Note: Descriptions are shown in the official language in which they were submitted.
CA 02320613 2000-09-21
1
SWITCH FOR OPTICAL SIGNALS
Field of the Invention
This invention relates to optical switchEa and is particularly
s concerned with switches for switching optical signals composed of light
of predetermined wavelengths, for example, Dense Wavelength Division
Multiplexed (DWDM) optical signals used in optical telecommunications.
Background of the Invention
Optical transmission systems achieve their end-to-end
connectivity by concatenating multiple spans between intermediate
switching nodes to achieve an overall end-to-end path. When the end-to-
end granularity of any given transmission path is a fraction of the capacity
of a given optical carrier, time division multiplexing is used to share the
15 overall bandwidth, mandating the use of electronic switching in the
intermediate nodes. However, the availability of Dense Wavelength
Division Multiplexing (DWDM), combined with the availability of high
capacity ports on data switches and routers, has increased the demand
for concatenation of individual spans to make end-to-end connections at
2o the wavelength level.
DWDM optical networks transmit multiple channel signals on
each optical fiber in the network; each channel signal is modulated light of
a predetermined wavelength allocated only to that signal. The result is a
plurality of optical carriers on each optical fiber, each optical carrier
25 carrying a channel signal separated from other carriers in optical
wavelength. Current DWDM optical networks typically convert channel
signals into electrical signals at every switching node in the network
because optical switches having sufficiently large enough port counts are
not available. To convert the channel signals to electrical signals,
3o transponders are used at every port of the switching node and for every
channel wavelength. As DWDM signals become denser, that is, as the
number of channels per optical fiber increases, the required accuracy of
the transponders, and hence the cost, also increasE;s. Moreover, as the
CA 02320613 2000-09-21
2
number of ports per switching node increases, the required number of
transponders also increases. Consequently, large networks carrying dense
DWDM signals require many costly transponders and are therefore costly
to build.
To overcome this problem it has been proposed to build large,
purely optical switches in various forms, to reduce or eliminate the need
for opto-electronic conversion in order to switch channel signals
electrically. Some effort has gone into conceiving methods of building
very large switches that offer full connectivity between all their ports.
to However, fabrication of these large optical switches has proven difficult.
Many attempts to create a large non-blocking optical switch
use a large number of small switch modules to create a multiple stage
switch. One example of this envisages building a '128 port x 1 28 port
switch out of three stages of multiple 16 x 16 crosspoint matrices, or a
~s 51 2 x 512 port switch out of three stages of multiple 32 x 32 crosspoint
matrices, in a three stage CLOS architecture. The above is based on the
availability of 16 x 16 or 32 x 32 switch matrices in the form of Micro-
Electro-Mechanical (MEM) switch matrices (e.g. "Free-space
Micromachined Optical-Switching Technologies and Architectures", Lih Y.
2o Lin, AT&T Labs-Research, OFC99 Session W14-1, Feb. 24, 1999). Other
multi-stage approaches use smaller matrices and more stages. Even the 3
stage CLOS architecture is limited to 512-1024 switched wavelengths
with 32x32 switch matrix modules, which, in today's 160 wavelength
per fiber DWDM environment, is only adequate to handle the output/input
2s to 3-6 fiber pairs (480-960 wavelengths). Furthermore, the optical loss
through each crosspoint stage (typically -5 dB with a 16x16 or 32x32
MEMs device) is compounded by the use of three stages, plus a complex
interconnect, to provide switch losses in the range of 15-18 dB.
Such multi-stage switches, even at three stages, have
3o significant problems. These problems include high overall optical loss
through the switch, since the losses in each stage are additive across the
switch, and there is the potential for additional loss in the complex
internal interconnect between the stages of the switch. Size limitations in
terms of the number of wavelengths switched can be overcome by going
3s to a five stage CLOS switch, but this further increases the loss through
the switch as well as adding to its complexity and cost. Using current
CA 02320613 2000-09-21
3
loss figures, the loss through a 5-stage switch wauld be in the order of
25-30 dB. This amount of loss is at or beyond the operating link budget
of modern high-bandwidth transponders. In addition, one of the major
cost-centres is the cost of the MEMs switch modules (or other small
s matrix modules). Sensitivity of the overall switch cost to the cost of the
MEMS modules is exacerbated by the fact that a CLOS switch requires a
degree of dilation (i.e. extra switch paths) to be non-blocking and that
each optical path has to transit three (or five) individual modules in series.
In U.S. patent 5,878,177 entitled "Layered Switch
1o Architectures for High-Capacity Optical Transport Networks" and issued
to Karasan et al., on March 2, 1999, another approach is disclosed. This
approach relies on providing signals received by a switching node with
access to any route leaving the node, but not access to every signal path
(fiber) on those routes. In this way, Karasan's switching node avoids the
t5 large number of switch points that a fully interconnected, or fully non-
blocking, switch fabric would require. Although this approach may be
adequate at the node level, or even for small networks, it adds further
complexity to network planning, which would become increasingly
difficult with larger networks.
2o Some prior art approaches attempt to generate large, general
purpose, non-blocking switches, which are then caupled to DWDM
multiplexers for coupling into output fibers. This results in substantial
waste of the capacity and capability of the non-blacking generic switches,
since the DWDM multiplexers are themselves blocking elements on all
2s their ports to any optical carrier except an optical carrier within the
specific passband of that port of the multiplexer. Hence the non-blocking
switch structure contains many crosspoints that direct specific input ports
carrying a given wavelength to output ports that cannot support that
wavelength, since it would be blocked in the WDM multiplexer. Such
3o crosspoints cannot be used in operation of the switch, and this wasting of
crosspoints makes inefficient use of expensive optical switching matrices.
Optical transmission networks that rely on electrical switching
and electrical regeneration at intermediate nodes require one pair of
transponders per wavelength channel at each intermediate switching
3s node. Consequently, as the number of wavelength channels per fiber
CA 02320613 2000-09-21
4
grows, the number of transponders and the resulting costs grow in
proportion to the number of wavelength channels.
Optical transmission networks that rely on "opaque" optical
switching and electrical regeneration at intermediate nodes experience the
same growth in transponder number and cost. (In "opaque" optical
switching, incoming optical signals are converted by transponders into
different optical signals that are switched optically before being converted
by further transponders to different optical signals for further
transmission.)
1o However, in optically switched networks that use cascaded
optical amplifiers to compensate for fiber loss on each span and for
optical insertion loss of the optical switches, each optical amplifier
simultaneously amplifies all wavelength channels an each fiber without
the use of transponders. Consequently, the number and cost of the
is optical amplifiers does not grow with the number of wavelength channels
per fiber, and the cost benefits of optically switched and amplified
networks relative to electrically switched and regenerated networks
increases with the number of wavelength channel:; per fiber.
Moreover, the cost advantages of optically switched and
2o amplified networks over electrically switched and regenerated networks
grow even faster as the maximum distance between electrical
regeneration points grows, because optically switched and amplified
networks can benefit from that increased optical rE:ach by eliminating
transponders. In contrast, electrically switched networks require a pair of
2s transponders per wavelength channel at each intermediate switching point
even if the optical range exceeds the distance between switching points.
Consequently, there is a substantial advantage in designing
optical transmission networks such that the majority of wavelength
channels can be routed end-to-end via optical switches and optical
3o amplifiers, without the use of transponders on a pE:r channel wavelength
basis at intermediate sites or nodes. This leads to a need, previously
unaddressed, for an optical cross-connect switch optimized for
establishing per-wavelength paths from end-to-end, as opposed to a large
opaque optical switching fabric designed to be located between banks of
3s transponders.
CA 02320613 2000-09-21
Summary of Invention
This invention aims to provide an improved cross-connect switch
which is well adapted for application to high capacity Wavelength Division
s Multiplexed (WDM) and Dense WDM (DWDM) transmission networks.
A first aspect of the invention provides a cross-connect switch
comprising a plurality of switching matrices and a wavelength-converting
inter-matrix switch. Each switching matrix has multiple input ports,
multiple output ports, at least one inter-matrix input port and at least one
to inter-matrix output port. Each switching matrix is operable to switch an
optical channel signal arriving on any input port to either any one of a
plurality of the output ports or an inter-matrix output port. Each switching
matrix is also operable to switch an optical channel signal arriving on any
inter-matrix input port to an output port. Each switching matrix is further
is operable to switch optical channel signals having a respective distinct
wavelength. The wavelength-converting inter-matrix switch is connected
between the inter-matrix output ports of the switching matrices and the
inter-matrix input ports of the switching matrices. The inter-matrix switch
is operable to switch a channel signal arriving from any inter-matrix
20 output port of any switching matrix to an inter-matrix input port of any of
a plurality of other switching matrices. In switching a first channel signal
having a first wavelength from an inter-matrix output port of a first
switching matrix to an inter-matrix port of a second switching matrix, the
wavelength-converting inter-matrix switch is operable to convert the first
2s channel signal having the first wavelength to a second channel signal
having a second wavelength.
Preferably, each switching matrix is operable to switch a channel
signal arriving on any input port to any of the output ports. Furthermore,
in such switches, the inter-matrix switch is operable to switch a channel
3o signal arriving from any inter-matrix output port of any switching matrix
to an inter-matrix input port of any of the other switching matrices. In this
way, when networked together, such cross-connect switches provide
increased flexibility in switching channel signals, thereby reducing the
complexity of network planning as compared to other approaches.
CA 02320613 2000-09-21
6
This arrangement between the switching matrices and the inter-
matrix switch enables the assignment of each switching matrix to a
respective channel wavelength of a WDM system. Channel signals having
a particular wavelength can be routed through the cross-connect switch
s in the switching matrix assigned to that respective wavelength. Because
this routing is through a single optical switching matrix, the optical loss
can be relatively low.
When the next span of an end-to-end path does not have a
particular channel wavelength available for a channel signal, the channel
to signal needs to be cross-connected to another channel wavelength. This
cross-connection requires transponders to perform the necessary optical
carrier wavelength conversion. This can be done by routing the channel
signal, of a first channel wavelength, through a first switching matrix
assigned to the first wavelength, to an inter-matri;K output port of the first
15 switching matrix. The channel signal is then routed from the inter-matrix
output port of the first switching matrix to the wavelength-converting
inter-matrix switch. The wavelength-converting inter-matrix switch
converts the channel signal of the first wavelength to a channel signal of
a second wavelength. The channel signal of the second wavelength is
2o then routed to an inter-matrix input port of a second switching matrix,
which is assigned to the second wavelength. The channel signal of the
second wavelength is then routed to an output port of the second
switching matrix, which completes the routing through the cross-connect
switch to the next span, as required. Since wavelength conversion is only
2s done as necessitated by network constraints, the cross-connect switch
requires substantially fewer transponders than switches that convert all
channel signals to electrical signals, or to a common channel wavelength,
prior to switching.
Each switching matrix may have multiple inter-matrix output ports,
3o and the wavelength-converting inter-matrix switch may comprise multiple
switching elements connected in parallel. In this case, each inter-matrix
output port of a particular switching matrix may be coupled to a
respective one of the switching elements of the wavelength-converting
inter-matrix switch. This arrangement provides multiple paths for routing a
35 signal from one switching matrix through the inter-matrix switch to
another switching matrix, thereby reducing potential for blocking in the
inter-matrix switch.
CA 02320613 2000-09-21
7
Moreover, the physical interconnection between the multiple
switching elements and the plurality of switching matrices may be
accomplished efficiently by orienting the switchinca elements into a first
set of parallel planes that are orthogonal to a second set of parallel planes
s into which the switching matrices have been oriented. For example, the
switching matrices could be implemented on horizontally oriented
switching cards and the switching elements fabricated on vertically
oriented convertor cards, or vice versa. This physical arrangement allows
the two orthogonal sets of parallel planes to be intersected by a third
to orthogonal plane, orthogonal to both sets of parallel planes, whereby each
switching matrix of the second set of parallel planes can be brought into a
proximal relationship, and optically interconnected, with each switching
element of the first set of parallel planes. For example, a midplane
representing the third orthogonal plane can be used to guide the switching
is cards and the convertor cards into a close physical arrangement, in which
the switching and convertor cards can be optically interconnected with
appropriate optical connectors on the cards and the midplane.
The inter-matrix switch may comprise at least one "add" input port
and at least one "drop" output port. In this case, the inter-matrix switch
2o is operable to couple an "add" input channel signal arriving at the "add"
input port to an inter-matrix input port of any switching matrix, and to
couple a channel signal arriving from an inter-matrix output port of any
switching matrix to the "drop" output port. These features enable the
cross-connect switch to "add" channel signals (i.e. to insert traffic signals
2s at the cross-connect switch) and "drop" channel signals (i.e. extract
traffic signals at the cross-connect switch) in addition to routing through
channel signals.
The cross-connect switch may further corrrprise a plurality of
wavelength division demultiplexers and a plurality of wavelength division
3o multiplexers. Each demultiplexer is operable to separate an optical input
signal into a plurality of output channel signals having respective distinct
wavelengths. The demultiplexer applies each output channel signal to a
respective input port of a respective switching matrix such that each
switching matrix receives only channel signals having a respective distinct
3s wavelength. Each multiplexer has a plurality of inputs, each respective
input of each multiplexer being coupled to an output port of a respective
switching matrix to receive a respective channel signal having a
CA 02320613 2000-09-21
8
respective wavelength. Each multiplexer is operable to combine channel
signals having distinct wavelengths into an optical output signal.
Such wavelength division demultiplexers and wavelength division
multiplexers are normally associated with the cross-connect switch and
s may be packaged as part of the cross-connect switch. In this case, the
wavelength division multiplexers and demultiplexers, implemented either
separately or in combination on circuit cards, could have an orthogonal
physical relationship with the plurality of switching matrices, in order to
achieve efficiency in interconnection as described earlier. The
~o demultiplexer receives an optical signal comprising multiple channel
signals, each channel signal comprising an optical carrier at a respective
distinct wavelength having a respective traffic signal modulated on the
carrier signal. The demultiplexer separates the channel signals onto
respective outputs for coupling to the switching matrices, each switching
Is matrix receiving only channel signals at one of the distinct wavelengths.
The multiplexer receives multiple channel signals, each having a different
respective wavelength from respective switching matrices and combines
the multiple channel signals for transmission on a single output fiber. In
this arrangement, every cross-point of every switching matrix is usable,
2o i.e. none of the cross-points route channel signals at a particular
wavelength to a WDM multiplexer port that is unable to pass channel
signals at that wavelength.
The wavelength-converting inter-matrix switch may comprise
multiple optical receivers, multiple optical transmitaers and an electrical
2s switch connected between the optical receivers and the optical
transmitters. The optical receivers are coupled to inter-matrix output
ports of the switching matrices, and are operable to convert channel
signals arriving from the inter-matrix output ports to electrical signals. The
electrical switch is operable to switch electrical signals from any optical
3o receiver to a plurality of the optical transmitters. The optical
transmitters
are operable to convert electrical signals to channel signals having
predetermined wavelengths.
In most practical wavelength-converting inter-matrix switches, the
electrical switch is operable to switch electrical signals from any optical
3s receiver to any or substantially any optical transmitter. The electrical
CA 02320613 2000-09-21
9
switch may be a single electrical switching element or multiple electrical
switching elements connected in series or in parallel.
In this arrangement, the electrical switch is used to couple a
receiver connected to a switching matrix assigned to a first wavelength to
a transmitter operating at a second wavelength and connected to a
switching matrix assigned to the second wavelength, thereby cross-
connecting a channel operating at the first wavelength to a channel
operating at the second wavelength.
Alternatively, the wavelength-converting inter-matrix switch may
to comprise an optical switch, and a plurality of optical transponders
connected to the switch. Each optical transponder is operable to convert
a channel signal having a first wavelength into a channel signal having a
second wavelength. The optical switch is operable to couple a channel
signal arriving from an inter-matrix output port of any switching matrix to
an inter-matrix input port of any of a plurality of other switching matrices
via an optical transponder.
The optical transponder may be a device having a receive half for
recovering an information signal from the incoming wavelength channel,
and a transmit half, having means to modulate the recovered information
2o signal onto a light source of a specific, fixed or tunable, wavelength for
output on a different wavelength channel. The optical switch may
comprise a single optical switching element or multiple optical switching
elements connected in series or in parallel for load sharing.
In most practical wavelength-converting inter-matrix switches, the
optical switch is operable to couple a channel signal arriving from an inter-
matrix output port of any switching matrix to an inter-matrix input port of
any or substantially any other switching matrix.
The optical switch may be coupled between the inter-matrix output
ports and the optical transponders. In this arrangernent, the optical switch
3o is used to couple a first channel operating at a first wavelength to a
selected transponder that converts the signal on the first channel to a
signal at a second wavelength. The transponder is connected to an inter-
matrix input port of the switching matrix that is assigned to the second
wavelength.
CA 02320613 2000-09-21
Alternatively, the optical switch may comprise plural optical
switching stages and the optical transponders may be coupled between
optical switching stages. For example, the optical switch may comprise a
multistage optical CLOS switch. The relatively high insertion loss of a
s multistage optical switch is acceptable in the inter-matrix switch because
the inter-matrix switch includes transponders that restore the optical
signal level as they convert an optical signal at one wavelength to an
optical signal at another wavelength. However attention must be paid to
an overall system loss budget to keep all components operating within
ro their specified range.
Some or all of the optical transponders may be tunable to transmit
channel signals of selectable distinct wavelengths. The use of tunable
transponders reduces the number of transponders that need to be
provided to allow for all possible wavelength conversion possibilities.
Is Each tunable transponder can be provisioned remotely for any of a
number of wavelength channels without requiring <~ visit to the switching
site to physically provision a wavelength channel. It can be
demonstrated statistically that a number of tunable transponders can
provide more combinations of channel configurations than the same
2o number of fixed wavelength transponders. Moreover, the use of tunable
transponders reduces the number of different transponder types that must
be stocked and inventoried.
However, tunable transponders are more expensive than fixed
wavelength transponders and currently have limited tuning range.
2s Consequently, some or all of the transponders may be fixed wavelength
transponders that are operable to transmit channel signals of a single
wavelength. Alternatively the tunable transponders may be arranged in
groups, each group covering the ports associated with a specific
wavelength band.
3o Another aspect of the invention provides an optical switching
matrix comprising first and second pairs of switching elements and a
plurality of optical combiners. Each pair of switching elements comprises
a first switching element and a second switching element. Each
switching element comprises a rectangular substrate having a plurality of
3s input ports on a first side, a first plurality of output ports on a second
side
opposite the first side and a second plurality of output ports on a third
CA 02320613 2000-09-21
11
side adjacent the first side and the second side. Each switching element
further comprises a plurality of optical diverters aligned between each
input port and a corresponding output port on the second side. Each
diverter is aligned with a respective output port on the third side and is
movable from a first position, in which the diverter allows an optical
signal incident from the input port to propagate in a direction toward the
respective output port on the second side, to a second position, in which
the diverter diverts an optical signal incident from the input port toward a
respective output port on the third side. For each of the first and second
to pairs of switching elements, each input port of the second optical
switching element is optically coupled to a respective output port of the
first optical switching matrix. Each combiner is coupled to a respective
output port of the first pair of optical switching elements and to a
respective output port of the second pair of optical switching elements.
Construction of larger switching matrices by assembly of smaller
switching matrices as described above, may be attractive until switching
matrices of the desired port count are readily available at attractive prices.
Moreover, the ability to assemble larger switching matrices from smaller
switching matrices enables modular construction of cross-connect
2o switches so that the size of the switch (and its installed cost) can grow
gracefully with capacity demands.
Accordingly, another aspect of the present invention provides a
plurality of switching matrices, each switching matrix being assignable to
a respective channel wavelength, as well as having multiple input and
output ports and at least one pair of inter-matrix input and output ports.
Additionally, each switching matrix has an expansion port for coupling to
an input port of an extension-switching matrix, which is also assignable to
the respective channel wavelength. In this way, each switching matrix
can be extended, thereby increasing its switching capacity and further
3o increasing the switching capacity of the cross-connect switch that
includes the extended switching matrices. For example, the size of a
switching matrix could originally be 32X32 and an extension switching
matrix of the same size could be coupled to it, via the expansion port, to
result in an extended switching matrix of size 32X64. A cross-connect
switch having a plurality of these extended switching matrices could be
coupled, via optical combiners, to another cross-connect switch having a
similar plurality of extended switching matrices. This would result in a
CA 02320613 2000-09-21
12
combined cross-connect switch with double the switching capacity and
port count of either of the original cross-connect switches.
Another aspect of the invention provides a wavelength-converting switch
for interconnecting optical switching matrices of an optical cross-connect
s switch, the wavelength-converting switch comprising an optical switch
and a plurality of optical transponders connected to the switch. Each
optical transponder is operable to convert a channel signal having a first
wavelength into a channel signal having a second wavelength. The
optical switch is operable to couple a channel signal arriving from an inter-
o matrix output port of any switching matrix to an inter-matrix input port of
any of a plurality of other switching matrices via an optical transponder.
The wavelength-converting switch can be used in the construction
of some embodiments of the cross-connect switch described above.
Another aspect of the invention provides a switching fabric for an
is optical cross-connect switch. The switching fabric comprises a plurality
of optical switching matrices. Each switching matrix has multiple inter-
node input ports and at least one intra-node input port for receiving
incoming optical channel signals, the incoming optical channel signals
having a wavelength that is particular to that particular switching matrix.
2o Each switching matrix also has multiple through output ports and at least
one intra-node output port. Each switching matrix is operable to switch
optical channel signals arriving on any input port to any of a plurality of
the through output ports and the intra-node output port.
In most practical switching fabrics, each switching matrix will
Zs operable to switch optical channel signals arriving on any input port to
any or substantially any of the output ports.
The switching fabric may further comprise an add/drop multiplexes
coupled to the intra-node input port and intra-node output port of each
switching matrix. The add/drop multiplexes is operable to couple, to the
3o intra-node input port of any switching matrix of the plurality of switching
matrices, optical channel signals having the wavelength that is particular
to that switching matrix. The add/drop multiplexes is also operable to
receive, from the intra-node output port of any switching matrix of the
plurality of switching matrices, optical channel signals having the
:a wavelength that is particular to that switching matrix.
CA 02320613 2000-09-21
13
Another aspect of the invention provides a method of cross-
connecting optical channel signals at an optical cross-connect switch
comprising a plurality of switching matrices. The method comprises
coupling each optical channel signal having a particular wavelength to an
input port of a particular switching matrix assigned to that particular
wavelength, and switching the optical channel signal in the particular
switching matrix to an output port selected according to a desired cross-
connection of the optical channel signal.
The optical channel signal may be switched to an intro-node output
~o port of the particular switching matrix when the optical channel signal is
to be cross-connected to an optical channel having a wavelength other
than the particular wavelength of the optical signal. In this case, the
optical signal may be coupled from the intro-node output port to a
wavelength converter for conversion to an optical channel signal having
~5 another wavelength. The optical signal at the other wavelength can be
coupled to an intro-node input port of another switching matrix, the other
switching matrix being assigned to that other wavelength. The other
switching matrix can switch the optical channel signal to an output port
selected according to the desired cross-connection of the optical channel
2o signal.
The optical channel signal may also be switched to an intro-node
output port of the particular switching matrix when the optical channel
signal is to be dropped at the cross-connect switch.
According to another aspect of the invention, the invention
25 provides an optical connection system for optically connecting circuit
cards via a midplane. The interconnect includes a first connector for
connecting a first plurality of optical fibers coupled to a first circuit
card.
The first connector has a first mounting means for mounting the first
connector adjacent an edge of the first circuit card.
3o A second connector is included for connecting a second plurality of
optical fibers coupled to a second circuit card. The second connector has
a second mounting means for mounting the second connector adjacent an
edge of the second circuit card. A first mating insert, disposed in the first
connector, is included for aligning the first plurality of optical fibers in
an
:5 optically coupled relationship with the second plurality of optical fibers.
A
second mating insert, disposed in the second connector, is included for
CA 02320613 2000-09-21
14
aligning the second plurality of optical fibers in an optically coupled
relationship with the first plurality of optical fibers. Finally, an alignment
ferrule is included for mounting in an opening in the midplane. The
alignment ferrule has an aperture for receiving the first mating insert on
one side of the alignment ferrule and the second mating insert on the
other side of the alignment ferrule. The aperture in the alignment ferrule is
oriented to pass through the opening in the midplane when the alignment
ferrule is mounted therein.
The alignment ferrule provides a means to align the mating inserts
o such that the final alignment features of the connectors, in this case a
pair of guide pins with corresponding sockets, can engage and provide the
final alignment of the optical fiber ends at the facE;s of the mating inserts.
The final alignment features provide translational alignment along
orthogonal axis parallel to the faces, as well as rotational alignment about
Is an axis perpendicular to the faces such that the multi-fiber ribbon cables
can be optically aligned.
According to yet another aspect of the present invention there is
provided an optical network comprising at least one optical cross-connect
switch wherein optical fibers couple the optical switching matrices to the
20 optical network via the input and output ports. Alternatively, or
additionally, where the optical cross-connect switch includes the
wavelength division multiplexers and demultiplexers, optical fibers couple
the wavelength division multiplexers and demultiplexers to the optical
network for respectively transmitting and receiving optical output and
25 input signals.
According to still another aspect of the present invention there is
provided a method of upgrading an optical cross-connect switch having a
plurality of switching matrices, each switching matrix assigned to a
respective channel wavelength and having multiple input and output
3o ports, the method comprising the steps of: providing each switching
matrix with an expansion port; providing a plurality of extension switching
matrices, each extension switching matrix having multiple input and
output ports; and coupling a respective extension switching matrix to
each switching matrix, via the expansion port and at least one of the
35 input ports of the respective extension switching rnatrix, to form a
plurality of expanded switching matrices.
CA 02320613 2000-09-21
Additionally, the optical cross-connect switch may be upgraded
further by providing another similarly upgraded optical cross-connect
switch having a plurality of the expanded switching matrices; and
coupling each output port of an expanded switching matrix of the optical
5 cross-connect switch to a respective output port of an expanded
switching matrix of the another optical cross-connect switch.
Other aspects of the invention comprise combinations and
subcombinations of the features described above other than the
to combinations described above.
Brief Description of the Drawings
Embodiments of the invention are described below, by way of
example only, with reference to the drawings in which:
is Fig. 1 a is a diagram of a prior art optical network;
Fig. 1 b is a diagram showing the nodes A and B of Fig. 1 a in
greater detail;
Fig. 1 c is a diagram showing the transponders and
regenerators of Fig. 1 b in greater detail;
Zo Fig. 2a is a diagram of an optical network in accordance with
an embodiment of the present invention;
Fig. 2b is a diagram showing the nodes A' and B' of Fig. 2a in
greater detail;
Fig. 3 is a functional block diagram of a cross-connect switch
2s in accordance with an embodiment of the present invention;
Fig. 4a is a functional block diagram of an embodiment of the
wavelength-converting switch shown in Fig. 3;
Fig. 4b depicts an embodiment of a physical arrangement for
the wavelength-converting switch of Fig. 4a;
CA 02320613 2000-09-21
16
Fig. 4c is a functional block diagram of components of the
electrical switch in Fig. 4b;
Fig. 4d is a functional block diagram, which provides further
detail on the wavelength-converting switch of Fig. 4b;
s Fig. 4e depicts another embodiment of a physical arrangement
for the wavelength-converting switch of Fig. 4a;
Fig. 5 is a functional block diagram of an embodiment of the
receiver transponder of Fig. 4;
Fig. 6 is a functional block diagram of an embodiment of the
1o transmitter transponder of Fig. 4;
Fig. 7 is a pictorial diagram of part of the optical switching
matrix of Fig. 3;
Fig. 8 is a functional block diagram of an embodiment of the
optical switching element of Fig. 3;
Is:,. Fig. 9a is a functional block diagram illustrating a second
embodiment of the optical switching matrix of Fig. 3;
Fig. 9b is a functional block diagram of a third embodiment of
the optical switching matrix of Fig. 3;
Fig. 9c is a functional block diagram of a fourth embodiment
20 of the optical switching matrix of Fig. 3;
Fig. 9d is a functional block diagram of a fifth embodiment of
the optical switching matrix of Fig. 3;
Fig. 9e is a functional block diagram of a sixth embodiment of
the optical switching matrix of Fig. 3;
2s Fig. 9f is a functional block diagram of a seventh embodiment
of the optical switching matrix of Fig. 3;
Fig. 9g is a functional block diagram of a eighth embodiment
of the optical switching matrix of Fig. 3;
CA 02320613 2000-09-21
17
Fig. 10 is a functional block diagram of a second embodiment
of the wavelength-converting switch shown in Fig. 3;
Fig. 1 1 is a functional block diagram of a third embodiment of
the wavelength-converting switch shown in Fig. 3;
Fig. 12 is a functional block diagram of an embodiment of the
converter module shown in Fig. 1 1;
Fig. 1 3 is a table of connections made by the interconnects A
and B in Fig. 1 1 ;
Fig. 14 is a table of connections made by the interconnect C in
to Fig. 1 1;
Fig. 1 5 is a table of connections made by the interconnect D
in Fig. 1 1;
Fig. 16a is a perspective view of a physical arrangement of the
cross-connect switch of Fig. 3, which includes the wavelength-converting
is switch of Fig. 4e;
Fig. 16b is a perspective view of another physical arrangement
of the cross-connect switch of Fig. 3, which includes the wavelength-
converting switch of Fig. 4e;
Fig. 17 is a perspective view of the optical connectors in Fig.
?0 16;
Fig. 18a is a line drawing plan view of the optical connectors
of Fig. 17 showing the connectors in a nearly connected position;
Fig. 18b is a side view of a portion of the connectors of Fig.
18a showing fiber polished at the face of each connector mating insert
2s and alignment pins and sockets;
Fig. 18c is a cross-sectional top view of the connectors and
alignment ferrule of Fig. 17 taken along the line AA in Fig. 18b;
Fig. 18d is a cross-sectional front view of the mating face of
the connector taken along the line BB in Fig. 18c;
CA 02320613 2000-09-21
18
Fig. 19 is a line drawing plan view of a second embodiment of
the optical connectors in Fig. 16;
Fig. 20 is a line drawing plan view of a third embodiment of
the optical connectors in Fig. 16;
s Fig. 21 is a perspective view of a switching shelf, which is a
portion of the cross-connect switch of Fig. 3 in a second embodiment;
and
Fig. 22 is a diagram of a fiber shuffle used in the switching
shelf of Fig. 21 .
0
Detailed Description
Referring to Figs. 1 a to 1 c a prior art optical network 1 will now be
described. In Fig. 1 a a network 1 includes six interconnected electrical or
is .. opaque switching nodes 2, labelled A to F. In Fig. 1 b nodes A and B are
shown in greater detail. For simplicity, a unidirectional representation of a
bi-directional network has been shown. In practice all links connecting the
nodes A to F would have companion links connected in inverse parallel to
carry return traffic, or the connecting links would be bi-directional links.
2o The network 1 includes electrical cross-connect switches 2
interconnected by spans comprising optical fibers and optical amplifiers 7,
which are spaced apart at appropriate intervals along the spans.
Alternatively, so-called "opaque" optical cross-connect switches could be
used in place of some or all of the electrical cross-connect switches 2. An
2s opaque switch is one that uses transponders between the links
connecting the switch to a network, such that the wavelength at which a
signal from the network is switched is independent from the wavelength
at which the signal is carried over the network. Each transponder is of
one of two forms, those being a receive transponder (Tr) and a transmit
3o transponder (Tt). The receive transponder (Tr) consists of a long-reach
line
wavelength receiver and a short-reach transmitter, which is usually a
single fixed-wavelength optical, or electrical, cross-office short-reach
transmitter. The transmit transponder (Tt) consists of a short-reach single
fixed-wavelength optical, or electrical, cross-office receiver and a long
CA 02320613 2000-09-21
19
haul optical transmitter working at the final line wavelength, either by
equipping the unit with the appropriate wavelength laser or by exploiting
tunable lasers. A bi-directional transponder (not shown here) is a
commonly packaged transmitter and receiver transponder with distinct
"line" and "office" sides. A regenerator (R) can also be made from
transponders, by placing them in series, such that: the short reach cross
office transmitter of the receiver transponder (Tr) directly drives the short-
reach, cross-office, receive port of the transmitter' transponder (Tt). A bi-
directional regenerator is two of these combinations of receiver
o transponder (Tr) and transmitter transponder (Tt) in an inverse parallel
configuration. The electrical cross-connect switches 2 each comprise
optical wavelength division (WD) demultiplexers 4 coupled to an electrical
switch fabric 2 via receive transponders (Tr) on the ingress side of the
switch 2. The receive transponders (Tr) convert demultiplexed line optical
is channel signals to electrical, or short-range optical signals, which are
fed
to the interfaces of, and are switched by the electrical switch fabric 2, or
opaque optical switch fabric. An optical pre-amplifier 7b may be coupled
to the input of a WD demultiplexer 4 to amplify received DWDM signals
before switching. At the egress side of the switch 2, optical WD
20' multiplexers 5 are coupled to the electrical switch fabric 2 via more
transmit transponders (Tt). Switched electrical signals are converted to
optical channel signals by the transmit transponders (Tt) at the egress
side, and the WD multiplexers 5 multiplex the optical channel signals into
DWDM signals, which are output by the switch 2. Electrical input signals
~5 to the optical network 1 are converted to optical signals by transmit
transponders (Tt) and multiplexed into a DWDM signal by WD multiplexer
5b. Conversely, DWDM signals are demultiplexed by the WD multiplexers
4b into optical channel signals, which are converted by the receive
transponders (Tr) and output from the network 1 as electrical signals.
o It is apparent that a pair of transponders is required for each
channel signal passing through the switch 2. Further, additional
transponders (T) are required to add or drop channel signals from the
switch 2. Still further, repeaters 3 comprising WD demultiplexers 4
coupled to WD multiplexers 5, via regenerators (R1, require an additional
pair of transponders per channel signal. These transponders (not shown)
are used in the regenerators (R) to perform O/E conversion of
demultiplexed channel signals before electrical regeneration, and E/O
conversion of regenerated electrical signals into optical channel signals,
CA 02320613 2000-09-21
which are then multiplexed into a regenerated DWDM signal. Thus a
regenerator (R) includes back-to-back transmitter and receiver
transponders and may also include reshaping/retiming functionality. With
such a network 1, any increases in the number of channel signals in a
s DWDM signal requires an additional pair of transponders in every switch 2
and repeater 3 in the network 1.
Extensive work is currently underway throughout the optical
communications industry to develop technology that will reduce the rate
at which an optical signal degrades with transmission distance. This work
o is being done to achieve ever longer amplified span lengths between
regeneration points, in order to eliminate the need for regenerators in all
but a few cases. The result will be requirement for a network
configuration that can exploit the resultant technology effectively. That is,
for an optical signal to travel long distances between overall network ends
is in an all photonic network, it must be possible to add, drop, and switch
traffic from intermediate nodes without reverting 1:o electrical (or opaque
optical) switching.
Fig. 2a shows an optical network 8 in accordance with an
embodiment of the present invention. The network 8 includes six
2o interconnected switching nodes A' to F' which are optical cross-connect
switches 10. The optical cross-connect switches '10 are in accordance
with another embodiment of the present invention and will be described in
more detail later. Fig. 2b shows the nodes A' and B' in greater detail. For
simplicity, a unidirectional representation of a bi-directional network has
2s been shown. In practice all links connecting the nodes A' to F' would
have companion links connected in inverse parallel to carry return traffic,
or the connecting links would be bi-directional links. The optical cross-
connect switches 10 are interconnected by spans comprising optical
fibers and optical amplifiers 7. The optical cross-connect switches 10
3o include photonic cross-connects 9 coupled to the spans via WD
demultiplexers 4 and optical pre-amplifiers 7b on the ingress side of the
switch 10, and WD multiplexers 5 on the egress side of the switch 10.
The photonic cross-connects 9 further include multiple optical switching
matrices and a wavelength-converting inter-matrix switch (not shown) as
will be described later with reference to Fig. 3. The wavelength-
converting inter-matrix switch converts a channel signal from one channel
wavelength to another channel wavelength as required, for example, by
CA 02320613 2000-09-21
21
span limitations in terms of available channel wavelengths. Transponders
are provided for this purpose. However, they are far fewer in number than
a pair per channel signal. Typically, transponders are provided for 25 % of
the channel signals that can be switched by the switch 10. This
percentage is determined by network engineering rules which will be set
up to favor finding end-to-end clear wavelengths or paths with minimum
wavelength conversion (i.e. lambda-hopping). Additional transponders for
adding channel signals to, or dropping channel signals from, the switch
would not normally be required since the transponders included in the
to wavelength-converting inter-matrix switch can also be used for this
purpose. Clearly, this optical network 8 requires fewer transponders than
the prior art optical network 1, the actual reduction being dependent upon
the network planning algorithms. This reduction in transponders leads to
savings in costs and power requirements for a given network
~s configuration, and as the configuration grows in switching nodes and
channel signals per DWDM signal.
In the prior art approach of Fig. 1, transponders are required for
every added, dropped and switched through wavelength irrespective of
reach or distance. In embodiments of the present invention, for example
2o as illustrated in Fig. 2, transponders are only needed to enter/leave the
optical domain, or because the system reach (i.e. the maximum allowable
distance between transponders) is too small for implementing a given
route, or because wavelength conversion is necessary to get around a
"blocked" wavelength.
25 Referring to Fig. 3 there is illustrated a cross-connect switch 10 in
accordance with an embodiment of the present invention. The cross-
connect switch 10 includes an input port 12 for receiving an optical signal
s, for example a DWDM optical signal from an optical telecommunications
network. The input port 12 is connected to an optical amplifier 14 via an
30 optical fiber. Unless stated otherwise, all connections internal to the
switch are made by way of optical fiber, which may or may not be
assembled into ribbon cables with multiple fibers and associated multi-
way connectors. The amplifier 14 amplifies the optical signal s, which
might, for example, be received from a fiber cable from the previous line
3s amplifier to compensate for the insertion loss of the span before it is
applied to a wavelength division (WD) demultiplexc:r 16. The
demultiplexer 1 6 divides the optical signal s into ita constituent channel
CA 02320613 2000-09-21
22
wavelengths. Each channel has a predetermined wavelength, 7~1 to ~,M
assigned to it. In the embodiment of Fig. 3 there are up to 160 such
channels. A respective P x P optical switching matrix 18 is provided for
each set of channels that have a common predetermined wavelength, one
s such channel coming from each of the WD demultiplexers 16. Fig. 3
shows M such optical switching matrices of which there are up to 160 in
the present embodiment (i.e. M = 160), since there are up to 160
channels on each inter-node long haul transmission fiber. At each channel
wavelength, light from the optical signal s at the channel wavelength is
to input into the respective optical switching matrix '18 for that channel. A
wavelength division (WD) multiplexer 20 aggregates a switched channel
from each one of the optical switching matrices into another optical signal
s' for outputting from an output port 24. Each output port 24 is
connected to the WD multiplexer 20 through an optical amplifier 22. The
is optical amplifier 22 amplifies the optical signal s' to compensate for the
insertion loss through the switch 10 before the optical signal s' is output
from the switch 10 into an optical telecommunications network, for
example.
Note that the cross-connect switch 10, in addition to performing
20 optical switch functions, also restores the level of the optical signals
for
transmission to the next cross-connect switch 10 or destination node.
Consequently, the cross-connect switch, as shown in its entirety in Fig.
3, replaces the entire WDM/WDD-transponder-cross-connect path that
would be required in an electrically switched and regenerated transmission
?5 network.
The switch 10 has a plurality of input ports 12 and respective
optical amplifiers 14 and WD demultiplexers 16 as well as a plurality of
output ports 24 and respective optical amplifiers 22 and WD multiplexers
20. Fig. 3 shows N input ports 12, each of which has an accompanying
30 optical amplifier 14 and WD demultiplexer 16. Fig. 3 also shows N
output ports 24, each of which having an accompanying optical amplifier
22 and WD multiplexer 20. In the present embodiment there are up to 24
input ports and 24 output ports, that is N=24. Hawever, expansion of
the switch 10 to provide a greater number of input and output ports is
35 possible, and will be described later. Also possible, are configurations in
which the number of input ports does not equal the number of output
ports. For example, rectangular (e.g. 16 x 32) optical switching matrices
CA 02320613 2000-09-21
23
18 could be used to map a reduced subset of transponders to a
provisionable subset of ports within the wavelength group of those
transponders.
An inter-matrix switch in the form of a wavelength-
s converting switch 28 with additional add/drop multiplexer capabilities is
connected across each of the optical switching matrices 18. That is, for
each P x P optical switching matrix 18, a number (K) of outputs of the
wavelength-converting switch 28 are connected individually to the same
number (K) of inter-matrix inputs of that optical switching matrix 18. As
o well, for each optical switching matrix 18, a number (K) of inputs of the
wavelength-converting switch 28 are connected individually to the same
number (K) of inter-matrix outputs of that optical switching matrix 18. In
the present embodiment, the number K is a variable over the range 0-16,
covering the extreme cases of all wavelengths needing conversion or
~s access to add-drop (K = 16) or no wavelengths neE:ding conversion or
access to add-drop (K=0). The particular value of K in any particular case
would be dependent on the location of the particular optical cross connect
switch in the network and details of the network engineering algorithms.
Typically, a practical value of K is K = 8, (i.e. 25°~0 of optical
switching
2o matrix 18 inputs/outputs, thereby permitting 33% of the remaining 24
inputs and outputs to be connected to the wavelength conversion/ add-
drop inter-matrix switch.) That is, in this embodiment of the switch 10,
the value of P = K + N. Other variations in the values of K, N, and P are
possible and would need to be planned for in conjunction with the
?a network engineering algorithms. The wavelength-converting switch 28
also has a capability of converting wavelengths. That is, it can receive
information on one wavelength and transmit the same information on a
different wavelength. This capability is useful for switching information
between channels as described further below.
3o It should be noted that the wavelength-converting switch 28 can
add/drop channel signals without performing wavelength conversion on
the added/dropped channels signals, hence it can function solely as an
add/drop multiplexer. Conversely, the wavelength-converting switch can
perform wavelength-conversion without performing an add/drop function;
;~ hence it can function solely as a wavelength-converting switch.
Moreover, the wavelength-converting switch 28 can perform both a
wavelength conversion function and an add function on the same channel
CA 02320613 2000-09-21
24
signal, and separate functions (i.e. add, drop, convert wavelength? on
different signals at the same time, as will be described later.
The switch 10 also includes a controller 26 for controlling each of
the optical switching matrices 18, the wavelength-converting switch 28
s as well as any tunable transponders or sources associated with the
wavelength-converting switch 28. For example, the controller 26 can set
up the optical switching matrix 18, assigned to channel one, to switch
light from the output of the WD demultiplexer 16,, connected to the
second input port 12, to the input of the WD multiplexer 20, connected
to to the first output port 24. The controller 26 is connected to each optical
switching matrix 18 via electrical cable, and controls each optical
switching matrix 18 using electrical control signals. The control signals
and link over which they are transmitted could also be optical in nature,
although the control of the crosspoint would likely remain electrical in
t5 nature. The control signals are generated by a real-time processor (not
shown) of the switch 10 which configures crosspoints of the switch 10 in
a manner similar to that used in an equivalent electrical switch operating
under Element Manager control from a central Network Manager.
Alternatively, the Element Manager may receive control signals from
2o configuration controllers distributed among network switching nodes.
An important design consideration of the switch 10 is balancing the
power gain/loss in the "through" path of the switch 10 with that in the
"wavelength conversion" path. The "through" path, or link path, is any
path through only one optical switching matrix 18, from amplifier 14 to
z5 amplifier 22. In such a path there is no optical regeneration and any
losses must be within the optical link budget allotted to the switch 10,
within the overall end-to-end optical link budget. In the wavelength
conversion path, i.e. any path through two or more optical matrices 18
and the wavelength-converting switch 28, there is typically optical
3o regeneration performed by transponders in the wavelength-converting
switch 28. It is important that this path, also between the optical
amplifiers 14 and 22 have a power loss/gain in the same range as the
through path. By adjusting the power levels of the transponders in the
wavelength-converting switch 28 the power loss/gain difference between
3s the two types of paths can be balanced.
CA 02320613 2000-09-21
In operation, the switch 10 is capable of three modes of switching,
they are port switching, channel switching, and switching that is a
combination of channel and port switching. The operation of the switch
10 in each of these three modes will be described further by way of
s example.
In the port switching mode, an optical signal s arrives at input port
one and is split into its constituent channels (1 to 160) by the WD
demultiplexer 16 assigned to that port. The controller 26 has set up the
optical switching matrix 18, for channel wavelength one, to switch
~o optical signals from its input from port number one to its output for port
number two. This causes light of wavelength ~.1 from the output of the
WD demultiplexer 16 assigned to channel wavelength one to be directed
to the input of the WD multiplexer 20 assigned to the output port two.
This light is aggregated with light from the other channels by the WD
~s multiplexer 20 into the signal s', which is output from the output port
two. Thus, information received by the switch 10 on channel one input
port one is switched to channel one output port two, and is outputted by
the switch 10.
In the channel-switching mode, each channel of the optical signal s
2o arrives at its respective optical switching matrix 18 from the WD
demultiplexer 16 as before. However, in this case the controller 26 has
set up the optical switching matrix 18 for channel two to switch its input
for port one to one of its outputs connected to the wavelength-converting
switch 28. For example, the optical switching matrix for channel two has
's been configured to switch its input for port one to the first input of the
wavelength-converting switch 28. Recall, that in the present embodiment
the wavelength-converting switch 28 has eight inputs and eight outputs
(K=8) connected to each optical switching matrix 18. The wavelength-
converting switch 28 is also configured by the controller 26 and
3o connected thereto by electrical or optical links (not shown). In this
example, the wavelength-converting switch 28 is configured to receive
information on its first input for channel two 7~2 and output the
information on channel three ~,3 at its first output port for channel three.
This optical switching matrix 18 is set up to direct the light from this
signal to its output connected to the WD multiplexer 20 for port one.
Consequently, information received by the switch '10 on input port one
channel two is output on output port one channel three. Thus, the switch
CA 02320613 2000-09-21
26
has performed channel switching, from channel two input port one to
channel three output port one.
In switching that is a combination of port and channel switching,
hereinafter referred to as port-channel switching, information arrives at
the switch 10 on a particular input port number, carried by a particular
channel wavelength, and leaves on another output port number, carried
by a different channel wavelength. The operation of port-channel
switching is the almost the same as channel switching except that in the
last switching step the signal is switched to another output port number.
to For example, in the previous example of channel switching, instead of
switching the channel signal back to output port one (on channel three) it
would be switched to any of the other output ports (e.g. output port
four).
The configuration control strategy used to control cross-connect
is switches 10 will favour port switching at the cross-connect switches 10
in preference to channel switching and port-channel switching. Channel
switching and port-channel switching will generally be used only when no
single wavelength channel is available from a source node to a destination
node. The need for channel switching and port switching can be reduced
2o by over-provisioning wavelength channels. Such over-provisioning has
less cost impact in an optically switched network than in an electrically
switched network since little of the required equipment is wavelength
specific.
A further capability of the switch 10 is the ability to add or drop
~a traffic using the add/drop multiplexer functionality of the wavelength-
converting switch 28. The wavelength-converting switch 28 has R add
inputs for adding traffic and also R drop outputs for dropping traffic. In
the present embodiment R ranges from 480 to 960, corresponding to a
20% traffic add/drop on a half to fully configured switch, depending on
3o the number of wavelength channels that are provisioned on the cross-
connect switch 10.
In Fig. 3, a signal sA, which is to be added to the traffic flow
processed by the switch, is shown being input to the wavelength-
converting switch 28. This signal sA could go through wavelength
_ _ conversion if necessary, as described above, before being output by the
wavelength-converting switch 28 into one of the optical switching
CA 02320613 2000-09-21
27
matrices 18. The signal sA is then output to one of the WD multiplexers
18 for aggregation into an optical signal, for example s', to be transmitted
from a corresponding output port 24. Also referring to Fig. 3, a signal sD,
which is to be dropped from the traffic flow processed by the switch, is
s shown being output from the wavelength-converting switch 28. This
signal sD, could also go through wavelength conversion if necessary,
before being output by the wavelength-converting switch 28 into other
optical communications equipment (not shown).
Fig. 4a illustrates, in a functional block diagram, an embodiment of
o the wavelength-converting switch 28 shown in Fig. 3. The wavelength-
converting switch 28 includes an electrical switch 30. A plurality of
transponder receiver sections for converting line optical signals to
electrical signals or short reach optical signals, as required by the
electrical switch, are connected to the inputs of the electrical switch 30.
is Specifically, M groups of K receiver transponders 32 for converting
optical signals from the optical switching matrices 18 are connected to
the electrical switch 30. As well, receiver transponders 38 for converting
optical signals (e.g. the signal sA) to be "added" to the traffic flow of the
transport system via switch 10 are also connected to inputs of the
2o electrical switch 30. In the present embodiment K =8 and M =160 and R
= 960, giving a cross-connect size of [(160 x 24) + 960] x [(160 x 24)
+ 960] = 4800 x 4800. Such a cross-connect may be implemented as a
single switch or as several (e.g. eight parallel planes of smaller (600 x
600)) switches, exploiting the lateral interplane cross-connection
25 inherently available in the optical switch matrices 18 to minimize any
resultant wavelength blocking. In addition, a plurality of transmitter
transponders for converting electrical signals to optical signals is
connected to the outputs of the electrical switch :30. That is, M groups of
K transmitter transponders 34 for converting electrical signals for the
30 optical switching matrices 18 are connected to the outputs of the
electrical switch 30. As well, transmitter transponders 36 for converting
electrical signals of dropped traffic into local cross-office optical signals
are also shown connected to outputs of the electrical switch 30.
It should be noted, that the transponders 36, 38 are optional and
3s would be required if optical signals (e.g. the signal sA) are to be added
to
the traffic flow or if dropped signals (e.g. the signal sD) are to be optical.
Further, it should be noted that the number of transponders connected to
CA 02320613 2000-09-21
28
the inputs of an optical switching matrix 18 does not have to be equal to
the number connected to the outputs of the same optical switching
matrix. Still further, the receiver transponders 32 need not be very
sensitive since they are receiving light that has been amplified by the
s optical amplifier 14 and then only attenuated about 5-10 dB by a WD
demultiplexer 16 and an optical switching matrix 18. However, the
wavelength-accurate transmitter transponders 34 are usually expensive,
due to their precision optical sources and the number of versions required
(i.e. one for each wavelength in the case of fixed transponders or one for
to each wavelength band in the case of tunable transponders, shown with a
control signal from the controller 26 to set the transmission wavelength
of the transponder). Hence, more of the receiver transponders 32 than the
transmitter transponders 34 may be provisioned to optimize the
wavelength conversion capability at the lowest coat.
is Operation of the wavelength-converting switch 28 will now be
explained by way of example. The receiver transponder 32 receives an
optical signal Sc 1 , on channel one (i.e. channel wavelength ~.1 ) from a
optical switching matrix 18 assigned to channel one and converts the
information in this signal to an electrical signal Ec1, which is input to the
2o electrical switch 30. The electrical switch 30 has a switching granularity
of the entire signal payload of each wavelength channel. The electrical
switch 30 switches the electrical signal Ec1 to one of its outputs assigned
to channel fifty. The transmitter transponder 34 receives the signal Ec1
and converts the information carried by it to an optical signal Sc50 having
2s a wavelength corresponding to channel fifty. This signal is output to the
optical switching matrix 18 that is assigned to channel fifty, which directs
it to a WD multiplexer 20, as described earlier. Thus, since the
information in the optical signal Sc1 has been switched to the optical
signal Sc50, a channel switching function, or wavE:length conversion
3o function (i.e. from channel wavelength ~,1 to channel wavelength x.50)
has been performed by the wavelength-converting switch 28. In a similar
manner the signal sA is directed into the electrical switch 30 via the
receiver transponder 38 and is forwarded into the one of the optical
switching matrices 18 via one of the transmitter transponders 34.
35 Likewise, the signal sD is directed out of the switch 10 via the receiver
transponder 32 and into the electrical switch 30, vvhere it is forwarded
into other communications equipment via the transmitter transponder 36.
CA 02320613 2000-09-21
29
With reference to Fig. 4b an embodimE:nt of a physical
arrangement for the wavelength-converting switch 28 of Fig. 4a will now
be described (details of the physical arrangement of the cross-connect
switch 10 will be described later with reference to Figs. 16a,b and Fig.
21 ). Fig. 4b shows the wavelength converting switch 28 connected to M
P x P optical switching matrices 18; one switching matrix 18 for each
distinct wavelength (i.e. M = 1601. The wavelength-converting switch 28
is physically implemented on four circuit cards C1 to C4, although it could
well be implemented on more, or fewer, cards as will be understood from
to the following description of this implementation. Each of the circuit cards
C1 to C4 includes a respective portion 30a to 30d, one forth in this case,
of the electrical switch 30. Each portion 30a to 30d is electrically
connected to a transmit bank 33 of the transmitter transponders 34, and
a receive bank 35 of the receiver transponders 32 on its card C1 to C4.
~5 Each of the banks 33, 35 is optically coupled to each of the optical
switching matrices 18 by a respective optical connection of width K/4.
Hence there are M~K/4 optical connections from the plurality of M optical
switching matrices 18 to each of the banks 33,3Fi. Add and Drop optical
connections of width R/4 are also provided to the receive bank 35 and
20 transmit bank 33, respectively. Each of the portions 30a to 30d is
electrically interconnected to each of the other portions 30a to 30d via a
high speed inter-card bus 31 of width K*~M+R. The details of this
interconnection will be described with reference to Fig. 4d.
In operation, optical signals from the switching matrices 18, or
2s from the add connections, are received by the receive banks 33 of the
cards C1 to C4, and are converted to electrical signals by receiver
transponders 32,36 in the receive bank 33 of the respective card C1 to
C4. The converted electrical signals are transmitted to the respective
electrical switch portion 30a to 30d on that card C1 to C4. The signals
3o are then either switched to the transmit bank 33 on the same card or to
the inter-card bus 31 where they are input to the electrical switch
portions 30a to 30d on the other cards. Signals switched to the other
cards can then be selected by the respective electrical switch portion 30a
to 30d on the other cards and switched to the transmit bank 33 of that
3s card. Signals switched to the transmit banks 33 are converted to optical
signals of an appropriate channel wavelength and transmitted to the
optical switching matrix 18 for that wavelength.
CA 02320613 2000-09-21
Fig. 4c is a functional block diagram of components of the
electrical switch in Fig. 4b. A Q x Q electrical switching fabric F1 has Q
inputs fully interconnected to Q outputs. That is, an electrical signal
arriving on any one of the Q inputs can be switched to any one of the Q
s outputs. The dimension Q equals (K~M+R)/4. Another Q x Q electrical
switching fabric F2 has Q inputs fully interconnected to Q outputs and Q
expansion outputs, as well as Q expansion inputs fully interconnected
with the Q outputs. The fabrics F1 and F2 can be interconnected into a
larger electrical switching fabric F3 by serially connecting three F2 fabrics.
to This is done by connecting the outputs of one fabric F2 to the expansion
inputs of the next fabric F2 and repeated until all three F2 fabrics are
serially connected. Next an F1 fabric is serially connected to the front of
the chain of F2 fabrics by connecting the outputs of the F1 fabric to the
expansion inputs of the first F2 fabric. The resulting fabric F3 has four
is sets of Q inputs, one set of Q outputs, and four sets of Q expansion
outputs although only the set of expansion outputs on the last F2 fabric is
used.
The operation of the fabric F3 is as follows. Any input of the
first three sets of Q inputs can be switched to any of the Q outputs of the
20 last F2 fabric. Additionally, any input of the last set of Q inputs can be
switched to any of the Q outputs of the last F2 fabric, or any of its Q
expansion outputs.
With reference to Fig. 4d further detail on the wavelength-
converting switch of Fig. 4b will now be provided. Each electrical switch
?s portion 30a to 30d is comprised of the fabric F3 on a respective circuit
card C1 to C4. The high-speed inter-card bus 31 is comprised of four
buses 31 a to 31 d of width Q. Each of the four buses 31 a to 31 d is
driven by the expansion outputs of a respective fabric F3 on one of the
cards C1 to C4, and is connected to the inputs of the fabrics F3 on the
3o remaining cards C1 to C4. Each of the buses 31 a to 31 d could in fact be
comprised of three individual interconnects, of width Q, for example high-
speed electrical interconnect or intra-system short reach optical
connections. In this case, each of the three individual interconnects in a
bus 31 a to 31 d would be point-to-point connection driven by one
3s switching fabric F3 on one card C1 to C4 and received by only one other
fabric F3 on another card C1 to C4.
CA 02320613 2000-09-21
31
The switching operation of the wavelength-converting switch
will now be described by way of example with reference to card C1 . The
electrical switch portion 30a on card C1 can receive electrical signals
from the receive bank 35 or from any of the buses 31 b to 31 d. Received
s electrical signals are either switched to the transmit bank 35 or to the bus
31 a connected to the expansion outputs of the electrical switch portion
30a. Electrical signals switched to the bus 31 a can be received by any of
the other electrical switch portions 30b to 30d and switched to their
respective transmit bank 35.
to Fig. 4e depicts another physical arrangement for the wavelength-
converting switch 28 of Fig. 4a. In this arrangement the electrical switch
30 is no longer partitioned between the circuit cards C1 to C4, but is
implemented as one electrical switch 30 residing on a circuit card, or
cards, which is physically parallel to the optical switching matrices 18.
is Each of the circuit cards C1 to C4 has a respective interface 37a to 37d,
which interfaces the transmit bank 33 and receive bank 35 of the card to
the electrical switch 30, either electrically or by short-reach optical
connections 39a, 39b of width Q (where Q=(K~M+R)/4). The remainder
of the topology and function of the circuit cards C'1 to C4 is as described
2o earlier with reference to Fig. 4b.
Fig. 5 illustrates, in a functional block diagram, an embodiment of
the receiver transponders 32, 38 of Fig. 4, both of which are identical in
structure. However, this need not be the case. The add-drop transponders
36, 38 may not need to be as high precision devicEa (i.e. high sensitivity
2s receiver, precise wavelength transmitter) as the transponders linked to the
switch matrices 18 unless they are going into another line system
directly. If they are feeding a Terabit router they may well be short reach
optics, for example, 1310 nm or 850 nm ribbon optics. The receiver
transponder 32 includes a long range receiver 32a connected to its input
3o for receiving an optical signal. The long range receiver 32a has enough
sensitivity to receive and detect data on optical signals that are at the
minimum specified power level and signal-noise ratio of the optical
communications network in which the switch 10 is used. A local interface
32b is connected at the output of the receiver transponder 32 and is in
3s communication with the long-range receiver 32a. The local interface 32b
receives data from the long-range receiver 32a that it has detected and
outputs this information in an electrical signal.
CA 02320613 2000-09-21
32
Fig. 6 illustrates, in a functional block diagram, an
embodiment of the transmitter transponders 34, 36 of Fig. 4, both of
which are identical in structure. The transmitter transponder 34 includes a
local interface 34a, connected to its input, for receiving an electrical
s signal and detecting data contained therein. A long reach transmitter 34b
of high precision is connected at the output of the transmitter transponder
34 and is in communication with the local interface 34a. The long reach
transmitter 34b receives the detected data from the local interface 34a
and outputs this information in an optical signal.
o Fig. 7 illustrates, in a pictorial diagram, an embodiment of the
switching matrix 18 of Fig. 3. The switching matrix 18 has P inputs and
P outputs. A subset K of the inputs are intra-node inputs and are for
receiving added or converted signals from the wavelength converting
switch 28, which provides both add/drop and wavelength conversion
Is capabilities as discussed previously. The remaining N inputs (i.e. N = P-K)
are inter-node inputs for receiving channel signals 'from other nodes.
Similarly, a subset K of the outputs are intra-node outputs and are for
transmitting signals to the wavelength-converting switch 28 that are to
be dropped or wavelength converted. The remaining N outputs are inter-
2o node outputs for outputting channel signal destined for other nodes.
Optionally, the switching matrix 18 has an expansion input port with P
inputs and/or an expansion output port with P outputs. These expansion
ports can be used to expand the size of the switching matrix 18 and/or
for interconnection with the wavelength-converting switch 28, as will be
2s described later with reference to Figs. 9a-9g.
A switching element 19 is shown in the form of a Micro-
Electro-Mechanical System (MEMS)-based switching element. A MEM
switching device is disclosed in an OFC99 paper entitled "Free-space
Micromachined Optical-Switching Technologies and Architectures", by Lih
3o Y. Lin of AT&T Labs-Research, and published in OFC99 Session W14-1 ,
Feb. 24, 1999 proceedings. The MEMs-based switching element 19
comprises optical diverters 48, 50 arranged in rows and columns to direct
light from an input on the perimeter of the arrangement of optical
diverters to an output also on the perimeter of the arrangement. The
35 MEMS-based switching element 19 has row outputs, which are in
alignment with the inputs and are on the opposite side of the arrangement
of optical diverters 48, 50 relative to the inputs. The MEMS-based
CA 02320613 2000-09-21
33
switching element 19 also has column outputs situated along paths at
right angles with paths between the inputs and row outputs. A self-
focusing collimating lens 52 at each input of the MEMS-based switching
element 19 directs light received from an optical fiber 54 into the
arrangement of optical diverters. At each row and column output another
self-collimating lens 56 and 56', respectively, receives light from the
arrangement and directs the light along a respective fiber 58 and 58'. The
controller 26 controls the state of each of the optical diverters, through a
mirror drive signal, in order to direct the light as required. Fig. 7 shows an
0 optical diverter 48, or mirror, in an activated state,. whereby, an optical
signal Sc1 entering the arrangement of optical diverters along a row is
redirected along a column to the self-collimating lens 56' at the respective
column output of the MEMS-based switching element 19. The other
optical diverters in the figure are shown in a non-activated state, for
example, optical diverter 50, whereby an optical signal Sc2 is not
redirected. The optical signal Sc2 passes through the arrangement and
enters the collimating lens 56 where it is passed along the fiber 58. The
switching matrix 18 is a self-contained switch circuit pack, providing all
the switching interconnect needs of all the ports and all the inter-matrix
feeds for one wavelength. It achieves this by incorporating, as part of its
functionality, an optical crosspoint array (i.e. using one or more MEMS-
based switching elements). One or more complete switching matrix can
be accommodated on a physical circuit pack.
Fig. 8 illustrates, in a functional block diagram, the MEMS-
zs based switching element 19. The MEMS-based switching element 19
could be an 8x8, 16x16, or a 32x32 array, and in this case it is shown as
a 32 x 32 array. MEMS switching devices are commercially available
components manufactured using silicon microelectronic processing.
MEMS switching devices can be "square" (i.e. the number of inputs
3o equals the number of outputs), resulting in an n x n array where "n"
conventionally equals 8, 16, 32, etc. MEMs switching devices can also be
"rectangular" (i.e. the number of inputs is not equal to the number of
outputs), resulting in an n x m array where n and m are conventionally 8,
16, 32, etc. The optical diverters 48, 50 of Fig. 7 are optically reflective
elements, for example mirrors. An optical diverter in an activated state
(e.g. the optical diverter 48 in Fig. 7) typically inserts a 3-7 dB loss in
optical power in the redirected signal (e.g. the signal Sc1 in Fig. 7),
depending upon the MEMS switching device port count, the quality of the
CA 02320613 2000-09-21
34
design and the fabrication of the parts. A signal that passes through a
MEM switching device into one of its row outputs (e.g. the signal Sc2 in
Fig. 7) usually has a lower drop in power, typically 1-2 dB, again
dependent upon device size and design. Although MEMS switching
s devices are shown in this embodiment of the active crosspoints of the
switching matrix 18, any matrix of optical diverters capable of directing
light of the required wavelength, and as desired, could be used.
Fig. 9a illustrates, in a functional block diagram, a second
embodiment of the switching matrix 18, shown for the first channel
~o wavelength. The switching element of the 32 x 32 optical switching
matrix 18 is comprised of four 16 x 16 MEM devices 19a1 , 19a2, 1 9b1 ,
and 19b2, which are controlled by the controller 2B. The MEM device
19a1 has sixteen inputs ip1 to ip16, which are connected to the WD
demultiplexers 16 of the ports one to sixteen, respectively. The MEM
Is device 19a1 has two sets of outputs, one set of column outputs
corresponding to the resultant path of an optical signal that has been
directed by an optical diverter in an activated state. The column outputs
are labeled opt to op16 in the figure. The other set of outputs are row
outputs corresponding to the path of an optical signal which is not
2o directed by any optical diverters. That is; all of the optical diverters in
the
path of the optical signal are in a non-activated state (e.g. the optical
diverter 50 shown in Fig. 7). The row outputs of the MEM device 19a1
are connected to the inputs of the MEM device 19a2. The column outputs
of the MEM device 19a2 are labeled op17 to op32. Likewise, the MEM
2s devices 19b1 and 19b2 are connected in a similar manner for inputs ip1 7
to ip32 and outputs opt' to op32'. Each of the column outputs opt to
op32 from the MEM devices 19a1 and 19a2 are combined with its
respective row output opt' to op32' from the MEM devices 19b1 and
19b2. This is done using thirty-two 2:1 combiners;: one of such combiners
30 70 is shown for port thirty-two. These combiners are single mode
compatible combiners. Both fused fiber couplers and Silica on Silicon
waveguide structures are appropriate. Both of these technologies will add
about 3 -3.5 dB of loss to the cross-switch budget. The combiner 70
combines the outputs op32 and op32' to produce an output op32". An
3s optical signal will appear at the output op32 or op32', depending on
which input an optical signal destined for port thirty-two is applied. For
example, an optical signal applied to the input ip1 will be output at one of
the outputs opt to op32, whereas an optical signal applied to the input
CA 02320613 2000-09-21
ip17 will be output at one of the outputs opt' to op32'. The output of
each 2:1 combiner is connected to the WD multiplexer 20 of its
respective output port, or to an input of the wavelength-converting switch
as described earlier. Fig. 9a shows the outputs of the combiner 70
s coupled to the input for the first channel wavelengi:h of the WD
multiplexer 20 and the resultant multiplexed signal is forwarded to the
optical amplifier 22 for that port. Using this arrangement an optical
switching element 19 of a given dimension can be implemented using
MEMs of smaller dimension. In this case, the switching element 19 is
realized using two pairs of smaller switching elements 19a1, 19a2 and
19b1, 19b2.
Fig. 9b to 9d illustrate, in functional block diagrams, other
embodiments of the optical switching matrix 18, shown for the first
channel wavelength. Several variations exist on the theme of using the
t5 through output port / third port of a MEMS device. One of these, already
described with reference to Fig. 9a, is to make 32 ;K 32 switches out of
16 x 16 switches. Such an approach is useful before 32 x 32 MEMs
devices are readily commercially available. Another variation, shown in
Fig. 9b, is to use four 32 x 32 MEMS modules to build a 64 x 64
2o switching element for each channel wavelength. The result is a 10240 x
10240 wavelength switch capacity, assuming 160 channel wavelengths
(160 x 64 = 10240?. Again, each pair of respective outputs, for example
op64 and op64', are combined with a combiner 70. The output of the
combiner 70 is either coupled to a WD multiplexer 20, as shown, or to an
2s input of the wavelength-converting switch 28. Still another variation,
shown in Fig. 9c, is to partition the cross-connect switch 10 such that
the initial implementation is 32 x 32 on an initial optical switching matrix
card 72, with the provision of an expansion port 7'.3. This expansion port
73 and an extension board 74 are used to extend the initial optical
3o switching matrix card 72 to the size of 32 x 64. In this case an identical
switch having another 32 x 64 optical switching card 75, can then be
used to create an expanded switch having double the capacity in terms of
port count. In this case two alternatives exist for coupling together the
outputs. The first is shown in Fig. 9c, where per port per wavelength
3s combiners 70 are provided, of which there would be sixty-four per
wavelength, and therefore 64 x 160 per switch. The output of each
combiner would be connected to a respective channel wavelength input
of a wavelength division multiplexer 20 assigned to the respective port
CA 02320613 2000-09-21
36
(as shown), or to an input of the wavelength-converting switch 28 (not
shown). The second approach, shown in Fig. 9d, is to reverse the
sequence of combining outputs of the MEMs and multiplexing the
combiner outputs. This approach can be used for outputs that are to be
s multiplexed and eventually coupled to output ports of the switch 10.
However, for outputs that are to coupled to the wavelength converting
switch 28, these outputs should be combined in rE;spective pairs (e.g.
op64 and op64') before being coupled to the wavelength-converting
switch 28. Fig. 9d shows forty-eight outputs of the switching matrix 18
to destined for output ports of the switch 10 and sixteen to be coupled to
the wavelength converting switch 28 (i.e. N =48 and K = 16). In this
second approach two banks of wavelength division multiplexers 20a, 20b
multiplex the output port destined outputs of the MEMs, one multiplexer
per port multiplexing M channel signals. That is, the first bank of
rs multiplexers 20a, multiplexes the outputs opt to op48 using one
multiplexer per port, each multiplexer multiplexing M channel signals of
distinct wavelength, and the second bank of multiplexers 20b, performing
the same function for ports opt' to op48'. The resultant multiplexed
signals of the banks 20a, 20b are combined on a port-by-port basis by
2o respective combiners 70a. Two such resultant multiplexed signals 77a
and 77b are shown in the figure. The output of each combiner 70 is then
applied to a respective optical amplifier 22 for the port. The outputs,
op49 to op64 and op49' to op64', are combined in respective pairs and
each combined output is coupled to an input of the wavelength-
2s converting switch 28 in the same manner as described earlier with
reference to Fig. 9c. For example, Fig. 9d shows the outputs op64 and
op64' connected to the combiner 70b, the resultant combined output of
which is then for coupling to the wavelength-converting switch 28. This
second approach increases the number of output 1NDM multiplexers 20a,
30 20b from 64 to 128, but reduces the number of combiners 70 from
10240 (i.e. 64 x 160) down to 2608 (i.e. 16 x 160 + 48). The second
approach also simplifies cabling. Both of the variations shown in Figs. 9c
and 9d use two-port and three-port MEMs to allow the optical
telecommunications switch to expand in the ports per wavelength
3s direction.
Fig. 9e illustrates, in functional block diagram, still another
embodiment of the optical switching matrix 18, shown for the first
channel wavelength. The optical switching element is comprised of four
CA 02320613 2000-09-21
37
three-port P x P MEMs, which are referenced generally as 19c. The three-
port MEMs 19c have interchangeable input and oul:put ports and an
expansion port, which acts as an input expansion port or an output
expansion port, depending on whether the input and output ports are
s "normal" or "reversed". When the expansion port (third port) acts as an
input expansion port it is aligned with the output port, as is the case with
the MEMs 19d2 and 19d4. When the expansion port acts as an output
expansion port it is aligned with the input ports, as is the case with the
MEMs 19c1 and 19c3. MEMs 19c1 and 19c3 have an expansion output
to port Eo as the third port, and MEMs 19c2 and 19c4 have an expansion
input port Ei as the third port. Operation of the MEMs 19c1 and 19c3 are
of the same as the MEMs discussed earlier with reference to Figs. 7 to
9d. In the case of the MEMs 19c2 and 19c4, the inputs of the expansion
input port Ei of these MEMs are physically aligned 'with respective output
t5 ports (O) of these devices. An optical signal from an expansion input port
(Ei) input will exit the MEMs from the respective output if none of the
deflection mirrors in the column corresponding to the output has been
activated into an upright position. In this way, any output of the MEMs
19c2, 19c4 can either emit an optical from its respective expansion input
2o ' port or from an input port (I) of the MEMs.
In Fig. 9e, the MEMs 19c1 is the original MEMs device, that is,
before addition of MEMs 19c2 to 19c4, to expand switching element 19.
Inputs 1 to P from input ports 1 to P of the switching matrix 18 are
connected to the input port (I) of the MEMs 19c1 . The input port (I) of
zs MEMs 19c2 is coupled to the expansion output port Eo of the MEMs
19c1, and output 1 to P of the switching matrix 18 are coupled to the
output port of MEMs 19c2. The output port (O) of MEMs 19c3 is coupled
to the input expansion port Ei of the MEMs 19c2, and has inputs P+ 1 to
2P from the now expanded switching matrix 18 coupled to its input port
30 (I). The input port (I) of the MEMs 19c4 is coupled to the expansion
output port Eo of the MEMs 19c3, and has outputs P+ 1 to 2P of the
expanded switching matrix 18 coupled to its output port (O1. In this
arrangement any of the inputs 1 to 2P of the expanded switching matrix
18 can be switched to any its outputs 1 to 2P. Hence the original P by P
3s switching element 19 has been expanded to a 2P by 2P switching
element without the use of combiners 70, which saves about 2-3dB in
optical power loss. Again, N inputs and N outputs of the switching
matrix 18 shown in Fig. 9e would be coupled to input and output ports of
CA 02320613 2000-09-21
38
the switch 10 through WD demultiplexers 16 and WD multiplexers 20,
respectively. Another K inputs and K outputs of the switching matrix 18
would be coupled to the wavelength-converting switch 28.
Fig. 9f illustrates, in functional block diagram, yet another
s embodiment of the optical switching matrix 18, shown for the first
channel wavelength. The optical switching element is comprised of four P
by P four-port MEMs 19d1 to 19d4, referred to generally as MEMs 1 9d.
Each of the MEMs 19d1 to 19d4 has an input port (I), an output port (O),
an expansion output port Eo, and an expansion input port Ei. In this
1n arrangement the original MEMs 19d1 provides support for P inputs and P
outputs before expansion of the switching element: 19. After expansion,
by the addition of the MEMs 19d2 to 19d4, the switching element 19
supports 2P inputs and 2P outputs. The interconnection of the MEMs
19d1 to 19d4 is the same as the MEMs 19c1 to 19c4 in Fig. 9e. The
t~ fourth port of the MEMs 19d1 to 19d4 is used for connection to the
wavelength-converting switch 28. The expansion output ports Eo of the
MEMs 19d2 and 19d4 have outputs 1 to P and P+ 1 to 2P, respectively,
which are coupled to inputs of the wavelength converting switch 28. The
input expansion ports Ei of the MEMs 19d3 and 19d1 have inputs 1 to P
2o and P + 1 to 2P, respectively, which are coupled to outputs of the
wavelength-converting switch 28. In this arrangement an output 1 to P
from the wavelength-converting switch 28 can be passed to a respective
output 1 to P of the switching element 19, via the MEMs 19d3 and 19d2.
Similarly, an output N + 1 to 2P from the wavelength-converting switch
~a 28 can be passed to a respective output P+ 1 to 2P of the switching
element 19, via the MEMs 19d1 and 19d2. An advantage of this
arrangement over the previous expansion arrangements described in Figs.
9a to 9e, is that input ports (I) and output ports (0) of the MEMs are not
required for connection to the wavelength-converting switch 28. This
3o connection is achieved through the expansion input ports (I) and
expansion output ports (O) of the MEMs 19d1 to 19d4. Therefore, a 2P
by 2P switching element 19 constructed as such with four-port MEMs
can provide full interconnection between its 2P inputs and 2P outputs as
well as provide connection of its 2P inputs to the wavelength-converting
switch 28. However, such a 2P by 2P switching element 19 no longer
has the ability to concentrate signals to be converted into a smaller
number of ports (i.e. less than 2P) connected to the wavelength-
converting switch 28. The lack of this ability places constraints on the
CA 02320613 2000-09-21
39
implementation of the wavelength-converting switch 28. However,
embodiments of the wavelength-converting switch 28 that are compatible
with these constraints will be discussed later in this document.
It should be clear that embodiments of the switching matrix 18
s shown in Figs. 9a to 9f that implement schemes far expanding the
switching capacity of an original switching matrix '18 exploit a particular
property of MEMs devices. That is, these embodiments make use of the
fact that the through path has a smaller loss ( -- 1 dB) than the switched
path ( ~ 5dB) in order to tandem multiple MEMs devices without incurring
to excessive losses. This is particularly important in the context of a
photonic switch having a link budget which the switched path and
through path losses have to remain within. Fig. 9g illustrates, in functional
block diagram, yet another embodiment of the optical switching matrix
18, shown for the first channel wavelength. The optical switching
~5 element is comprised of one P by P four-port MEMs 19d. The switching
matrix 18 provides full interconnection between its P input ports and its P
output ports, that is any one of the P inputs can be switched to any one
of the P outputs. The switching element 19 further provides
interconnection of all P inputs to the wavelength-converting switch 28,
2o through the expansion output ports Eo of the MEMs 19d. Furthermore,
the switching element 19 provides interconnection of all P outputs from
the wavelength-converting switch 28, through the expansion input ports
Ei of the MEMs 19d. However, it should be noted that each input of the
MEMs 19d corresponds to a respective expansion port output, which is
2s aligned with the particular input, hence an optical signal arriving at an
input can not be switched to a different expansion port output. Likewise,
each expansion port input of the MEMs 19d is aligned with a respective
output and can not be switched to a different output. Such a 2P by 2P
switching element 19 no longer has the ability to concentrate signals to
3o be converted into a smaller number of ports (i.e. less than 2P) connected
to the wavelength-converting switch 28. The lack of this ability places
constraints on the implementation of the wavelength-converting switch
28. However, embodiments of the wavelength-converting switch 28 that
are compatible with these constraints will be discussed later in this
35 document.
Fig. 10 illustrates, in a functional block diagram, a second
embodiment of the wavelength-converting switch 28. The wavelength-
CA 02320613 2000-09-21
converting switch 28 includes K channel convertors 80. Each channel
convertor 80 has M inputs and M outputs. There is one input and one
output for each channel wavelength. Each optical switching matrix 18 for
a particular channel wavelength has an inter-matrix output connected to
s the input for the corresponding channel wavelength and an inter-matrix
input connected to the output for the corresponding channel wavelength
of each channel convertor 80.
As noted above with reference to Fig. 3, typically 25% of the
inputs and outputs of the optical switching matrix 18 are connected to
o the wavelength-converting switch 28. Thus, a channel convertor is
required for each of these input/output connections. Accordingly, K=8 in
the case of the optical switching matrices 18 being 32 x 32 matrices.
Furthermore, as shown in Fig. 3, there are R inputs/outputs on the
wavelength-converting switch 28 for add/drop traffic.
~s As shown in figure 8, the R drop outputs and the R add inputs
are distributed equally among the K cards, resulting in R/K drop outputs
and R/K add inputs per card. Each channel convertor 80 has a M x L
optical switch 82 connected to the inputs of the wavelength-converting
switch 28. Since there are K channel convertors 8c7, the number of inputs
2o that the wavelength-converting switch 28 has is K*~M, which equals
1280 inputs. The wavelength-converting switch 28 also has 1280
outputs. A subset L of the outputs of the M x L optical switch 82 are
each connected to respective optical receivers 84. In this embodiment the
number L equals 32, but could be any number less than or equal to M.
2s The M x L optical switch 82 directs input optical signals to the optical
receivers 84 according to its connection map, which is determined by the
controller 26. Each optical receiver 84 converts a received input optical
signal into an electrical signal and outputs the electrical signal to a
selector 90 (or small electrical switching fabric). The selector 90 includes
3o the R/K inputs for adding channel signals and the R/K outputs for dropping
channel signals. Optionally, a bus 91 interconnects the selector on each
of the K channel convertors 80. The bus 91 has a width of K~L (i.e. L
connections driven by a selector 90 on each of the K channel convertors
80). The controller 26 controls the operation of adding and dropping
35 channel signals via the selector 90. This operation is shown as being done
with electrical signals however the selector 90 could include opto-electric
conversion capabilities to add/drop optical channel signals. The selector
CA 02320613 2000-09-21
41
90 forwards electrical signals to a tunable optical source 86. There are L
tunable optical sources 86. Each tunable optical source 86 is operable
over a range of M channel wavelengths. Currently available tunable
sources typically have a tunable range of thirty-two channel wavelengths,
s however this range is increasing. Each tunable optical source 86 receives
an electrical signal from the optical receiver 84 to which it is connected
and outputs an output optical signal, which contains information present
in the input optical signal, to an L x M optical swit<;h 88. The output
optical signal has a channel wavelength equal to the wavelength at which
to the controller 26 has set the tunable source 86. The L x M optical switch
88 directs output optical signals, according to its connection map set by
the controller 26, to the outputs of the channel convertor 80.
The operation of the channel convertor 80 will now be further
explained by way of example. An optical signal Sc'I , of channel one
is wavelength, is applied to the first channel convertor 80 at the first input
of the M x L optical switch 82. The optical signal Sc1 is directed by the
M x L optical switch 82 to the first optical receiver 84. The first optical
receiver 84 converts the information contained in the optical signal Sc1
into an electrical signal Ec1 . The electrical signal Ec;1 passes through the
2o selector 90 and is applied to the first tunable source 86. The first
tunable
source 86 has been set to output an optical signal at the channel twenty
wavelength. The first tunable source 86 outputs an optical signal Sc20,
which contains the information in the electrical signal Ec1, to the first
input of the L x M optical switch 88. The L x M optical switch 88 directs
2s the optical signal Sc20 to the twentieth output of l:he channel convertor
80. The optical signal Sc20 is then further directed by the optical
switching matrix 18 assigned to channel twenty.
In the case that the optical switching matrix 18 shown in
either of Figs. 9e or 9f is used in the switch 10, then the wavelength-
3o converting switch 28 would comprise P channel convertors 80 (i.e. K = P)
and the bus 91 with a width P~L. The bus 91 would then provide a
similar degree of interconnectivity between the ports of the wavelength-
converting switch 28, as the embodiment described earlier with reference
to Fig. 4a.
3s A physical implementation of the embodiment of the
wavelength-converting switch 28 shown in Fig. 10 would easily be
CA 02320613 2000-09-21
42
realized in a manner similar to that shown in Fig. 4b. That is, the channel
convertors 80 would be implemented on separate circuit cards with the
bus 91 interconnecting the cards, and each of the cards connected to
each of the switching matrices 18.
s Fig. 1 1 illustrates, in a functional block diagram, a third
embodiment of the wavelength converting switch shown in Fig. 3. This
embodiment includes K/S, S*M x S*~M channel convertors 92, where
M = 160 and S = 2 in Fig. 1 1 . Selection of a value for S will be explained
later. The structure of the channel convertor 92 will be explained with
1o reference to Figs. 13 to 15 which provide tables specifying
interconnections between components of the channel convertor 92. The
channel convertor 92 has S~M inputs and S*~M outputs. The inputs and
outputs are arranged in S banks, in this case there are two banks, bank1
and bank2. Each bank has an input and an output for each of the channel
is wavelengths. That is, there are M inputs and M outputs per bank. The
inputs/outputs of the banks connect to inter-matrix: outputs/inputs of the
appropriate switching matrices according to their channel wavelength. In
the present embodiment of the switch 10, where K; =8, the wavelength-
converting switch includes four 2M x 2M channel convertors 92.
2o The channel convertor 92 includes an optical switch100
(OXC-A), which comprises five 32 x 32 optical switches OXC1 to OXCS.
The optical switch 100 is connected to the first bank of inputs via an
interconnect 102 (interconnect A). The interconnect 102 connects the
inputs of bank1 , to the inputs of the optical switch 100 according to table
2s 1 in Fig. 13. For example, table 1 shows that the input is connected to
the input 11 of the 32 x 32 optical switch OXC1. The remaining inputs of
the optical switch OXC1 are connected to every fifth input of bank1 (e.g.
12 connected to 13 to and so on). Similarly, the optical switches OXC2 to
OXC5 have inputs connected every fifth input of bank1 starting at input
3o to ~~, respectively, as shown in tablet . The channel convertor 92 also
includes another optical switch 104 (OXC B) and another interconnect
(interconnect_B) connected in a similar manner to the inputs of bank2, as
shown in tablet of Fig. 13.
The channel convertor 92 further includes a wavelength
3s convertor 108 which is connected to the optical svvitches 100, 104
(OXC A and OXC B) via an interconnect 110 (interconnect C). The
CA 02320613 2000-09-21
43
wavelength convertor 108 includes ten convertor modules 106, labelled
G 1 to G 10. Each convertor module 106 includes up to 32 tunable
transponders, a 32 x 32 optical switch, 32 inputs labelled I(1:32) and 32
outputs labelled O(1:32). The convertor module 106 will be described in
s more detail later. The value of S, referred to earlier, is chosen to match
the range of the tunable transponders to the size of the optical switches
in the convertor modules G 1 to G 10. In this case, the transponders have
a range of 16 channel wavelengths and the optical switches are 32 x 32,
hence S is 32/16 = 2. The interconnect 1 10 connects the inputs of the
to wavelength convertor 108 to the outputs of the two optical switches
100, 104, as shown in table 3. For example, the first ten outputs of the
optical switch OXC1 (01 to 010) are connected to the first input of the
convertor modules G1 to G10, respectively. Likewise, the second and
third sets of ten outputs (01 1- 020 and 021- 03G) of the optical switch
t5 OXC1 are connected to the second and third inputs of the convertor
modules G1-G10, respectively. The remaining two outputs 031 and 032
of the optical switch OXC1 are connected to the thirty-first inputs of the
first and second convertor modules G 1 and G2, according to table 3. The
remaining optical switches OXC2 to OXC5 of the optical switch 100
20 (OXC_A) are connected in a similar manner as shown in table3 of Fig. 14.
Similarly, the outputs of the optical switch 104 (O:XC B) are connected to
the wavelength convertor in a similar manner, as shown in table3.
The outputs of the wavelength convertor 108 are connected
to the two banks of outputs via another interconnect 1 12
25 (interconnect-D), as shown in table4 of Fig. 15. For example, the first
sixteen outputs 01-016 of the convertor module G1 are connected to the
first sixteen outputs of bank1, respectively. The remaining sixteen
outputs 017-032 are connected to the first sixteen outputs of bank2,
respectively. The remaining convertor modules G2 to G10 are connected
3o in a similar manner to the remaining outputs in the banks, according to
table4.
The wavelength convertor 108 also has add inputs and drop
outputs for adding/dropping channel signals. Since the wavelength
converting switch 28 provides R of each such inputs/outputs then each
35 2M x 2M channel convertor, and hence each wavelength convertor 108,
provides 2R/K add inputs and 2R/K drop outputs. This will be explained in
more detail later.
CA 02320613 2000-09-21
44
The controller 26 controls the operation of the optical
switches 100, 104 (OXC A and OXC-B) and the wavelength convertor
108.
The interconnects 102, 103, 1 12 (interconnect A,
s interconnect-B and Interconnect_D) would typically be implemented as
optical fiber connection. However, the interconnect 110 (interconnect D)
could be optical fiber, but could additionally includE: several 32 x 32
optical switches which interact with the optical switches 100,104 and
the optical switches in the convertor modules G 1 to G 10 to create a
to standard CLOS arrangement.
Operation of the third embodiment of the wavelength-
converting switch 28 will now be further explained by way of example. A
channel signal Sc1 of a first wavelength arrives at the first input of
bankl . The interconnect 102 (interconnect A) connects the signal Sc1 to
t5 the first input 11 of the optical switch OXC1 which routes the signal Sc1
to its tenth output 010. The interconnect 110 (interconnect C) connects
the signal Sc1 to the first input 11 of the tenth convertor module G10.
The tenth convertor module G10 receives the signal Sc1, converts it to
another channel signal Sc160 of the 160 channel wavelength, and routes
2o the signal Sc1 60 to its sixteenth output 016. The interconnect 1 1 2
(interconnect_D) connects the signal Sc160 to the 160th output of bank1
where it is output from the wavelength-converting switch 28, into one of
the ports on the optical switching matrix 18 associated with the channel
wavelength.
Fig. 12 illustrates in a functional block diagram an embodiment
of a converter module 106 shown in Fig. 1 1 . The convertor module 106
includes receivers 1 14, connected to the inputs 11 to 132, a selector 1 1 5
connected to the outputs of the receivers 1 14 for add/drop capability, and
3o tunable transponders 116, connected to the outputs of the selector 1 1 5.
The tunable transponders 1 16 are tunable over a range of sixteen channel
wavelengths in this embodiment. The convertor module 106 is
provisionable for up to 32 tunable transponders. However, options exist
where some of the tunable transponders can be replaced by fixed
3s transponders. Each receiver 1 14 can receive a channel signal of any of
the M wavelengths, and convert the received channel signal to an
CA 02320613 2000-09-21
electrical signal. The selector 1 15 is used to add/drop electrical signals
in/out of the convertor module. There are U add inputs and U drop
outputs shown in Fig. 12, where U = 2R/1 OK. The value for U is derived
from the total number (R) of add/drop inputs/outputs for the wavelength
s converting switch 28 divided by the number of S~M x S*M channel
convertors (K/2) divided by the number of wavelength convertor modules
(10). The outputs of the transponders 116, are connected to the inputs of
a 32 x 32 optical switch 1 18. The optical switch 1 18 routes each
channel signal it receives to an output according to the wavelength of the
to channel signal being routed. Operation of the tunable transponders 1 16
and the optical switch 118 is under control of the controller 26.
Figs. 13 to 15 are tables which respectively show the
connections made by the interconnects A and B, interconnect C, and
interconnect D of Fig. 11.
Is A physical implementation of the embodiment of the
wavelength-converting switch 28 shown in Fig. 11 would easily be
realized in a manner similar to that shown in Fig. 4b. That is, the channel
2M x 2M convertors 92 would be implemented on separate circuit cards
with each of the cards connected to each of the switching matrices 18.
2o With reference to Figs. 16a and 17, thE; physical arrangement
of the switch 10 will now be described. The basic switch physical
structure includes two arrays of circuit cards arranged physically
orthogonal to each other. One plane of the physically orthogonal
arrangement consists of per lambda switching circuit cards 216 while the
2s other orthogonal plane consists of I/0 circuit cards 202 (i.e.
tributary/WDM cards) and wavelength convertor circuit cards 214, which
also have add-drop ports. Only one wavelength convertor circuit card 214
is shown in Fig. 16a for clarity, however there could be several as
described earlier with reference to Fig. 4b. This arrangement facilitates an
3o array of optical connections between the cards, with every I/O card 202
and convertor card 214 having access to every switching card 216. The
arrangement also eliminates any need for an optical backplane since all
the optical connections simply pass straight through a midplane 206, the
function of which is primarily to provide mechanical alignment for the
3s optical connections and electrical interconnect between the cards.
CA 02320613 2000-09-21
46
An input fiber 200 is coupled to the WD demultiplexer 16 on
an I/O circuit card 202 and an output fiber 204 is coupled to the WD
multiplexer 20 on the same I/O card. I/O circuit cards 202 are held in
mechanical alignment with respect to the switching circuit cards 216 by
s the midplane 206. This alignment is accomplished via alignment ferrules
210, which are mounted on and pass through the midplane 20, and by a
plurality of optical connectors 208, 218 mounted adjacent an edge 203,
21 7 of the I/0 circuit cards 202 and switching circuit cards 216,
respectively. The controller 26, implemented on a controller circuit card
l0 21 2, and wavelength converting switch 28, implernented on a convertor
circuit card 214, are also aligned by optical connecaors 208 on the cards
21 2, 214 which are inserted into the alignment ferrules 210 on the
midplane 206. Alternatively, the controller card 212 and convertor card
214 could be interface cards connected to a centr<~I controller 26 and
15 central wavelength converting switch 28. These alternatives will be
described later in more detail.
Additionally, as is commonly used with circuit cards and
midplanes, other hardware such as tracks and clamps (not shown) are
used to hold the cards. There is a plurality of such I/O cards, however
20 only one is shown in the Fig. 16 for clarity. Furthermore, there can also
be a plurality of convertor circuit cards 214, depending on the size of the
wavelength-converting switch 28 and how it is partitioned into circuit
cards. All of the aforementioned circuit cards are arranged in a standard
orientation with respect to the midplane 206 without the need for
Z5 propagating optical signals along a backplane structure. That is, the
circuit
cards are spaced apart at standard intervals, are substantially parallel to
each other, and are perpendicular to the midplane :206. Besides providing
mechanical alignment, the midplane 206 also provides electrical
connectivity and power to the I/0 cards 202, controller cards 21 2,
3o convertor cards 214 and switching cards 216.
There can be several optical switching matrices 18 per
switching card 216. There is a plurality of switching circuit cards 216.
Fig. 1 6a shows sixteen switching circuit cards 216 covering channel
wavelengths 1 to 160. The switching circuit cards 216 are arranged in
3s the previously mentioned standard orientation with respect to the
midplane 206. However, each of the switching circuit cards 216 is on
the opposite side of the midplane 206 with respect to the I/0 circuit cards
CA 02320613 2000-09-21
47
202, convertor circuit cards 214, and controller circuit cards 212, and is
also in a perpendicular orientation with respect to the same cards. In this
way, each I/O circuit card 202 is in close physical proximity to each
switching circuit card 216 and can be communicatively coupled via
respective optical connectors 208, 218 on the cards and by way of the
alignment ferrules 210 on the midplane 206.
For cross-connect switches 10 having a large number (P) of
input/output ports, or a large number (M) of channel wavelength per port,
the switch 10 can be configured with shelves, each shelf containing a
to subset of the circuit cards. Additionally, it may be desirable to include
more than one optical switching matrix 18 on a switching circuit card
216, as shown in Fig. 16. For example, referring to the embodiment
described earlier in which M=160 and P=32, each switching circuit card
216 could include ten optical switching matrices 18, each matrix 18 for a
separate channel wavelength. In this case, sixteen switching circuit cards
216 would be required to support 160 channel wavelengths, ten
switching circuit cards 216 per shelf.
Fig. 16b is a perspective view of another physical arrangement
of the cross-connect switch of Fig. 3, which includes the wavelength-
2o converting switch 28 of Fig. 4e. Only one wavelength convertor circuit
card 214 has been shown for clarity, although there could be several such
cards (e.g. four) as described earlier with reference to Fig. 4e. The circuit
card 214 optically connects to the switching circuit cards 216 via
connectors 208, 218 (not shown) and alignment ferrules 210 as
described earlier with reference to Fig. 16a. Transmit bank 33 and receive
bank 35 are coupled to the connectors 208 via optical fibers. The
interface 37 is connected to the transmit bank 33 and receive bank 35 as
described earlier with reference to Fig. 4e. A connE;ctor 209, either optical
like the connector 208, or an electrical connector, couples the interface
37 to the electrical switch 30 via the electrical, or optical, bus 39. The
electrical switch 30 is provided on a switch circuit card 21 5 and connects
to the bus 39 via an alignment ferrule 21 1, similar to the alignment ferrule
210, in the case of an optical connection, or a double-ended male type
connector for an electrical connection. A connector (not shown)
corresponding to the connector 209 is provided on the switch circuit card
21 5.
CA 02320613 2000-09-21
48
With reference to Figs. 17 to 18c, the aptical connectors 208,
218 and alignment ferrules 210 will now be described in further detail.
The optical connector 208 is mounted adjacent the edge 203 of an I/O
circuit card 206. The optical connector 208 is comprised of a housing
220 mounted on the I/O circuit card 202 via elongated, or slotted,
through holes 222 and bolts or rivets (not shown). The longitudinal axis
of the holes 222 is aligned with the edge 203 of the circuit card 202.
Mounting the housing 220 to the circuit card 202 in this manner allows
movement of the connector 208 along a portion of the edge 203 of the
circuit card 202, as shown by arrows (A) in Fig. 17. The range in
movement of the connector 208 should be sufficient to allow the
connector 208 to be brought into alignment with the alignment ferrule
210 and inserted in it. Typically, this range is in thE; order of a
millimetre.
The housing 220 houses a mating insert 224 having a mating face 226,
is which faces in the same direction as the edge 203 of the circuit card
202. The mating face 226 has a pair of sockets 22.8 for receiving
alignment pins 249 from the corresponding optical connector 218
mounted adjacent an edge 217 of a switching circuit card 216. The
alignment pins 249 are precision tungsten pins, or another hard, durable
20 material. Other positive engagement, or alignment, features could be used
as well or instead. An optical fiber ribbon cable 229 having a plurality of
optical fibers 230 is held in the mating insert 224. Each fiber 230 has an
end 232 that is flush with the mating face 226. A pair of leaf springs 234
mounted in the housing 220 provides flexible biasing of the mating insert
25 224 in the direction that the mating face 226 faces. The mating insert
224 is mounted in the housing 220 such that it is moveable in the
direction of the biasing and in the opposite direction, as shown by arrows
B in Fig. 17. The result is the mating insert 224 can move in a direction
transverse to the edge 203 of the circuit card 202 and in a plane parallel
3o to the plane of the circuit card 202. The biasing helps ensure the optical
fibers of the connectors 208, 218 remain in a communicatively coupled
relationship when the connectors 208, 218 are in the alignment ferrule
210.
The corresponding optical connector 218 mounted on the edge
35 217 of the switching circuit card 216 is similar in structure to the
optical
connector 208 described above. The difference is that it does not contain
the sockets 228, but instead includes the alignment pins 249 and has
been rotated by 90 degrees with respect to the card 216. The optical
CA 02320613 2000-09-21
49
connector 218 has a housing 236, which houses a mating insert 238
having a mating face 240. The housing 236 is mounted adjacent the edge
217 of the switching circuit card 216 via slotted through holes 242. This
is done in a manner that allows movement of the housing 236 along the
s edge 217 of the switching circuit card 216, as shown by arrows C in Fig.
17. A pair of leaf springs 243 provides biasing in the direction of the
mating face 240. The mating insert 238 is mounted such that it is
moveable in the direction of the biasing and in the opposite direction, as
shown by arrows D in Fig. 17. An optical fiber ribbon cable 244 having a
o plurality of optical fibers 246 is held in the mating insert 238. Each fiber
246 has an end 248 which is flush with the matinc3 surface 240 such that
they achieve an optically coupled relationship with a respective fiber 230
of the optical fiber ribbon cable 229 when the mating surfaces 226, 240
are brought into contact with each other. The plurality of optical fibers
is 230, 246 are connected to optical components such as WD
demultiplexers 16, WD multiplexer 20, and optical switching matrices 18
on their respective circuit boards.
The alignment ferrule 210 is mounted on the midplane 206
and extends through an opening 250 therein. There are a plurality of
2o openings 250 in the midplane 206 for mounting a plurality of the
alignment ferrules 210, one of such openings 250 is shown without an
alignment ferrule 210 in Fig. 17. These openings 250 are located at the
intercepts of the switching circuit cards 216 and the convertor circuit
cards 214 (or I/O cards 202, controller cards 212) to provide a path for
25 optical connections between the cards 216, 214. The alignment ferrule
210 has an aperture 252 for receiving the mating inserts 224, 238. The
alignment ferrule 210 has a chamfered inner edge 254, 256 around the
periphery on either side of the aperture 252 for assisting the mating
inserts 224, 238 into alignment. Alternatively the aperture 252 could
3o have a tapered inner surface which gradually reduces the size of the
aperture, reaching a minimum at, or near, the midpoint of the aperture
252 (as shown in Fig. 18c). In this case the mating inserts 224 and 238
could further have chamfered, or sloped, corners 226a,b and 240a,b on
their mating faces 226, 240, respectively. The mating face 226 has a
3s ridge 258 aligned with the sockets 228 and the mating face 240 has a
corresponding groove 260 aligned with the pins 249. The ridge 258 and
the groove 260 are for assisting the mating faces ;?26, 240 into
alignment such that the pins 249 can be inserted into the sockets 228,
CA 02320613 2000-09-21
thereby aligning the fiber ends on the polished faces 232,248, to
establish an optical connection between the plurality of optical fibers 230,
246.
Both connectors 208, 218 require two degrees of movement
s within the plane of the midplane 206 unless the alignment ferrule 210 is
provided this freedom of movement (shown as arrows E in Fig. 1 7) by the
manner in which it is mounted on the midplane 206. In the case where
the alignment ferrule 210 is fixedly mounted on the midplane 206, a small
amount of flexing of the circuit card (202, 212, 214, and 217) provides
»> one degree of movement while the moveable manner in which the
respective connector 208, 218 is mounted on its card (as described
earlier) provides the other degree of movement.
With reference to Fig. 18c the operation of the alignment
features of the optical connectors 208, 218 and alignment ferrule 210
is will now be discussed in further detail. Precision in alignment in the
order
of at least 1-2 microns is required to optically connect the polished faces
of the optical fibers 229, 244. Alignment progresses in three stages; each
stage providing a finer degree of precision in the alignment. The first
stage is provided by the mechanical interaction of the chambered, or
2o sloped, corners 226a,b and 240a,b with the corresponding chambered, or
tapered, surfaces 254, 256 of the alignment ferrule 210. This first stage
provides approximately one millimeter of alignment precision. The second
stage of alignment is provided by mechanical co-operation between the
ridge 258 and corresponding groove 260 on the mating faces 226, 240
?s of the mating inserts 224, 238, respectively. This stage provides
approximately 20-100 microns of alignment accuracy. The final stage of
alignment is provided by the engagement of the pins 249 in the sockets
228. This final stage provides approximately 1-2 microns of alignment
accuracy. The details of the alignment of the fibers 229, 244 within
3o respective alignment structures 224a,b and 238a,b will be described with
reference to Fig. 18d.
Fig. 18d is a cross-sectional front view of the mating face 240
of the connector 218 taken along the line BB in Fig. 18c. Two etched
silicon wafer slice alignment structures 238a,b are housed in the mating
:~ insert 238. Each structure 238a,b has fiber groovE;s 253 for aligning
fibers 246 and pin grooves for aligning pins 249 etched on one of its
CA 02320613 2000-09-21
51
planar surfaces. The fibers 246 are stripped of their protective cladding
before installation in the fiber grooves 253. The fiber grooves 253 are V-
shaped with a side dimension (a) equal to approximately 120 microns to
accommodate a 1 25 micron fiber 246 with allowance for an epoxy fill
s 251 between the structures 238a,b. The thickness of the epoxy is set by
compressing the structure 283a,b together thereby clamping the pins 249
and fibers 246 in position. The etched V-shaped grooves on the silicon
are dimensioned such that the silicon clamps firmly on to the fibers when
a gap of about 5-7 microns exists between the wafers. Hence the V-
to shaped grooves clamp the fibers into their locations with high precision.
The etched pin grooves 255 are also V-shaped and have a side dimension
(d) equal to about 245 micron to accommodate a tungsten pin 249 of just
under 250 microns in diameter and a maximum length in the order of
2000 microns (2mm) of protrusion beyond the mating insert 238. The pin
is could also be of square cross-section with a thickness of just under 250
microns. The fiber grooves 253 are spaced apart at regular intervals (b),
measured from center to center of adjacent grooves, the interval (B) equal
to about 250 microns. This spacing results in a surface distance (c)
between the grooves of about 80 microns. The pins 249 and fibers 246
20 are fixedly held between the alignment structures 238a,b by the epoxy fill
251 when hardened. The epoxy fill 251 also holds the alignment
structures 238a,b together in addition to forces provided by the mating
insert 238 when the structures 238a,b and insert .238 have been
assembled together. Optionally a "float" space 257 between the mating
zs insert 238 and alignment structure 238a,b, which is housed in an
aperture in the mating insert 238, can be provided to allow the pins 249
to carry out the final alignment without fighting the other alignment
features. The matting insert 224 of the connector 208 has the same
structure except that the pins 249 are replaced by sockets 228, which
3o may additionally have ferrules inserted therein.
With reference to Fig. 19, a second embodiment of the optical
connectors 208, 218 and alignment ferrule 210 will now be described in
further detail. The alignment ferrule 210 includes shutters 270a, 270b;
each mounted at opposite ends of the aperture 252 via respective biased
;~ hinges 272a, 272b and covering the aperture 252 in a closed position.
The shutters prevent particulate contaminants from entering the aperture
252, as well as solving a problem of "eye-safety" endemic in modern
optical communication systems due to the optical intensity used, by
CA 02320613 2000-09-21
52
blocking potentially harmful invisible infrared light emissions from a
partially equipped shelf (e.g. when, or after, a circuit card has been
removed from the midplane 206). Each shutter has a short side that
extends past its respective biased hinge and a long side that covers the
s aperture 252. The shutters have a dust seal 271 on their inner surfaces,
which prevents particulate contaminants from entering the aperture 252
of the alignment ferrule 210. The biased hinges 270a, 270b, are mounted
along an outer edge of the alignment ferrule 210 so that their longitudinal
axis are parallel to the midplane 206. An activation arm 274 disposed on
to the optical connector 208 opposite the circuit card 202 opens the shutter
270a outwardly against the closing force of the biased hinge 272x. This
is done by applying a force on the short side of the shutter 270a as the
optical connector 208 is brought into connection with the alignment
ferrule 210. The dust seals 271 come into contact with the exterior side
is face of the respective optical connector 208, 218 when the shutters
270a, 270b are fully open. This is to prevent contaminants from entering
the aperture 210 both on insertion and withdrawal of the optical
connectors 208, 218 and requires that, as well as cleaning the optical
face of the connector (as would be normal procedure), the area protected
20 by the shutter should be cleaned prior to circuit card insertion. Arrows
labelled A and B indicate the motion of the activation arm and the shutter
270a, respectively. An aperture 276 in the midplane 206 receives the
activation arm 274 as the optical connector 208 is inserted in the
alignment ferrule 210. The optical connector 218 has a similar activation
2s arm 278 for outwardly opening the shutter 272b against the closing force
of the biased hinge 270b (see arrows C and D). A similar aperture 280
receives the activation arm 278 when the optical connector 218 is
inserted in the alignment ferrule 210. When either of the optical
connectors 208 or 218 are removed from the alignment ferrule 210, the
3o respective shutter 270a, 270b returns to the closed position thereby
protecting the corresponding optical connector 218 or 208, and alignment
ferrule 210 from foreign matter.
Fig. 20, depicts a third embodiment of the optical connectors
208, 218 and alignment ferrule 210 in which a pair of outwardly opening
3s shutters 290a,b are mounted at each end of the aperture 252 via
respective biased hinges 294. Fig. 20 shows only one side of the
alignment ferrule 210, however identical shutters would also be included
on the other side. The biased hinges 294 bias the shutters to a closed
CA 02320613 2000-09-21
53
position such that the aperture 252 is covered by the shutters 290a,b.
The hinges 294 are mounted along an outside edge of the aperture 252,
their longitudinal axis parallel to the midplane 206 'when the alignment
ferrule 210 is mounted therein. The shutter 290a has an arm 291
s extending past its respective hinge 294 at an acute angle (e.g. 50
degrees) with respect to the alignment ferrule 210 for engaging the
activation arm 278. Arc-shaped members 296 having a toothed edge
296a are disposed at upper ends of the shutters 290a, 290b. The toothed
edge 296a of the member 296 disposed on the shutter 290a engages the
to corresponding toothed edge 296a of the member 296 disposed on shutter
290b, causing the shutter 290b to open outwardly when the shutter
290a opens outwardly. As the optical connector 218 is brought towards
the aperture 252, the activation arm 278 engages the shutter arm 291
causing the shutters 290a,b to open outwardly. In the opened position
is the shutters 290a,b reside alongside and in contact with either side of the
housing 236. Arrows labelled A and B denote the motion of the optical
connector 218 and shutters 290a/b, respectively. The shutters 290a,b
each include a dust seal 292 on their surface adjacent the aperture 252.
Each dust seal comes into contact with an exterior side of the connector
20 housing 236 when the shutter 290a,b is fully open. This helps to prevent
entry of contaminants into the aperture 210 both during insertion and
withdrawal of the connector 218. When the optical connector 218 is
removed, the shutters 290a,b return to their closed position with the dust
seal maintaining contact with the alignment ferrule 210 and between the
2s shutters 290a,b, thereby protecting the aperture 252 from foreign
material.
Fig. 21 shows a physical arrangement of a switching shelf 310 of
the switch 10. In this case the switch 10 has been partitioned into a
plurality of switching shelves 310. The physical arrangement of the
3o switch 10 comprising switching shelves will be described later. The
switching shelf 310 is similar to the switch 10 of Fig. 16 except that it is
configured to handle only 40 channel wavelengths instead of 160, which
reduces the demands on technology density for an easier implementation.
This reduction in channel wavelengths per shelf 310 also reduces the
35 number of fibers in the optical connectors 208, 218 and the alignment
ferrules 210, making the connectors 208,218 easier to manufacture. The
switching shelf 310 includes 10 includes ten switching circuit cards 216,
with each circuit card 216 having four optical switching matrices on-
CA 02320613 2000-09-21
54
board. This allows each switching circuit card 216 to switch four channel
wavelengths. The WD multiplexer 20 and WD demultiplexer 16, on the
I/O circuit card 202, each have a forty- channel wavelength capacity. Five
fiber shuffle (FS) modules 312 are included on each I/O circuit card 202
s to provide interconnection between the WD multiplexer 20 and WD
demultiplexer 16 and the optical connectors 208. Each optical connector
210 is an eight-way ribbon connector. The details of the fiber shuffle
module 312 are shown in Fig. 22. The controller circuit card 21 2 and
controller 26 would only control this switching shelf 310 and would
provide an interface to a central controller for controlling the entire switch
10.
The card 214 in Fig. 21 could be a wavelength converting switch
consisting of a number of receiver and transmitter transponders and an
electrical fabric between them. Preferably, it would be an interface and to
is a centrally located wavelength connecting switch 28 to prevent blocking
problems associated with a large number of small switches. Considering
the wavelength-converting switch 28 of Figure 4, i:he transponder
elements 32, 34 would preferably be on the card 2.14 with the electrical
switch 30 in a separate shelf, connected either electrically or via low cost
2o short wavelength, short reach ribbon optics. An optical crosspoint may be
included on the card 214 to permit tunable lasers to be connected into
various planes as needed, within the wavelength range of each tunable
laser. Considering the wavelength-converting switch 28 of Figure 10,
each group of channel converters 80 would be implemented as one card
2s 214. Considering the wavelength-converting switch 28 of Figure 1 1 , all of
the channel convertors 92 would be implemented in a separate shelf, with
the card 214 being an interface card with no functionality. Alternatively in
the latter case, we may place optical cross-connect elements for OXC A
100 and OXC-B 104 along with elements for the wavelength converter
30 108 on each card 214 and create a cabling juncture pattern at
interconnect C 1 10 between the cards. This would require that
interconnects A, B, D (102, 103, and 112) be partitionable into multiple
parallel circuit packs.
Fig. 22 illustrates the fiber shuffle module 312. The module 31 2 is
3s comprised of two fiber ribbon cables 312a and 31:?b, and has two input
ports and two output ports. Each cable 312a, 312b has eight optical
fibers and enters a respective input port of the module 312. The module
CA 02320613 2000-09-21
31 2 divides the cables so that four fibers from a cable go to one output
port and the remaining four fibers go to the other output port. The fiber
ribbon cable 312a is for connection of eight channel wavelengths from
the WD demultiplexer 16, at the input of the module 312, and for
s connection of two switching cards 216 at the output of the module. For
example, via ribbon cable 312a, channel wavelengths 1 to 8 are coupled
from the WD demultiplexer 16 to the first switching circuit card 216, for
channel wavelengths 1 to 4, and to the second switching circuit card
21 6, for channel wavelength 5 to 8. Similarly, via ribbon cable 312b,
o channel wavelengths 1 to 4, from the first switching circuit card 21 6, and
channel wavelengths 5 to 8, from the second switching circuit card 21 6,
are coupled to the WD multiplexer 20. There are five fiber shuffle modules
31 2 per I/O circuit card 202 in order to connect the forty channel
wavelengths multiplexed/demultiplexed by the card 202 to a respective
15 switching card 216.
Numerous modifications, variations, and adaptations may be made
to the particular embodiments of the invention described above without
departing from the scope of the invention, which is defined in the claims.
For example, the electrical inter-matrix switch 30 in Fig. 4 could be
2o replaced with an optical switch which can be made out of a multi-stage
array of MEMS devices since the cross-connect is sitting between
transponder banks and is not part of the line system optical reach budget
where loss is critical.
The optical loss of a 32 x32 MEMS is likely to be about 5-8 dB, so
~5 the insertion loss of one pass through a switching matrix 18 does not
approach the inter-amplifier link budget (approximately 24 dB~.
The plurality of receiver transponders 32 could convert optical
signals to short reach optical signals if the output signals of the cross-
connect switch 10 are only required to be routed 1:o terminal equipment at
3o the same node.
The leaf springs 234, 243 of the optical connectors 208,218 could
be any type of component suitable to provide flexible biasing of the
mating inserts 224, 238. For example, coil springs, pads of elastomeric
material, or formations of flexibly resilient plastic are a few of the many
;~ alternatives that could be used in place of the leaf springs 234, 243.
CA 02320613 2000-09-21
56
The slotted through holes 222,242 could be replaced by other
mounting means that allow some movement of the optical connector with
respect to the circuit card upon which it is mounted in order to align
mating faces 226, 240 with the respective sides of the aperture 252. For
s example, the circuit cards 202,216 could have slotted holes in which
pins, bolts, or rivets fastened securely to the optical connectors 208, 218
could move.
