Capella 2034
`JDS Uniphase v. Capella
`IPR2015-00739
`
`1
`
`

`
`U.S. Patent
`
`Jul. 31, 2012
`
`Sheet 1 of 15
`
`Us 8,233,794 B2
`
`
`
`Fig. 1a
`PRIOR ART
`
`2
`
`

`
`U.S. Patent
`
`Jul. 31, 2012
`
`Sheet 2 of 15
`
`Us 8,233,794 B2
`
`Fig. 1b
`PRIOR ART
`
`
`
`3
`
`

`
`U.S. Patent
`
`Jul. 31, 2012
`
`Sheet 3 of 15
`
`Us 8,233,794 B2
`
`PRIOR ART
`
`21 a
`21b
`21 c
`21 d
`21 e
`
`2";
`
`23
`
`E
`
`252:
`25b
`25c
`25d
`25e
`251‘
`
`First 513389
`(Mxl)
`
`Second stage
`(1xN)
`
`PRIOR ART
`
`Fig. 2a
`
`4
`
`

`
`U.S. Patent
`
`Jul. 31, 2012
`
`Sheet 4 of 15
`
`Us 8,233,794 B2
`
`21!: Q
`
`25
`
`21 d O
`213 O
`
`0 Input or output port
`0 Common port
`0 Block port
`
`Fig. 2b
`
`23
`
`21 b Q
`
`21 c
`
`21d 0
`21 e 0
`
`26
`
`0 Input or output port
`0 Common port
`0 Block port
`
`5
`
`

`
`U.S. Patent
`
`Jul. 31, 2012
`
`Sheet 5 of 15
`
`US 8,233,794 B2
`
`First stage
`
`Second stage
`
`O = Input or output port
`
`. = Block port
`
`
`
`32
`
`34
`
`6
`
`

`
`U.S. Patent
`
`Jul. 31, 2012
`
`Sheet 6 of 15
`
`US 8,233,794 B2
`
`First Stage
`
`Second stage
`
`O = Input or output port
`
`0 = Block port
`
`7
`
`

`
`U.S. Patent
`
`Jul. 31, 2012
`
`Sheet 7 of 15
`
`Us 8,233,794 B2
`
`Second stage
`
`O = Input or output port
`
`0 = Block port
`
`
`Fig. 6
`
`Second stage
`(dual 1 XN)
`
`First stage
`(Mx2)
`
`8
`
`

`
`U.S. Patent
`
`Jul. 31, 2012
`
`Sheet 8 of 15
`
`US 8,233,794 B2
`
`72
`
`O = Input or output port
`
`0 = Common port
`
`
`
`9
`
`

`
`U.S. Patent
`
`Jul. 31, 2012
`
`Sheet 9 of 15
`
`Us 8,233,794 B2
`
`1x3 WSS
`
`50 GHZ
`
`83
`
`
`
`(odd channels)
`
`100 GHz AWG
`
`85a
`
`Express
`channels
`
`100 GHz AWG
`
`(even channels)
`
`88
`
`89
`
`Fig. 8a
`
`
`
`10
`
`

`
`U.S. Patent
`
`Jul. 31, 2012
`
`Sheet 10 or 15
`
`US 8,233,794 B2
`
`86
`
`83
`
`85a
`
`85b
`
`850
`
`87a
`
`87b
`
`Fig. 8c
`
`91a
`
`91b O
`
`96
`
`Fig. 9a
`
`11
`
`11
`
`

`
`U.S. Patent
`
`Jul. 31, 2012
`
`Sheet 11 or 15
`
`Us 8,233,794 B2
`
`91c
`
`96
`
`I
`
`Fig. 9b
`
`12
`
`

`
`U.S. Patent
`
`Jul. 31, 2012
`
`Sheet 12 or 15
`
`Us 8,233,794 B2
`
`\
`
`S\
`
`O = Output port
`o = 1.1,, port
`
`Fm.uJ
`
`0,
`
`0 O
`
`O O O
`
`O O O
`
`O Q
`
`0
`
`/.
`
`101a
`
`101b
`
`1010
`
`101d
`
`101e
`
`101f
`
`101g
`
`101h
`
`10fi
`
`10H
`
`101k
`
`101|
`
`101m
`
`101n
`
`Second stage
`
`13
`
`

`
`U.S. Patent
`
`Jul. 31, 2012
`
`Sheet 13 or 15
`
`Us 8,233,794 B2
`
`
`
`Second stage
`(separate 1xN’s)
`
`115b
`
`
`
`14
`
`

`
`U.S. Patent
`
`Jul. 31, 2012
`
`Sheet 14 or 15
`
`Us 8,233,794 B2
`
`1319 0 ©1379
`
`O=Inputport
`
`15
`
`

`
`U.S. Patent
`
`Jul. 31, 2012
`
`Sheet 15 or 15
`
`Us 8,233,794 B2
`
`1418
`
`
`
`
`
`
`
`141b O O
`
`1410 O O 1470
`
`141d 0 O
`
`me Q»-
`
`141f Q Q 1471‘
`
`Q:1,,,,,,m
`
`O O
`
`=Outputport
`
`Fig. 14
`
`
`
`16
`
`

`
`US 8,233,794 B2
`
`1
`HITLESS MXN WAVELENGTH SELECTIVE
`SWITCH
`
`CROSS-REFERENCE TO RELATED
`APPLICATIONS
`
`The present invention claims priority from U.S. Provi-
`sional Patent Application No. 6lr'095,324, filed Sep. 9, 2008
`and U.S. Provisional Patent Application No. 61r"l 17,408, filed
`Nov. 24, 2008, both of which are incorporated herein by
`reference.
`
`TECHNICAL FIELD
`
`invention relates to optical wavelength
`The present
`switches configured for hitless or quasi-hitless operation in
`fiberoptic telecommunications, in particular to wavelength
`selective optical switches based on micro-electro -mechanical
`systems (MEMS) for use in reconfigurable optical add-drop
`modules (ROADMS) for dense wavelength division multi-
`plex (DWDM) systems.
`
`BACKGROUND OF THE INVENTION
`
`Advanced optical network technologies such as Dense
`Wavelength Division Multiplexing (DWDM) form the foun-
`dation for fiberoptic telecommunications networks, enabling
`worldwide traffic aggregation and metro and regional net-
`work consolidation. Such optical fiber networks often use
`reconfigurable optical add-drop modules (ROADMS) to
`deliver new flexibility to DWDM networks by enabling
`dynamic, transparent optical wavelength addfdrop fi.mction-
`mg.
`In general networks, the degree (D) of a network node is
`usually taken to mean a measure of how many network nodes
`are connected immediately adjacent to that node.
`Wavelength selective switch (WSS) technology incorpo-
`rated in ROADMs may use multi-degree (nD) ROADM
`architectures with a broadcast-and-select architecture. An
`optical splitter distributes wavelengths to a ‘drop’ path fixed
`wavelength demultiplexer and to each express direction. For
`each outgoing direction, a WSS is used to selectively combine
`‘add’ wavelengths from an ‘add’ path fixed-wavelength mul-
`tiplexer with channels selected from ch express direction.
`In many current architectures, a particular transmitter can
`send signals in only one output direction, towards only one
`adjacent network node. However, a ‘degreeless’ architecture
`is preferable, in which a particular transmitter can send sig-
`nals to any direction, that is, to any adjacent node. As the
`optical fiber networks evolve toward ‘degreeless’ architec-
`tures, many implementations of these architectures require a
`MxN WSS function, as described by Peter Roorda and Bran-
`don Collings (“Evolution to Colorless and Directionless
`ROADM Architectures", OFC 2008, paper NWE2). WSS
`technology is well-suited to extending ROADMS to allow
`automated assignment of the addfdrop wavelength, a func-
`tionality often referred to as colorless switching. Colorless
`ROADM architectures address the full automation of wave-
`
`length assignment, but the outbound direction of the tran-
`sponders remains fixed.
`For example, a power splitter may be used to broadcast the
`‘add’ wavelengths to a WSS for each direction, and another
`WSS is used to select the direction for the associated ‘drop’
`wavelength. Using this architecture with a colorless MUXJ
`DEMUX and amplifiers for compensating insertion loss, an
`‘addfdrop’ port can be assigned to any wavelength and
`coupled to any direction in a firlly automated fashion.
`
`10
`
`15
`
`20
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`2
`
`A basic simplified structure ofa I xN WSS using arrays of
`adj ustable reflectors. The adjustable reflectors may be MEMS
`minors that can be tilted in 1-dimension about one axis, as
`shown in top view in FIG. 1a. This has been described by
`Ducellier et al. in U.S. Pat. No. 6,707,959 issued Mar. 16,
`2004, which is incorporated herein by reference. Port switch-
`ing in the plane of the drawing is effected by suitably tilting
`MEMS mirrors of a modifying (MEMS) array about an axis.
`In this example according to prior art, a basic simplified
`wavelength switching module 102A comprises a light redi-
`recting element, such as a spherical reflector 120, used to
`receive a beam of light comprising wavelength multiplexed
`signals from a fi'ont-end unit 122 and to re-image the beam
`onto a micro-electro-mechanical systems (MEMS) array 126
`aflerreflection offa diffraction grating 124. Due to the optical
`dispersion of the diffraction grating 124, a separate image is
`formed on the MEMS mirror array 126 for every wavelength
`multiplexed signal present in the beam of light.
`Each MEMS mirror of the array is arranged so that it
`reflects the image corresponding to a wavelength multiplexed
`signal back to the front-end unit 122 via the diffraction grat-
`ing 124 and the spherical refiector 120. The mirrors are fab-
`ricated to enable tilting about an axis perpendicular to the
`plane of FIG. 1a by means of a suitable controller.
`FIG. lb shows the front end of the WSS of FIG. 1a in
`greater detail. Four ports can be used for inputs or outputs
`with MEMS mirrors used for coupling a particular wave-
`length multiplexed signal between any two ports. For
`example, this WSS could be used as a 3x1 switch, i.e. with 3
`input ports and 1 output port.
`In FIG. 1b, an optically equivalent from end of the wave-
`length switching module 102A of FIG. 1a comprises four
`inputfoutput ports 132A-D such as optical fibers, each carry
`wavelength multiplexed signals. The light beams from the
`optical fibers 132A-D are collimated by lenses 134A-D
`before passing through a switching lens 136, which converts
`the spatial separation between the ports 132A-D to an angular
`separation at an intersection point 150. Since the rest of the
`optics in the wavelength switching module 102A serve to
`re-image intersection point 150 onto the MEMS array 126,
`for the purposes of this description each tilting MEMS minor
`corresponding to a particular wavelength multiplexed signal
`can be considered as being located at the intersection point
`150.
`
`Details of the imaging and dispersing optics are well
`known in the art, for instance as described by Bouevitch et al.
`in U.S. Pat. No. 6,810,169 issued Oct. 26, 2004, which is
`incorporated herein by reference.
`FIG. lc illustrates in greaterdetail such a prior art 2x2 WSS
`structure based on tilting MEMS minor arrays in conjunction
`with optical circulators. Two optical inputs, ‘IN’ 11 and
`‘ADD’ 21, carrying wavelength multiplexed signals entering
`bi-directional ports 31 and 32 through circulators 10 and 20
`are focused by lens 35 into beams 41 and 42, respectively,
`onto an intersection point N. A concave mirror 40 re-images
`intersection point N via a diffraction grating 50 and transmis-
`sion path correction element 100 onto MEMS tilting mirrors
`61, 62 of a MEMS array 60. After reflection off the MEMS
`tilting minors 61, 62 the beams return by essentially the same
`route to the intersection point N from where they are colli-
`mated into the bi-directional ports 31 and 32, fortransmission
`through circulators 10 and 20 into ‘EXPRESS’ 12 and
`‘DROP’ 22 outputs, respectively.
`There are essentially two possible choices for the axis
`about which the MEMS minors are tilted: vertical or hori-
`zontal. In principle there is no difference between the two,
`however factors such as optical beam cross-section, spot
`
`17
`
`17
`
`

`
`US 8,233,794 B2
`
`3
`shape, preferred operating configuration, switch module
`geometry and similar would in practice determine the choice.
`Thus FIG. lb could be regarded as a top view for the case
`where the MEMS mirrors are tilted about a vertical axis,
`whereas the same figure can be regarded as side view for the
`case where the MEMS minors are tilted about a horizontal
`
`axis. The latter case will be used without loss of generality in
`the following description with the understanding all the
`embodiments would be equally functional for the vertical axis
`case.
`
`FIG. 2a shows a possible way of configuring a MxN WSS
`from a M><l WSS and a 1xN WSS. This example illustrates 5
`input ports and six output ports forming a 5x6 WSS. In this
`configuration, the single output of the Mxl WSS and the
`single input of the 1xN WSS are connected together.
`Each wavelength multiplexed signal can be routed inde-
`pendently, however only one instance of each wavelength
`multiplexed signal can be passed from an input port to an
`output port. In other words, the configuration exhibits “wave-
`length blocking", which poses an obstacle to achieving true
`arbitrary configurability.
`As achieving true arbitrary configurability is considerably
`more difficult, network designers are likely to accept the
`wavelength blocking restriction in their designs.
`Another potentially more serious problem with the con-
`figuration of FIG. 2a is “hitting” during switching operations,
`which results in unwanted transient signals appearing at the
`output ports during switching operations.
`It is an object of the invention to provide an MxN WSS
`which performs wavelength switching without “hitting”, i.e.
`a hitless MxN WSS.
`
`A fitrther object is to provide a hitless M:-<N WSS that can
`take advantage of low-cost, manufacturable MEMS mirrors
`which are tilted only in I-dimension.
`Another object of the invention is to provide a method of
`operating such MxN WSS in a hitless or qua si-hitless manner.
`
`SUMMARY OF THE INVENTION
`
`Accordingly, the present invention relates to an optical
`switching device which performs wavelength switching with-
`out “hitting”.
`An aspect of the present invention relates to a first stage
`Mxl WSS concatenated with a second stage 1xN WSS using
`1D tilting MEMS mirror arrays as switching elements,
`employing one block port on the first stage WSS to obtain
`quasi-hitless M><N switching.
`Another feature of the present invention provides for addi-
`tional block ports in either the first stage or the second stage
`to obtain completely hitless MxN switching.
`In particular, the present invention relates to an optical
`switching device, comprising a first wavelength selective
`switch stage with a plurality ofinput ports, each input port for
`receiving a respective DWDM signal; a first common port for
`transmitting one of the respective DWDM signals; a first
`adjustable reflector for selecting an optical path for a selected
`one of the respective DWDM signals between one of the
`plurality of input ports and the first common port; and a first
`input block port disposed between the first common port and
`at least one of the input ports for providing the first adjustable
`reflector an interim position in which no signal is input to the
`first output port.
`The optical switching device also includes a second wave-
`length selective switch stage comprising a second common
`port optically coupled to the first common port for receiving
`the selected DWDM signal; a plurality of output ports for
`transmitting the selected DWDM signal; a second adjustable
`
`10
`
`15
`
`20
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`4
`
`refiector for selecting an optical path for the selected DWDM
`signal between the first common port and one of the plurality
`of output ports; and a switch controller for optically coupling
`the block port and the first common port when the second
`adjustable reflector is adjusted between settings.
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`The invention will be described in greater detail with ref-
`erence to the accompanying drawings which represent pre-
`ferred embodiments thereof, wherein:
`FIG. 1a is a top view schematic diagram of a prior art
`wavelength selective switch (WSS);
`FIG. lb is simplified equivalent schematic diagram of a
`prior art WSS;
`FIG. 1c is a top view of a prior art WSS showing greater
`detail;
`FIG. 2a is a block diagram ofa Mxl WSS and a lxN WSS
`concatenated to create a MxN WSS functionality;
`FIGS. 2!: and 2c are schematic cross-sectional views at
`A-A’ from point 150 of FIG. lb showing input, output and
`block ports ofa hitless WSS in a ‘block’ and ‘transmit’ setting,
`respectively, according to the present invention;
`FIGS. 3 and 4 are cross-sectional views at A-A’ from point
`150 of FIG. lb showing input, output and block ports of a
`fiilly hitless WSS according to the present invention;
`FIG. 5 is a cross-sectional view at A-A’ from point 150 of
`FIG. lb showing input, output and block ports of a quasi-
`hitless WSS according to the present invention;
`FIG. 6 is a block diagram ofa cascaded Mxz and dual 1xN
`WSS to provide Mx2N WSS functionality according to the
`present invention;
`FIG. 7 is a cross-sectional view at A-A’ from point 150 of
`FIG. lb of the first and second stage WSS's of FIG. 6 accord-
`ing to the present invention;
`FIG. 8a is a block diagram of a network node design using
`a 1x3 WSS for odd-even separation of channels.
`FIGS. 8!: and 8c are simplified drawings of a 1x3 WSS
`showing two approaches to making it hitless;
`FIGS. 9a-9c are simplified drawings of a 3x2 WSS with a
`single 1-dimensionally tilting MEMS minor array shown in a
`‘block’ setting, an input port coupled to one of the output
`ports, and the same input port coupled to the other output port;
`FIG. 10 is a simplified drawing ofa dual 1xN WSS using a
`single lD tilting MEMS mirror array according to the present
`invention;
`FIG. 11 is a block diagram of a cascaded Mx2 and two
`separate l><N WSS‘s to provide M><2N WSS fiinctionality
`according to the present invention;
`FIGS. 12a and 125 are simplified drawings of the Mx2
`WSS as used in FIG. 11, showing two different pairs of inputs
`coupled to the pair of outputs, respectively;
`FIGS. 13 and 14 are simplified drawings of an alternative
`port configuration according to the present invention for a
`dual I ><N WSS using a single ID tilting MEMS mirror array
`where ports are arranged in two parallel rows rather than one
`single row.
`
`DETAILED DESCRIPTION
`
`To address the problem associated with “hitting" in wave-
`length selective switches (WSS), the present invention pro-
`vides structures modified to include block ports and methods
`of operating them in fully hitless as well as quasi-hitless
`modes.
`For example, “hitting" is said to occur when a particular
`wavelength multiplexed signal gets switched from one output
`
`18
`
`18
`
`

`
`US 8,233,794 B2
`
`5
`
`port to another output port, such that interfering signals
`momentarily appear at intervening output ports during the
`switching operation.
`In conventional 1 ><N WSS devices based on MEMS mirror
`arrays, hitless operation can be achieved by using biaxial
`tilting MEMS minors. While normally a Mxl or a lxN WSS
`device results in a hitting WSS, in this disclosure it is shown
`how a cascaded configuration of two WSSS using uniaxial
`tilting MEMS mirrors can be used for hitless or quasi-hitless
`operation. It is necessary to design the WSS devices appro-
`priately and operate them in the correct sequence to achieve
`hitless operation.
`Furthermore, as described below, there are various options
`to achieve hitless operation. Options are described which
`allow the greatest number ofusable input and output ports for
`a given MEMS mirror tilt angle capability.
`FIG. 2a shows a possible way of configuring a MXN WSS
`from a M><l WSS and a lxN WSS. This example illustrates 5
`input ports and six output ports forming a 5x6 WSS.
`It has a first stage comprising a Mxl WSS 22 and a second
`stage comprising a lxN WSS 24 concatenated through a
`common port to achieve the required MxN functionality. M
`denotes the number of input ports 21a-21e and N denotes the
`number ofoutput ports 25a-25;". In this exemplary case, M=5
`and N=6, giving a 5x6 WSS.
`Each of the input ports 21a-2le can accept a lightwave
`beams comprising several wavelength multiplexed signals, as
`is typical in a DWDM network. The Mxl WSS 22 of the first
`stage comprises a MEMS mirror for each wavelgth in its
`operating wavelength band. By appropriate angling or tilting
`under the control ofa first controller (not shown), each mirror
`selects a corresponding wavelength multiplexed signal from
`one of the M input ports 21¢:-21e and reflects it onto the
`common port 23. The 1xN WSS 24 performs this process in
`reverse by receiving the input wavelength multiplexed signals
`selected by the first stage Mxl WSS 22 from the common port
`23. Each corresponding MEMS mirror reflects to one ofthe N
`output ports 25a-25fby appropriate angling or tilting under
`the control of a second controller (not shown).
`One concern for network operators is “hitting” during
`switching operations. “Hitting” occurs when spurious tran-
`sient signals appear at the output ports during switching
`operations.
`In certain network configurations, “hitting” during switch-
`ing operations can cause interference in other signals. For
`example, in FIG. 2a, if in the second stage WSS 24 a wave-
`length multiplexed signal is switched from output port 25::
`to output port 25c, it will momentarily appear on output port
`25b. Another wavelength multiplexed signal k may also be
`routed to output port 256, carrying signal traific. Ifoutput port
`25!: is connected to a receiver with no wavelength discrimi-
`nation, the wavelength multiplexed signalj will interfere with
`the reception ofthe wavelength multiplexed signal k, causing
`a temporary network interruption for the signal at wavelength
`k.
`
`This is an unacceptable situation for network operators.
`Note that even if a signal at a particular wavelength on a
`particular input port is not being used, it may be present and
`therefore could cause interference if it is inadvertently routed
`to an undesired output port during a switching operation. In
`general one must assume that all possible wavelength signals
`are present at all inputs.
`This problem can be explained using the following
`example. FIG. 2b shows port detail of the first stage WSS in
`FIG. 2a as a schematic cross -sectional view of switch ports on
`the front end, similar to the view seen at A-A‘ fi'om point 150
`in FIG. lb. In accordance with this invention, a block port 2'.-'
`
`10
`
`15
`
`20
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`6
`has been added to the priorart first stage WSS in FIG. 2a. The
`five input ports 21a-212, the output or common port 23 and
`the block port 2'.’ appear end-on in a collinear arrangement.
`For the purposes of this example, an adjustable refiector such
`as a MEMS mirror 26 is also included, although it would in
`fact be located above the plane of the drawing. When tilted
`about an axis parallel to the plane of the drawing, the MEMS
`mirror 26 can provide optical coupling for a lightwave hm
`between the respective ports. The MEMS minor 26 may also
`be tiltable about another axis or even two axes in two dimen-
`sions depending on the functionality required by a particular
`application.
`In FIG. 2b the MEMS mirror 26 is configured in a setting
`where the block port 27 is coupled to the output port 23. Now,
`as the MEMS mirror 26 is reconfigured to the setting shown
`in FIG. 2c where input port 21c is coupled to the output port
`23, input ports 21:: and 211:: will momentarily be coupled to
`the output port 23. This momentary appearance of signals
`from input ports 21a and 21b at the output port 23 is called
`“hitting”.
`In this example the MEMS mirror 26 must have sufficient
`angulartilt range to couple 6 ports (input ports 21a-212, block
`port 27) to the output port 23. The block port 2'? could be a real
`port in the WSS 22, chosen to be unused in a particular
`application, or the lens and fiber that would nomially occupy
`this location could be absent. It is important that even if the
`block port 27 is not physically present, the MEMS mirror 26
`must have suflicient angular tilting range to access the block
`port 27 if the block function is required in the WSS 22 (no
`input to output couplings). In FIGS. 2b and 2c, the optical
`paths could be reversed, such that the output port 23 is used as
`an input port, and input ports 21a-212 become output ports; in
`this case the device is referred to as a lxN WSS (1 x5 in the
`example of FIGS. 2!: and 2.9).
`Using the WSS 80 presented in FIGS. 8a and 8b as an
`example, a possible way of preventing “hitting” is described
`here. A wavelength multiplexed signal or an optical channel
`beam entering through input port 83 is redirected by a biaxi-
`ally tilting MEMS mirror 86 from output port 85:: to 85¢
`along a ‘2-dimensional path’ 87¢ as illustrated in FIG. 8b.
`Before tilting the MEMS minor 86 about a first axis to per-
`form switching, a controller mirror first tilts the MEMS mir-
`ror 86 about a second axis to direct the optical channel beam
`away from the output ports, so as to prevent the optical chan-
`nel beam from hitting output port 855;.
`However, for low manufacturing cost it is desirable to
`obtainhitless operation ofthetwo stage MxN WSS ofFIG. 2a
`with only I-dimensional tilting MEMS. In accordance with
`instant invention, this can be accomplished by incorporating
`one or more block ports into the constituent WSS’s.
`FIG. 3 is a view similar to those in FIGS. 2!; and 2c,
`howeverthe port configuration shown provides hitless opera-
`tion ofa 4x5 WSS. The first stageWSS 32 has input ports 31a,
`31c, 31d, 31], two block ports 31b, 31c, and an output or
`common port 31g, while the second stage WSS 34 has five
`output ports 35b-35;”, one block port 35a, and an input or
`common port 35g. The first stage WSS 32 and the second
`stage WSS 34 is concatenated by connecting the output port
`31g with the input port 35g.
`Of course the number of ports can be increased or
`decreased depending on the angular tilt range of the MEMS
`mirrors and the number ofports required in a particular appli-
`cation. In FIG. 3, the first stage WSS 32 has a block port 31!:
`(3142) adjacent to every input port 31a, 31-: (Sold, 31)‘), respec-
`tively. As can be seen fiom FIG. 3, block ports between each
`pair of input ports are not necessary, for example there is no
`block port between input ports 31c and 31d. The switching
`
`19
`
`19
`
`

`
`7
`
`US 8,233,794 B2
`
`between ports is effected by means ofthe adjustable reflectors
`or MEMS mirrors in the first and second stage under the
`control of a switch controller 39.
`
`LNINJ
`
`During reconfiguration of a particular wavelength multi-
`plexed signal frorn output port i to output port j, interfering
`signals at the same wavelength will momentarily appear at
`output ports i and j. This is ‘quasi-hitless’ operation. This
`should not affect any other network traffic, since that wave-
`length is unused on ports i and j during the switching opera-
`tion, and is intentionally present on those ports before and
`afier the switching operation. The only potential issue may be
`security, if the wavelength multiplexed signal momentarily
`routed to the incorrect ports is of a confidential nature and if
`there is a possibility that it can be intercepted by an unin-
`tended receiving party. The switching procedure is:
`1. switch first stage 52 to the block port 51:: (interfering
`signals at the same wavelength from other input ports
`will momentarily appearat the currently configured out-
`put port during this operation);
`. switch second stage 54 to a desired output port; and
`. switch first stage 52 to a desired input port (interfering
`signals at the same wavelength from other input ports
`will momentarily appear at the final output port during
`this operation).
`The arrangement of FIG. 5 achieves the greatest possible
`numberof input and output ports for a given MEMS tilt angle
`range, since only one port needs to be reserved as a block port.
`Note that an arrangement where the block port is on the
`second stage WSS 54 does not provide hitless switching.
`FIG. 6 shows a further arrangement which increases the
`number of available output ports without increasing the
`required MEMS mirror tilt angle range. A cascaded MX2
`WSS is cascaded with a dual lxN WSS to provide Mx2N
`WSS functionality. This configuration provides double the
`number of output ports normally addressable by a single
`MEMS mirror, at the expense of halving the number of avail-
`able input ports.
`The first stage WSS 62 is configured as an Mx2 WSS with
`M input ports 61, while two second stage WSS 64a and 64.’:
`are configured as a dual 1xN WSS, each with N output ports
`for a total of 2N output ports 65. The two stages are intercon-
`nected with two common ports 63.
`FIG. 7 illustrates the port configurations needed to achieve
`the Mx2N function, 3x12 in this example. The first stage
`WSS 72 comprises three input ports 71b-71a‘ and two output
`ports 71a and 71e, while the dual 1x6 second stage WSS 74
`comprises two input ports 75a, 75}: and twelve output ports
`755-753 and ‘75t'-75». To interconnect the two stages output
`ports 71:: and 712 are connected to input ports 75:1 and 75h,
`respectively. One MEMS mirror simultaneously provides an
`optical connection between the input por1 75a and the six
`output ports 75b-75g as well as between the input port 75}:
`and the six output ports 75:‘-75:1.
`In a particular fiber optic network node configuration used
`by some network suppliers, a 1x2 WSS is used to select
`signals on wavelength multiplexed channels which are to be
`sent to an express port and channels which are to be dropped
`at that node. For further demultiplexing the dropped channels
`into separate receivers, an array waveguide (AWG) can be
`used. With channel spacings of 50 GHz, it is common to use
`an interleaver to separate the channels into ‘odd’ and ‘even’
`channels onto separate fibers, the odd channels and the even
`channels now being spaced by 100 GHZ.
`This is advantageous because it is much sierand cheaper
`to manufacture anAWG for separating 100 GHZ spaced chan-
`nels compared to an AWG for separating 50 GHZ spaced
`channels. The cost savings in the AWG’s offsets the cost of
`the added interlver. This type of node configuration is used
`rather than simply a 1><N WSS because, particularly for net-
`works with 50 GHZ channel spacing, the number of dropped
`
`10
`
`15
`
`At least one third of the potential input ports must he used
`as block ports to achieve one block port adjacent to every 5
`input port. The second stage WSS 34 has only one block port
`35a. The switching procedure for reconfiguring a particular
`wavelength from a first combination of input and output ports
`to a second combination of input and output ports is:
`1. switch first stage 32 to a block port adjacent to the current
`input port;
`2. switch second stage 34 to the block port 35a:
`3. switch first stage 32 to a block position adjacent to the
`final input port;
`4. switch second stage 34 to the desired output port; and
`5. switch first stage 32 to the desired input port.
`For example, if the initial setting for a wavelength multiplex
`signal j couples input 31d to output 35;‘; and the desired
`setting afier reconfiguration is to connect input 31:: to output
`35b, the switching procedure would be:
`1. switch first stage 32 from input port 31d to block port
`31:2;
`2. switch second stage 34 from output port 35fio block port
`35a;
`3. switch first stage 32 from block port 312 to block port
`31b;
`4. switch second stage 34 from block port 35a to output
`port 35!); and
`5. switch first stage 32 from block port 31b to input port
`310.
`
`20
`
`30
`
`With this sequence, at no time do any undesired signals
`emerge fi'om any output port. In FIG. 3, both the first stage
`WSS 32 and the second stage WSS 34 must have a MEMS
`mirror tilt angle range capable ofaddress ing six distinct ports;
`the first stage has only 4 usable input ports because two port
`locations must be reserved as block ports, and the second
`stage has only five usable output ports because one port must
`be assigned as a block port.
`FIG. 4 shows a similar arrangement, this one providing
`hitless operation for a 5x4 WSS. The first stage WSS 42 has
`input ports 41b-41_f, an output port 41g, and a block port 41a,
`while the second stage WSS 44 has five output ports 45a, 45c,
`45d and 45f, an input port 45g and two block ports 45!), 452.
`The switching between ports is effected by means of the
`adjustable reflectors or MEMS mirrors in the first and second
`stage under the control of a switch controller 49.
`The switching procedure for this configuration is:
`1. switch second stage 44 to a block port adjacent to the
`current output port;
`2. switch first stage 42 to the block port 41a,-
`3. switch second stage 44 to a block position adjacent to the
`final output port;
`4. switch first stage 42 to the desired input port; and
`5. switch second stage 44 to the desired output port.
`The main difference between the configurations of FIGS. 3
`and 4 is that the one in FIG. 3 provides the greater number of
`output ports, while the one in FIG. 4 provides the greater
`number of input ports.
`FIG. 5 shows another arrangement which has only one
`block port, yet still allows ‘quasi-hitless’ operation. The first
`stage WSS 52 has five input ports 51b-5 lfi an output port 51g,
`and a block port 51a, while the second stage WSS 54 has six
`output ports 55a-45fl an input port 553, and no block ports.
`The switching between ports is efiected by means of the
`adjustable reflectors or MEMS mirrors in the first and second
`stage under the control of a switch controller 59.
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`20
`
`20
`
`

`
`US 8,233,794 B2
`
`9
`channels (which could be 40 or more at a single node) can
`exceed the current capability for l><N WSS (which is in the
`range of 9 to 16 ports). For some WSS manufacturing meth-
`ods, for example those using liquid crystal polarization
`switching, the complexity and cost of the WSS increases as
`the port count is increased, because of the need to add addi-
`tional liquid crystal switching elements to accommodate an
`increased number of output ports.
`For WSS devices using tilting MEMS minor arrays how-
`ever, the addition of extra ports only requires an increased
`MEMS mirror tilt angle to be able to select among an
`increased number ofports, and thus the added complexity and
`cost is minimal. For a MEMS-based WSS, the cost difference
`between a 1x2 WSS and a 1x3 WSS is very small. Therefore,
`if a MEMS-based WSS is used, thenode configuration shown
`in FIG. 8a becomes preferable, eliminating the need for an
`interleaver. The added cost ofthe 1x3 WSS comparedto a 1x2
`WSS is less than the cost of the interleaver, resulting in a
`lower overall cost and eliminating the insertion loss of the
`interleaver.
`
`In the configuration of FIG. 8a, a DWDM signal which is
`carrying traflic on several wavelength multiplexed channels
`within a given wavelength band at a 50 GHZ spacing is
`applied to an input port 83 of a 1x3 WSS 80. Only odd-
`numbered channels 855 are directed to the ‘odd’ port, and
`only even-numbered channels 85:: are directed to the ‘even’
`port. Express channels 85a are directed to the ‘express’ port.
`As the odd-numbered channels 85!) are spaced at 100 GH2, a
`100 GHz AWG 81 may be used to produce demultiplexed
`channels 88. Similarly the even-numbered channels 85:: are
`demultiplexed by a 100 GI-I2 AWG 82 into demultiplexed
`channels 89.
`
`One potential difficulty with a 1x3 WSS based on MEMS
`tilting minors is