signal components passing through in one direction only or first in one
direction, then in the other direction, sequentially. Further the present
invention comprehends permitting signals components to pass through in
one direction but the same signal components being blocked in the other
direction.
To facilitate the manipulation of optical signals it is preferred
to have the signals at a predetermined power. Some of the components of
the router according to the present invention discussed above have the
effect of altering or reducing the power, such as the signal processor 14
which splits and combines the signals. Others introduce transmission losses
such as the signal selectors. Generally it is desirable to have all signals
leave the switch according to the present invention with approximately the
same power as they arrived with, or more precisely, with the same power at
which optical signals are provided within the optical network. This power
level may be referred to as an operating power level. Thus, it is preferred
according to the present invention to use a signal power amplifier as needed
to boost the power signal so that the optical signals leaving the switch 10
will
conform with operating power requirements. This is referred to as lossless
switching or routing.
A preferred bidirectional optical amplifier 60 according to the
present invention is shown in greater detail in Figure 6. In this sense an
optical amplifier is any device which provides optical gain to an incident
CA 02326800 2000-11-22
-16-
optical signal. As shown a first coupler 62 is provided on a first side 64 and
a second coupler 66 is provided on a second side 68. The couplers 62 and
66 are conventional optical connections which permit the passage of an
optical signal in either direction through the coupler. For example, they can
be collimating lenses to collimate or collect optical signals entering or
exiting
waveguides 22 or 26. Located between the first and second optical
couplers is a doped body 70, which comprises the power enhancing or gain
block part of the amplifier 60. The doped body 70 may be made from any
appropriate optically transmissive material, such as fused silica, and good
results have been achieved in the present invention with phosphate glass.
The planar body is doped with an appropriate rare earth, examples of which
include erbium, neodymium, praseodymium, ytterbuim or mixtures thereof.
Erbrium is most preferred because it can provide relatively constant gain to
signal within the C and L bands, which is standard for optical signals. This
amplifier may be referred to as an OFSA (Optical Free Space Amplifier). It
will be appreciated that while a preferred form of the body is a planar body,
it may in fact be formed in other shapes or configurations. However, for the
reasons set out below the preferred planar body provides adequate results
in the present invention.
The amplifier 60 includes a pump source 72 which pumps
power into the gain block 70. The preferred pump source is a laser tuned
to a wavelength in the high absorption bands of the doped material, and as
will be appreciated by those skilled in the art such wavelength will vary for
different compositions of doped glass. As will be appreciated by those
skilled in the art, the pumping of the gain block 70 at the pump frequency
causes a population inversion among the electron energy levels of any
dopant or codopant, from a low energy to a high energy. When impinged by
incident light (i.e. the signal) the electrons return to the low energy state
releasing photons. The released photons amplify the optical signal passing
through the gain block 70. Thus, an at least partially unguided input optical
CA 02326800 2000-11-22
-17-
signal 71, for example, is amplified to an at least partially unguided output
optical signal 73, which is shown schematically at higher power.
Although typically optical amplifiers are found located within
waveguides such as a length of optical fibre, the at least partially unguided,
but directed, configuration of the preferred amplifier 60 has certain
advantages. One advantage is that it permits the pump laser 72 to be
directed on the gain block at an incident angle remote from the direction of
the information carrying signal and most preferably transverse to the
direction of the information carrying signal. Significantly this allows any
desired pump wavelength to be used including wavelengths within the signal
bandwidth. Further, stray photons known as ASE, which are produced as
the information carrying signal passes through the gain block 70 are
permitted to pass out through the sides of the gain block 70, where they do
not interfere with the information carrying signal, rather than being captured
and propagated by a wave guide, for example. This reduces the noise in
the amplified signal and is especially desirable in a bidirectional amplifier
as
comprehended by the present invention. Thus, according to the present
invention the preferred amplifierwill accept input light from eitherwaveguide
22 or 26, at an input power of Pin. This input light is then amplified to Pout
by the time the light reaches the other coupler on the other side of the gain
block 70. The gain may therefore be simply expressed as G = PoutIPin.
Having described the above noted components which form the
switch architecture, a switch having a broadcast configuration according to
the present invention can now be understood.
Figure 2 shows a first embodiment of a switch architecture
according to the present invention which is referred to as a port to port
replication type switch. In this example, a switch having a number of ports
12 is shown. It will be understood that the diagram of Figure 2 represents
one plane from the generalized architecture set out in Figure 1 of the signal
distribution matrix 16.
CA 02326800 2000-11-22
-18-
As shown, each port has a signal path indicated as 80, 82, 84
and 86 for port N. The signal paths are indicated with double ended arrows
indicating that optical signals may pass in either direction either from or to
the ports. Next is shown a respective signal selector 18 for selecting
specific signal components to pass either into or out of the signal delivery
matrix 16. It will now be appreciated that the signal selectors act as gates
in both directions, managing the signals entering the signal delivery matrix
as well as managing the signals exiting the signal delivery matrix at any
given port.
The signal delivery matrix 16 in this embodiment comprises
two components in combination. In particular, the signal delivery matrix
includes a plurality of symmetrical splitters 88. These devices are
characterized as ones which have three nodes, 90, 92 and 94 as shown.
Thus, an input signal received at any node (say 90) will be split into two
informationally identical signals one of which is sent out to each of the
other
two nodes (92 and 94). As can now be appreciated, for a passive splitter 14
this has the effect of reducing the power of each of the split signals, to
approximately one half of the power of the input signal less internal power
losses. Alternately, if an active or lossless splitter 14 is used there would
be
no change of power. Therefore associated with each passive symmetrical
splitter 88 is a power amplifier 96, preferably of the type described above,
which is tuned to deliver a predetermined amount of gain to raise the signal
power to equal the input signal power. In this way, all signals input into any
of the successive symmetrical splitters 88 will have the same power,
because they will have been boosted by the interposed optical amplifiers 96.
In this manner, any given signal can be communicated to all other ports at
the same power as the power at which they started. Also shown is a second
amplifier 100 which is optionally located outside of the signal selector. It
will
be appreciated by those skilled in the art that the preferred optical
amplifiers
of the present invention can be located at various points of the architecture
CA 02326800 2000-11-22
-19-
as required to supplement the signal power as needed, depending upon the
switch architecture.
It can now be appreciated how the switch architecture of the
first embodiment operates. A signal is received at port 1 and then is split
into K identical, but lower power, signals. If there were no losses through
the signal processor 14, the power of each of the K signals would be
reduced to 11K. In practice, there will be losses, but these losses can be
managed by means of amplification as noted previously. The splitting of the
input signal then provides an identical signal to input into K signal
matrices.
At each of the N ports, each of the K signals is then passed to a signal
selector which is capable of selecting or deselecting any specific signal
components from the multiplexed signal. All signals which have been
selected are then communicated (broadcast) to every other port of the
switch by passing through successive symmetrical splitters 88 and
associated power amplifiers 96. The symmetrical splitters then ensure that
the selected signal is delivered to every other port in the system and at the
same power as when it entered the signal delivery matrix 16. In addition, at
any given matrix port, the signal components can be either routed into the
port, or not routed into the port by the bidirectional signal selector at that
matrix port. Thus any given signal component can be selected at one matrix
port and then delivered to any other matrix port and be permitted to exit at
that matrix port. Thus the present invention provides that every signal can
be presented to every other matrix port, because it is not known in advance
at which switch ports any given signal is required. Once this is determined
the signal components can be permitted to pass out of the matrix port 15;
combined and passed out to the signal carrier through the switch port 12.
Further, by providing K signal delivery matrices, signal mixing of the same
wavelengths is avoided and yet enormous flexibility of signal routing is
provided.
A second embodiment of the signal delivery matrix of the
present invention is shown in Figure 3. This embodiment is referred to as
CA 02326800 2000-11-22
-20-
a direct replication type and like numbers refer to like components as in
Figure 2. Under this architecture, the signals are received coming into the
port in the same manner and then one or more signal components are
selected from the signal by a signal selector 18. Then the signals are
passed through a second signal processor 102, which divides the selected
signal into N - 1 informationally identical signals, where N is the number of
switch ports 12. The second signal processor 102 may be of the same type
as 18 described above. Each of the N nodes of the second signal processor
is connected by an optical pathway to a second signal processor associated
with each of the other ports in the switch 10. Thus, each signal which is
selected and passed into the signal delivery matrix is delivered to every
other port. Thus, at each delivery port, a signal selector 18 will further
select
or deselect signals to permit the further selected signal or signal component
to pass through and thus out of the port.
Again it is preferable to ensure that the signals being passed
through the switch 10 have an even power. Thus, again there is preferably
provided a power amplifier to boost the signal to ensure that the signals are
uniform. To this end the power amplifier may optionally be provided in the
signal connection between ports, or at any point between the port and the
feed optical fibre into the switch as shown schematically by 104. Since the
power amplifier of the present invention is bidirectional and since it will
only
boost the power to a predetermined maximum, the amplifier could also be
positioned on the input connection to the optical fibre at 100. Incoming
signals may be amplified and outgoing selected signals may also be
amplified and the total amplification or power gain can be set to any
predetermined level, but most preferably will be set to a relevant operating
power level before re-entering the network from the switch 10.
A third embodiment of the signal delivery matrix of the present
invention is now shown at Figure 4 which may be referred to as a star
replication type. In this embodiment there is shown an additional
component, namely a circulator 110, associated with each switch port. A
CA 02326800 2000-11-22
-21-
circulator 110 is a known device which may be considered to have three
nodes or connection points indicated as A, B and C. It is characterized by
being able to receive input signals at any connection point and then to pass
the signals out at the next adjacent connection point on the circulator. The
circulator 110 has the effect of dividing the signal path into an input signal
path into the signal delivery matrix and an output path from the signal
matrix,
which are essentially in parallel. Associated with each of the input and
output paths are an input signal selector 120 and an output signal selector
140. It will be appreciated that input and output are chosen for ease of
reference and that either of the signal selectors 120, 140 could be either an
input or an output selector. Most preferably the signal selectors are of the
preferred type previously described.
The next element of this embodiment is a bidirectional
broadcast coupler sometimes referred to as a star coupler. This coupler is
shown schematically as 130 and essentially comprises a pair of back to
back signal processors 132, 134 of the 1 to K splitter/combiner type such as
previously described. In this manner every input signal selector is
connected by an optical pathway to every output signal selector of every
other node. It will be appreciated that two back to back signal processors
are provided to permit all of the selected signals received in the signal
delivery matrix 16 as outputs from the output signal selectors 140 to pass
through a splitter so as to deliver to every other port the selected signals.
As can now be appreciated, the architecture according to the
present invention is characterized in that for each embodiment, any signal
can be directed or routed from any of the N ports to any one or more of the
other N ports. Further, through the use of the signal selectors the routing
can be changed at a rate equal to the switching speed of the shutter
element, which is presently estimated to be 20 nanoseconds (20 X 10-9
seconds). To achieve such shutter control, however, requires a control
system for the switch, which is shown in schematic in Figure 7.
CA 02326800 2000-11-22
-22-
Essentially the control system will receive signalling
information from the network at 200, for example, by a control channel for
this purpose. In addition the control system will have a set of desired signal
properties 202 against which a set of measured signal properties 204 can
be compared. The measured signal properties can include one or more of
polarization, power, and wavelength of a given optical signal component for
example and even polarization mode dispersion (PMD). In addition, an input
into the control system may be optical headers 206 which are embedded or
carried by the signal itself. All of these inputs are fed into the control
system
208. As well, the control system 208 will rely on the switch characterization
data shown at 210. Based on the foregoing inputs the control system 208
will change any controllable parameters as needed such as a gain of an
optical amplifier or a condition of a particular signal selector to achieve a
desired result shown as 212.
Figure 8 shows a further aspect of the present invention.
Figure 8 shows schematically a lossless scalable splitter/combiner or
replicator combiner. In Figure 8 a main port is shown at 200, with a signal
path 202 to an auxiliary port 204. These signal paths are formed by lossless
couplers 206. A lossless coupler is desirable because it permits field
installed additional connections within the switch to improve capacity. A
lossless coupler 206 permits additional matrices to be added to the switch
as more capacity is added to the signal in terms of additional signal
components. Provided enough excess capacity is built into the lambda
selector, additional signal component processing can be achieved, making
the present invention field scalable. Figure 8 describes a scalable structure
by virtue of adding more couplers 206. Thus, over time, rather than
replacing a whole switch to increase capacity, all that is required is to add
further switch ports and associated matrix connections to achieve higher
bandwidth capacity.
The switching of signals according to the present invention can
now be described. First fibres will be connected to the switch, with a single
CA 02326800 2000-11-22
-23-
fibre being connected to each of the switch ports. The number of ports can
be set according to the connection demand of the location the switch is to
be used in and a large number of optical fibres can be optically connected
through the switch so as to permit a signal to pass from one fibre to another
fibre. Once the fibres are connected, the control system is advised so that
each connection is known and assigned.
The next step is to route or switch an inbound signal for
example. The full multiplexed signal is passed into port 1. Then, the signal
is split (or copied), and passed through a signal selector where the specific
signal components are selected or deselected. The next step is to internally
broadcast the selected signal components throughout the signal delivery
matrix, in a way that permits the selected signal components to be delivered
to each and every other matrix port. Then the selected signals are further
selected or deselected to determine which if any signal components should
be passed out through each port. In this sense, it will be understood that
any fibre connection is a bidirectional connection with signals and signal
components passing both into and out of the switch therethrough, again
either simultaneously or sequentially.
As discussed above, the present invention preferably
accomplishes this switching on an essentially lossless basis, meaning that
an optical signal leaving the switch will be emitted at a predetermined power
level. Most preferably, the power of the switched signals (i.e. the
predetermined power level) will be set at a power value which is suitable for
combining with optical signals at that position in the optical network.
It will now be appreciated that the present invention provides
a switch architecture that enables frequency reusability within a given switch
according to the present invention. For example, in a four port system,
wavelength 1 can be routed from port 1 to port 2 and at the same time,
wavelength 1 carrying different information can be routed from port 4 to port
3. Thus, much more capacity is added to the signal carrying network
CA 02326800 2000-11-22
-24-
because the signal components can be used as needed in a point to point
routing scheme.
It will also be appreciated that the present invention therefore
permits the development of a unlimited cross-connection mesh network
architecture for signal transmission networks, rather than a connected ring
architecture as is presently used in the field although it can be used in a
ring
architecture also.
Another feature of the present invention is that the routing
takes place at the trunk level. Although individual signal components are
routed after being demultiplexed in the signal selector, multiple switching of
the multiplexed signals takes place by virtue of the K signal delivery
matrices. Switching takes place because K informationally identical copies
of the signal components collectively or at the trunk level are used, rather
than as in the prior art separating the signals into signal components and
then producing copies of each signal component individually. Since a
plurality of selected signals are broadcast throughout the signal delivery
matrix at the same time, the routing of signals through the switch is both
fast, efficient and significantly less costly per wavelength.
Further, an important advantage provided by the present
invention is that the architecture is bidirectional as an overall structure.
What is meant by this is that any signal received at any port can in fact be
sent to any other port. Thus, unlike the MEMS devices and the like currently
employed, the present switch invention is not limited to a defined set of
input
ports and a defined set of output ports. In the present invention most
preferably any port can act as an input or an output port, even
simultaneously for different signal components.
It will be appreciated by those skilled in the art that while the
present invention has been described in relation to various preferred
embodiments, there are other variations which may be made which do not
depart from the broad scope of the invention as defined by the attached
claims. Some of these have been discussed above and others will be
CA 02326800 2000-11-22
-25-
apparent to those skilled in the art. For example, although the OFSA
amplifier is preferred other forms of amplification can be used to achieve the
same result. Also, while the signal selector as described is preferred for its
speed and bi-directionality, other devices for signal selection are also
possible. What is considered important for the present invention is to
provide a signal delivery matrix in combination with signal component
selection and de-selection in a way that any signal components received in
any port may be collectively transmitted to any other port or ports.
