|||||||l||||||||||||||||||||||||||||||||||||||||||||||||||||||||l|||||l|||
`
`USUOG’198941B2
`
`(12) United States Patent
`Smith et a1.
`
`(10) Patent N0.:
`(45) Date of Patent:
`
`US 6,798,941 B2
`Sep. 28, 2004
`
`(54) VARIABLE TRANSMISSION MULTI-
`CHANNEI. OPTICAL SWITCH
`
`("15]
`
`Inventors: David A. Smith, Atlanta, GA (US);
`John E. Golob, Atlanta, GA (US);
`Fnriborr. Fat-halt, Alpharetta, GA (US)
`
`(73) Assignee: Mover. Networks, Inc., Norcross, GA
`(US)
`
`( * ~J Notice:
`
`Subject to any disclaimer, the term of this
`patent is extender] or adjusted under 35
`U.S.C. 154{b) by 78 days.
`
`(21] App}. No: 091951312
`
`(22]
`
`(65]
`
`((10)
`
`Filed:
`
`Sep. 20, 2001
`
`Prior Publication Data
`
`US 2113210011627 A1 Jun. 13, 2002
`
`Related U.S. Application Data
`Provisional application No, 601234,683, filed on Sep. 22,
`20110, and provisional application No. 61]126'?,285, tiled on
`Feb. '1'. 21.101.
`
`Int. Cl.’
`(51)
`(52) U.S. CI.
`(58)
`Field of Search
`
`G028 6126; GUEB (1142
`385118; 385119
`
`385118, 19, 15,
`385133; 3981156
`
`(56]
`
`References Cited
`U.S. PATENT DOCUMENTS
`
`4,596,062 A ‘
`4,911,526 A
`5,414,540 A
`5,621,829 A
`5,745,271 A
`5,771,320 A
`5,796,479 A
`5,828,300 A
`5,915,063 A
`5,960,133 A
`
`3981156
`911987 IaBudde
`350196.24
`311991} l-Isu et al.
`359131..1
`.
`5111195 Patel et at.
`41109? Ford ............. 385122
`411998 Ford ct al.
`3591130
`611098 Stone
`335116
`811998 Derickson et at.
`3561326
`1011998 Itenryet at.
`385120
`611099 Colhotlrnc 1:1 :1].
`3351140
`911999 'I‘onlilinson
`385118
`
`
`
`
`
`6,091,859 A
`385117f
`9112000 Solgaard et al.
`6,101.25“) A
`385116
`812000 Laor
`..........
`
`(1,111,686 A *
`. 359133113
`312111.11
`'royohara
`
`6,204,946 Bl
`11116361116161.
`.
`.. 3591131
`312001
`
`6,263,133 131
`1126111 Bishopelal.
`385115
`
`FOREIGN l’tflt'l‘liN'l~ DOCUMEN'J‘S
`
`1:1I
`1313
`JP
`wo
`wo
`
`o 984 311
`o 729 044
`55159402
`W10 02125358
`wo 00120899
`
`312111111
`112003
`1211980
`31201111
`41211111
`
`O'l'l—IER PUBLICATIONS
`
`Lin et 211.. “Micro—electro—mechanical systems (MEMS) for
`WDM optical-crossconnect networks“. Militoty ("011111111111-
`cott’ort Conference Proceedings, 1909. MILCOM 1909.
`IEEE Atimtfic City, NJ. USA 31, IEEE. US, Oct. 31. 199‘),
`954—957 pp.
`G. Jeong et £11., "Comparison of Wavelength—interchanging
`and wavelength—selective cross—connects
`in mulliwave-
`length all—optical networks". IEEP. 1996. 156—163 pp.
`
`* cited by examiner
`
`Pit-1111611111 Exonn‘ner—Chandrika Prasad
`(74) 111101719» Agent, or Firm—Charles S. Guenzcr
`
`(57)
`
`ABSTRACT
`
`A meld—wavelength or white—light optical switch including
`an array of mirrors tillahle about tWo axes, both to control
`the switching and to provide variable power transmission
`through the switch, both for optimization and for power
`equalization between wavelength channels in a multi-
`wavelcngth signal. The output punter ot' a channel
`is
`monitored,
`thereby allowing feedback adjustment of the
`transmitted power. The mirrors are preferably formed in a
`micro electromechanies system array to he tiltable in
`orthogonal directions and having electrostatically controlled
`tilting by two pairs of electrodes beneath the mirrors. Input
`power of the separate channels may also be monitored.
`
`37 Claims, 11 Drawing Sheets
`
`
`
`FNC 1009
`
`

`

`US. Patent
`
`$1313.28, 2004
`
`Sheet 1 0f 11
`
`US 6,798,941 B2
`
`
`
`FIG.
`
`1
`
`24-
`
`24
`
`24
`
`22
`
`(PRiOR ART)
`
`FIG. 2
`
`26
`
`25
`
`25
`
`

`

`US. Patent
`
`Sep. 28, 2004
`
`Sheet 2 0f 11
`
`US 6,798,941 B2
`
`
`
`FIG. 3
`
` w
`
`(PRIOR ART)
`
`FIG. 4
`
`

`

`US. Patent
`
`$013.28, 2004
`
`Sheet 3 0f 11
`
`US 6,798,941 132
`
`72
`
`
`
`
`
`
`
`"""'"-'
`'flll!
`athmw
` '.
`
`
`
`
`
`
`
`:'-"=‘
`
`
`
`

`

`US. Patent
`
`Sep. 28, 2004
`
`Sheet 4 0f 11
`
`US 6,798,941 132
`
`
`NETWORK
`
`FROM
`
`TO
`NETWORK
`
`SWITCH
`
`COMMANDS
`
`140
`
`/
`
`

`

`US. Patent
`
`$913.28, 2004
`
`Sheet 5 0f 11
`
`US 6,798,941 132
`
`160
`
`
`
`FIG.
`
`

`

`US. Patent
`
`Sep. 28, 2004
`
`Sheet 6 0f 11
`
`US 6,798,941 132
`
`170
`
`
`
`SWHCH
`COMMANDS
`155
`158
`234
`‘
`
`
`228
`
` CONTROLLER
`
`FROM
`
`NETWORK
`
`

`

`US. Patent
`
`Sep.28,2004
`
`Sheet? ofll
`
`US 6,798,941 B2
`
`
`
`164 166
`
`

`

`US. Patent
`
`801). 28, 2004
`
`Sheet 8 0f 11
`
`US 6,798,941 132
`
`240
`
`
`
`

`

`US. Patent
`
`Sep. 28, 2004
`
`Sheet 9 0f 11
`
`US 6,798,941 B2
`
`260
`
`
`
`

`

`US. Patent
`
`Sep. 28, 2004
`
`Sheet 10 of 11
`
`US 6,798,941 B2
`
`
`
`

`

`US. Patent
`
`Sep. 28, 2004
`
`Sheet 11 0f 11
`
`US 6,798,941 B2
`
`
`*-3'!2
`
`300
`
`

`

`US 6,798,941 B2
`
`1
`VARIABLE TRANSMISSION MULTI-
`CHANNEL OPTICAL SWITCH
`
`REL/(I‘ll!) APPLICATIONS
`
`This application claims benefit of U.S. Provisional Appli-
`cations Nos. $934,683. filed Sep. 22, 2000, and 603'267,
`285, filed Feb. 7, 200].
`
`BACKGROUND OF THE INVENTION
`
`1. Field of the Invention
`
`The invention relates generally to optical switches. In
`particular, the invention relates to optical switches used in
`mu lti—channel optical communications networks and having
`controlled transmissivity for ditlerent channels.
`2. Background Art
`Modern communications networks are increasingly based
`on silica optical fiber. which ofi'ers very wide bandwidth
`including several transmission bands usable for communi-
`cations. In a conventional point-to-poinl optical communi-
`cations link, at the transmitter, an electrical data signal is
`used to modulate the output of a semiconductor laser
`emitting, for example, in the 155(tntrt band. and the modu-
`lated optical signal is impressed on one end of the silica
`optical fiber. Other transmission bands at 850 um and 1310
`nm are also available. On very long links, the optical signal
`may he amplilied along the route by one or more optically
`pumped erbium—doped fiber amplifiers (EDFAs) or other
`optical amplifiers. Al the receiver, the optical signal from the
`fiber is detected, for example, by an optical p-i-n diode
`detector outputting an electrical signal in correspondence to
`the modulating electrical signal. The transmission band—
`width of such systems is typically limited by the speed of the
`electronics and opto-electronics included in the transmitter
`and receiver. Speeds of 10 gigabits per second (Gbs) are
`available in fielded systems, and 40 Ghs systems are reach—
`ing production. l-‘urther increases in the speed of the elec-
`trnnics will be difficult. These speeds do not match the
`bandwidth inherent in the fiber, which is well in excess of
`one terabit per second. Furthermore, such fast optical trans-
`mitters and receivers are expensive and may require special
`environmental controls.
`
`Transmission capacity of fiber systems can be greatly
`increased by wavelength division multiplexing [WDM) in
`which the optical signal is generated in a transmitter includ-
`ing multiple semiconductor lasers emitting at different
`respective wavelengths within the transmission band. The
`1550 nm transmission band has a bandwidth of about 35 nm,
`determined by the available amplification band ol'an EDFA.
`Other amplifier types and amplification bands are being
`commercialized so that
`the available WDM spectrum is
`growing each year. In a WOM system, each laser is modu-
`lated by a dilT'erent electrical data signal, and the ditferent
`laser outputs are optically combined (multiplexed) into a
`multi-wavelength optical signal which is impressed on the
`optical fiber and which together can be amplified by an
`EDFAwithout the need to demultiplex the optical signal. At
`the receiver, an optical dcmultiplexer, such as one based on
`a diffraction grating, an arrayed waveguide grating, or a
`thin-film filter array, spatially separates the different wave-
`length components, which are separately detected and outv
`put as respective electrical data signals. For an N wavelength
`WIJM wavelength grid, the fiber capacity is increased by a
`factor of N using electronics of the same speed. Dense
`WDM (DWDM) systems are being designed in which the
`WDM comb includes 40, 80 or more wavelengths with
`wavelength spacings of under 1 nm. Current designs have
`
`10
`
`15
`
`Id Ut.
`
`30
`
`be u:
`
`40
`
`45
`
`50
`
`55
`
`till
`
`65
`
`2
`
`is,
`wavelength spacings of between 0.4 and 0.8 nm, that
`frequency spacings of 50 to 100 (31-12. Spectral packing
`schemes allow for higher or lower spacings, dictated by
`economics, bandwidth, and other factors.
`Point-to-point WDM transmission systems as described
`above enable very high transmission capacity in networks
`having simple connectivity. However, a modem commu ni~
`cations network 10, such as that illustrated in FIG. 1, tends
`to be more complex and requires that the WDM concept be
`expanded to cover not only transport but also switching in a
`complex communication network. This network 10 has
`multiple switching nodes 12 switching signals between
`multiple terminals 14 at the edges of the network 10. Fiber
`optic links 16 interconnect
`the switching nodes 12 and
`terminals 14. The switching nodes 12 should be capable of
`switching single-wavelength WDM channels in difl'erent
`directions with the directions being changeable over some
`time period. The network diagram of FIG. 1 is highly
`conceptual but emphasizes the switching requirement ofthe
`illustrated complexly connected network. Such a WDM
`network 10 achieves high capacity through multi-
`wavclength channel
`transport along complexly fiber-
`interconnected routes. To achieve a flexible dynamic inter-
`connection wilhin the network II],
`it is desirable that
`the
`fiber infrastructure be wired differently for diEerent wave—
`lengths. As a result, the optical switches 12 should not only
`be dynamically reconfigurable between multiple pens but
`also the switch state should be able to be wired
`simultaneously, yet independently in each wavelength chau-
`ncl. That
`is,
`the cross-connects 12 should be wavelength
`selective. Channel paths 20 are illustrated in FIG. 1 in which
`three wavelength channels at wavelengths 11, 1.2, M can
`enter a switching node 12 on a single fiber 16 but be
`switched at its outputs to three difierent fibers 16. Also note
`that
`this architecture allows the reuse of wavelengths
`between difierent pairs of terminals 14, for example. the
`wavelength 7t, as illustrated. Frequency reuse further
`increases the trafiIc capacity of the network 10. Such switch-
`ing requirements apply as well to a more regularly config-
`ured ring network having switching nodes distributed
`around a fiber ring with a terminal associated with each
`node.
`
`The practice in the recent past has been to form each link
`16 in the network 10 as a separate point-to-point WOM
`system so that each switching node 12 includes an optical
`receiver and an optical
`transmitter for each wavelength
`channel. The electrical data signals derived from the optical
`receiver are spatially switched by conventional electronic
`switches and then converted back to optical form for trans-
`mission on the next link. However, a point—to~point design
`does not integrate well into a complex network such as in
`FIG. l. The required number of optical receivers and trans-
`mitters become very expensive. Also, such a system requires
`the electronics and opto—electronics at each switching node
`to be operating at
`the highest data rate supported by the
`network. Such a system is unduly costly when some end-
`to—end links require only modest data rates, and the system
`is difficult to upgrade since all the nodes must be upgraded
`at the same time.
`in
`there is much interest
`For these and other reasons,
`all—optical communication networks in which each switch—
`ing node demultiplexcs the multi~wave length WDM signal
`from an input fiber into its wavelength components, spatially
`switches the separate singlewwavelcngth beams in different
`directions, and multiplexes the switched optical signals for
`retransmission on one or more output
`fibers. Thus,
`a
`wavelength-routing node will generally switch WDM chan-
`
`

`

`3
`
`4
`
`US 6,798,941 B2
`
`nels in an all-optical manner unless there is a specific need
`to electrically regenerate a specific subset of channels, for
`example,
`to remove accumulated optical noise through
`electrical regeneration or to perform wavelength conversion.
`Indeed, the goal is to reduce the number ofconversions from
`optical to electrical and back to optical at each intervening
`node in a fiber optic end—to—end link.
`An all-optical wavelength-selective switching node 12
`may be implemented by a wavelength cross connect (WXC)
`such as a 3x3 WXC 22, represented in the simple schematic
`diagram of FIG. 2, coupling three input ports 24 to three
`output ports 26, each port being typically equated with a
`transmission fiber in the network. The WXC 22 has the
`
`capability ol'switching any wavelength channel on any input
`port 24 to the corresponding wavelength channel on any
`output port 26. The design is not
`limited to a 3x3 cross
`connect with three input and three output ports and may be
`generalized to a WxW cross connect with W input ports and
`W output ports, where W is greater than 1. Systems under
`development have values of W up to [2, and further repli-
`cation is possible. An unequal number of input and output
`ports is possible but will not be further discussed here.
`Although other designs are possible,
`the MC 22 is
`typically accomplished, as illustrated in the schematic dia-
`gram of FIG. 3, by splitting the WDM channels into their
`wavelength components and switching those wavelength
`components by optical elements generally insensitive to the
`wavelength values, often referred to as white-light elements.
`The multivwavelength signal on each optical input port 24
`enters a respective one of 3 (more generally “0 optical
`demultiplexers 28 which separate the multi-wavelength sig-
`nal into its N wavelength channels, here illustrated as N-4,
`and outputs single—wavelength switching beams 30. The
`single-wavelength switching beams 30 of the same nominal
`wavelength enter a respective one of N white-light cross
`connects 32 J
`to SIN, which can switch any input switching
`beam 30 to any of its output switching beams 34 without
`regards to wavelength except that the switching beams 30,
`34 in any plane have the same wavelength or color. W
`optical multiplexers 36 each receive N single-wavelength
`output switching beams of N different wavelengths and
`combines them into a multi-wavelength signal impressed on
`one of the W output ports 26. By such an arrangement, the
`N wavelength channels on each of the W input ports 24 are
`simultaneously and independently routed to selected ones of
`the W output ports 26. The wavelength multiplexers and
`demultiplexers may be accomplished in a number of ways
`typically including dispersive elements such as Bragg
`gratings,
`thin-film interference filter arrays, and arrayed
`waveguide gratings (AWGs} that spatially separate wave-
`length components.
`‘l‘here are a number of ways of achieving the wavelength
`routing functionality represented by the wavelength cross
`connect 22. The illustrated stnictu re of FIG. 3 represents a
`replicative approach using N distinct white—light cross con—
`nects 321, to 32N. A more integrated design, schematically
`illustrated in FIG. 4, uses the same input and output ports.
`multiplexers 36, and demultiplexers 28, but uses fibers 42,
`44 on the switch side of the demultiplexers 28 and multi-
`plexers 36 to route the single-wavelength signals to a large
`white-light optical cross connect 46 having WN inputs and
`WK outputs. The optical cross connect 46 needs to be
`configured so that it connects inputs to outputs of the same
`nominal wavelength. Any cross-wavelength connection
`results in loss of that signal at the output multiplexers 36 and
`perhaps corruption of another signal.
`Solgaard et al.
`in US. Pat. No. 6,097,859 (hereinafter
`referred to as Solgaard) disclose a multi-wavelength cross
`
`10
`
`15
`
`id Ut.
`
`30
`
`be J:
`.
`
`40
`
`45
`
`50
`
`55
`
`till
`
`65
`
`connect switch based on an array of micro electromechani-
`cal system {MEMS) mirrors.
`In this device, multiv
`wavelength WDM signals are received on typically two or
`more input ports. An input lens systems collimates these
`beams and directs them to a diffraction grating that reflects
`different wavelength channels at difi'erent angles. The result-
`ing two—dimensional array of beams, in which input beams
`are separated in one dimension and the wavelength channels
`are separated in the other dimension, is imaged onto an array
`of electronically actuated MEMS micro—minors. Each beam
`is reflected by its micrmrnirror at
`a selected angle that
`depends upon the voltage applied to the mirror actuator.
`Because switching is performed between corresponding
`wavelength channels of different
`fibers,
`in the simplest
`design the mirrors need to tilt only in a single dimension.
`A more integrated design following the Solgaard design
`includes the wavelength multiplexing and MEMS switching
`in a single unit. A 2-input, 2—output, 7—wavelength switching
`system 50 is schematically illustrated in FIG. 5. Two input
`fiber waveguides 52, 54 and two output fiber waveguides 56,
`53 are aligned linearly parallel to each other to couple into
`two free-space input beams 60. 62 and two free-space output
`beams 64. 66. A lens 68 collimates the input beams 60, 62
`to both strike a dilTraclion grating 70.
`Considering the first input beam 60, the dillraction grating
`70 angularly disperses it into a fan-shaped collection 72 of
`beams angularly separated according to wavelength, as is
`well known in the art and taught by Solgaard. That is, the
`grating 70 acts as a wavelength-dispersive element. The
`wavelengths of the signals on the one input fiber 52, as well
`as on all the other libers 54, 56, 58 correspond to the WOM
`wavelengths of one of the standardized grids, for example,
`the ITU grid, and each optical carrier signal of the different
`wavelengths on the separate fibers is modulated according to
`its own data signal. Each of the beams in the collection 72
`of beams corresponds to one of the wavelength channels of
`the ITU grid. A lens 74 focuses these beams toward a first
`row 75 of tiltable input mirrors 76, typically formed as a
`two—dimensional array in the plane of a MEMS structure.
`The mirrors 76 of the first row 75 are associated with the
`
`wavelength channels of the first input fiber 52 while those in
`second row 77 are associated with the second input fiber 54.
`The mirrors 76 are also arranged in a second dimension in
`which each column 78 is associated with one of the wave-
`lengths A, through a, for the illustrated 7-wavelength sys-
`tem. The mirrors 76 described to this point are input mirrors.
`Similarly arranged output mirrors 80 in rows 82, 84 are
`output mirrors. The mirrors 76, 80 are tiltable about respec-
`tive axes lying generally horizontally in the illustration so
`that the input mirrors 76 direct each input beam 60, 62 beam
`toward a folding mirror 86. Depending upon the tilt angle of
`the respective input mirror 76. the folding mirror 84 reflects
`that beam to the output mirror 80 in a selected one of the
`output rows 82, 84. The two illustrated connections show
`coupling to output mirrors 80 located alternatively in the
`third and fourth rows 82, 84. The output mirrors of the third
`row 82 are associated with respective wavelength channels
`on the first output fiber 56 while those of the fourth row 84
`are associated with the wavelength channels on the second
`output fiber 58. The optics are arranged and controlled such
`that an optical signal from an input mirror 76 is reflected
`only to one of the output mirrors 80 in the same column 78,
`that is, associated with the same WDM wavelength. The
`input and output mirrors 76, 80 typically have the same
`construction and difl'er only by their placement in a two—
`dimensional array in a single MEMS structure. Practically
`speaking. in this configuration, the designation of input and
`
`

`

`5
`
`6
`
`US 6,798,941 B2
`
`output mirrors is arbitraxy and the input and output rows
`may be interleaved.
`In some applications, it is possible to dispense with the
`folding mirror 86 and to use only a single set of micromir-
`rors to directly reflect a wavelength-separated input beam
`back to a selected output fiber although this configuration
`presents problems with uniformity of coupling.
`Each output mirror 80 is also tiltable in correspondence to
`the tilt angle of the input mirror '76 to which it is coupled
`through the folding mirror 86 so that the same optics 68, 70,
`74 used to focus and demultiplex the beams from the input
`fibers 52, 54 are also used to multiplex the wavelength-
`separated output beams onto the two output fibers 56, 58.
`That is, the diffraction grating 70 acts as both a demulti-
`plexer on the input and a multiplexer on the output.
`By means of the illustrated optics and MEMS micromir-
`ror array, a wavelength channel on either of the input fibers
`52, 54 can be switched to the same wavelength channel on
`either of the output fibers 56, 58. It is of course understood
`that
`the described structure may be generalized to more
`input and output fibers and to more WDM wavelengths.
`Another system 90. as illustrated in the schematic dia-
`gram of FIG. 6, provides much of the functionality of the
`system 50 of FIG. 5. Input fibers 92 are arranged in a first
`linear array 94 and output fibers 96 are arranged in a second
`linear array 96. The system includes a first two-dimensional
`array 100 of input mirrors 102 and a second two-
`dimensional array 106 of output mirrors 108. In both arrays
`100, 106. the mirrors 102, 108 are arranged in row directions
`according to fiber and in column directions according to
`wavelength. The beams are directly coupled between the
`input and output mirrors 102, 108 without the use of a
`folding mirror. However, such a coupling mirror may be
`advantageously applied between the two mirror arrays 100.
`106 and eliminate the need for separate gratings and further
`allow the input and output fibers 92—98 to be placed in a
`single linear array. Advantageously. separate demultiplexing
`and multiplexing gratings 110, 112 are provided on the input
`and output sides respectively, and birefringent wave plates
`are inserted so as to substantially eliminate polarization
`dependence within the switch, as is well understood by those
`lt'l art.
`
`A large white-light cross connect may have a structure
`similar to that illustrated in FIG. 6 but without the dilfraction
`
`grating. Awhile-light system 120 illustrated in the schematic
`illustration of FIG. 7 includes a substantial number ofinput
`fibers 122 bundled together in a two-dimensional array 124
`and preferably a like number of output fibers 126 bundled
`together in another two—dimensional array 128. One of the
`input fibers I22 concentrates its beam at one of the input
`minors 102 of the input mirror array 100. Similarly,
`the
`output mirror array 106 has its output mirrors 108 near the
`focus of the output fibers 126. An optional mirror 130
`couples the input and output mirrors 102, 104, Each input
`mirror 102 is tiltable about two axes to allow it to direct its
`
`input beams to any ones of the output mirrors 106. Each
`output mirror 106 is similarly tilted in a complementary
`fashion to direct
`the beam towards the output fiber 126
`associated with that output mirror 103.
`Another unillustrated white—light system resembles the
`folded white»light system 120 of FIG. 7 but with the input
`and output
`fibers arranged in a same one- or
`two-
`dimensional array and with the mirror arrays 100, 106
`integrated into a single one- or two-dimensional array.
`A complex WDM or white-light network is subject to
`many problems. The dilferent optical signals which are
`
`propagating on a particular link or being optically processed
`may have originated from dilferent sources across the net—
`work. In a WDM system,
`the WDM wavelength output
`power may vary from transmitter to transmitter because of
`environmental changes, aging, or dilferenees in power
`injected into the WDM stream. Different optical sources for
`either a WDM or white-light system are additionally subject
`to different amounts of attenuation over the extended net-
`work. Particularly,
`for
`a wavelength-routed transparent
`network,
`the WDM spectrum on a given fiber contains
`wavelength components which generally have traversed
`many diverse paths from difierent sources and with dill'erent
`losses and different impairment accumulation such as deg-
`radation of the optical signal-to-noise ratio or dispersion
`broadening. Further, wavelength multiplexing and demulti-
`plexing usually rely on optical etfects, such as diffraction or
`waveguide interference, which are very sensitive to absolute
`wavelength, which cannot be precisely controlled.
`EDFAs or other optical amplifiers may be used to amplify
`optical signals to compensate loss, but they amplify the
`entire WDM signal and their gain spectrum is typically not
`flat. Therefore, measures are needed to maintain the power
`levels of different signals to be the same or at
`least
`in
`predetermined ratios.
`In a complex WDM or white-light network, a signal may
`be switched multiple times. Each switching event needs to
`maximize transmission of the optical signal and minimize
`cross-talk between channels. A maximum of 10 dB attenu-
`ation through the switch and a minimum of 30 dB channel
`isolation are typical requirements. However, MEMS cross-
`connects and their associated optics are subject to internal
`variations of optical characteristics and misalign ments, holh
`integral to the device and as a result of both manufacturing
`and environmental variation and non-uniformity and of
`mechanical stress, all ofwhich result in switch states having
`a significant variation and instability of insertion loss when
`aligned according to their nominal settings.
`As described above, it is well known that WI)M systems
`must maintain a significant degree of uniformity of power
`levels across the WUM spectrum, so that dynamic range
`considerations at receivers and amplifier, non-linear ellects,
`and cross talk impairments can be minimized. As a result,
`serious attention must be payed to equalization of power
`levels across the spectrum. This equalization should be
`dynamic and under feedback control since the various
`wavelength components vary in intensity with time and due
`to changes in optical channel routing history among the
`components. One object of this invention is to provide
`means of equalization at multiple-fiber WDM switching
`nodes where many beams from diverse sources are inter-
`changed among the fibers.
`Bishop et al. have disclosed in US. Pat. No. 6,263,123 a
`pixellated WDM cross connect using a two-dimensional
`array of micromirrors in which a signal beam of a particular
`carrier wavelength is distributed to a plurality of the micro—
`mirrors. The number of mirrors reflecting light to an output
`port determines the transmission coeflieient
`through the
`switch. When such a pixellated intensity control is used in a
`“TDM cross-connect,
`the mirrors are arranged in a two-
`dimensional array and are tiltable about
`two axes. The
`described system is used primarily for characterizing the
`optical signal, not for controlling it. Similarly, Deriekson et
`al. have disclosed in US. Pat. No. 4,796,479 a system for
`monitoring the intensities ot'the WDM channels in a WDM
`cross connect with the main emphasis on determining the
`ratio of signal to noise. It would be desirable to integrate the
`capability of such systems into an optical network without
`unduly increasing the cost and complexity.
`
`10
`
`15
`
`14 UL
`
`30
`
`be u:
`
`40
`
`45
`
`50
`
`55
`
`(it)
`
`65
`
`

`

`7
`SUMMARY OF THE INVENTION
`
`US 6,798,941 B2
`
`8
`
`An optical switching system includes a plu ralily of optical
`switching elements controllable in two difl'erent scales or
`dimensions depending upon the switch architecture and
`designed to be able to effect both switching and control of
`the transmission coellicienl. The poorer adjustment is par-
`ticularly useful in a wavelength-division multiplexing opti-
`cal switch in which the power is adjusted between different
`wavelength channels combined in an output path such as an
`optical fiber.
`An example of such a switching element is a mirror
`tiltable about two orthogonal directions or tiltable in one
`dimension according to a fine and a coarse resolution. In one
`embodiment, tilting about a major axis is used to control
`switching between optical ports, and tilting about the minor
`axis is used to cortlrol the amount of optical power passing
`through the switch.
`Such mirrors are advantageously formed in a micro
`electromechanical system (MEMS) array. A mirror may be
`tilted by two pairs of electrostatic actuator electrodes posiw
`tioned beneath the mirror on opposed sides of Ivvo torsion
`beams supporting the mirror.
`A portion of the switched output signal may be diverted
`to an optical power monitor to enable feedback control of the
`power adjustment. Dynamic power equalization advanta-
`geously involves monitoring of power levels for the indi-
`vidual WDM wavelength channels on each fiber in the
`system. This information provides feedback [or the power
`equalization mechanism.
`Input power may be advanta-
`geously also monitored for each wavelength channel.
`The minor axis tilt may also be used for optimizing
`transmission through the optical switch. After the position of
`maximum transmission is established, the transmission may
`be deluued. An example of two—axis switching is tltc use of
`a major axis optimized for switching between fibers and a
`minor axis for adjusting transmission, especially insertion
`loss, of the same optical channel. Another example is the use
`of a coarse control of one or more major axes for establish—
`ing a switch state connection in combination with a line
`control along one or more minor axes (which may or may
`not be the same as the major axes) to moderate the degree
`of coupling of a wavelength channel between the chosen
`fibers for that wavelength service.
`to white-light
`The feedback control applies as well
`switching systems in which the same mirrors are used for
`switching and for transmission optimization and power
`equalization.
`The minor axis tilt may be used to increase the high
`insertion loss for any optical connection in the off state and
`may further be used to turn the switch to a hard off during
`switching between discrete optical paths.
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`1 is a network diagram of a complexiy connected
`FIG.
`optical communications network.
`FIG. 2 is a general block diagram of a 3x3 wavelength
`division multiplexing optical cross connect (WOXC).
`FIG. 3 is a more detailed block diagram of the WOXC of
`FIG. 2
`
`FIG. 4 is a block diagram ofa more integrated version of
`the WOXC of FIG. 2.
`
`FIG. 5 is a schematic diagram of a WOXC using one array
`of rniorornirrors and a folding mirror.
`FIG. 6 is a schematic diagram of a WOXC using two
`arrays of micromirrors and no folding mirror.
`
`10
`
`15
`
`Id Ln
`
`30
`
`be u:
`
`40
`
`45
`
`50
`
`55
`
`(ii)
`
`65
`
`FIG. 7 is a schematic diagram of a white-iight optical
`cross connect
`
`FIG. 8 is a block diagram of an optical switching system
`including an optical power monitor and feedback control.
`FIG. 9 is a schematic plan view of a 2x2 wavelength
`division multiplexing (WD M) cross connect using an exter-
`nal optical power monitor.
`FIG. 10 is a sectioned orthographic view of a concentrator
`included in the cross connect of FIG. 9.
`
`FIG. 11 is a schematic plan view of a 2x2 WDM cross
`connect using an internal external optical power monitor.
`FIG. 12 is a block diagram block diagram of the optical
`switching system of FIG. 8 additionally including an input
`optical power monitor.
`FIG. l3 is a schematic plan view of an improved of the
`WDM cross connect of FIG. 9 additionally including moni-
`toring the input power.
`FIG. 14 is plan view of two-axis tiltable mirror usable
`with the invention.
`
`FIG. 15 is a cross sectional View taken along View line
`15—15 of FIG. '12.
`
`FIG. 16 is a schematic orthographic view ol‘a mirror array
`and its fold mirror.
`
`FIGS. 17 and 18 are cross-sectional views illustrating two
`kinds of mismatch in optically coupling two beams.
`FIG. 19 is an elevational view of a concentrator 01" FIG.
`10 illustrating a method of switching between two output
`beams.
`
`DETAILED DESCRIPTION OF THE.
`PREFERRED EMBODIMENTS
`
`The invention includes an optical switching system in
`which the output power on particular channels is monitored
`to enable power adjustments in the optical switch,
`for
`example. adjustment of the transmission coefficient for an
`optical channel transmitted through the switch.
`A block diagram of one embodiment of a controlled
`transmissivity optical switching system 140 ofthe invention
`is illustrated in FIG. 8. A 2x2 optical cross connect (OXC)
`142 links two input Iibers 144, 146 connected to input ports
`IN}, {NZ to two output fibers 148, 150 connected to output
`ports OUT], OUT} However,
`the invention is directly
`applicable to a iarger number of input and output fibers.
`In the wavelength division multiplexing (WDM) embodi-
`ments of the invention. each fiber 144, 146, 148, 150 is
`capable of carrying a multi-wavelength WDM optical sig-
`nals at wavelengths RYAN. The OXC 142 is capable of
`switching the separate wavelength components on each
`input fiber 144, 146 to either of the output fibers 148, 150.
`This. architecture applies as well to a WDM addfdrop mul-
`tiplexer (WADM) in which the inpu