UDC 621.383:62l.391.6
`
`Photonic Networking Using Optical
`Add Drop Multiplexers and Optical
`Cross-Connects
`
`CTerumi Chikama OHiroshi On aka OSatoshi Kuroyanagi
`(Manuscrzpt received May 8, 1999)
`
`The photonic network will enable the construction of high-capacity and flexible
`optical communication systems for the future data-centric era. Optical add drop mul-
`tiplexers (OADMS) and optical cross connects (OXCs) along with already mature DWDM
`systems are key technologies for photonic networking. Prototype systems of 0ADMs
`based on the acousto-optic tunable filter (AOTF) and 0XCs based on PLC optical switch-
`es have been demonstrated.
`This paper provides a perspective of the latest optical path layer technologies.
`
`1.
`
`Introduction
`
`In the 21st century, there will be an explo—
`
`sive growth in the amount of information being ‘
`transmitted by digital services such as electronic
`commerce, software distribution, and digital
`video/music distribution services. The capacity
`required to handle all this information will be pro
`vided using new communication technologies.
`IP/ATM and photonic networking are key—enablers
`for realizing terabit capacities and efiective and
`reliable use of networks. Current transport tech-
`nologies based on the SONET/SDI-I format are
`already in wide use in today’s networks. The
`transport network has a layered structure as de~
`fined by the International Telecommunication
`Union—Telecommunication Standardization Sector
`
`(ITU-T). It consists of a circuit, transmission me-
`dia, and path layer. The maturity of OC48, O0 192,
`and extremely dense WDM forces us to rethink
`the strategy for cost—efiicient bandwidth manage—
`ment using these layers. The introduction of an
`optical path layer with high bit rate TDM pipes
`multiplexed by DWDM which can be managed by
`QADM and OXC will be effective for overall net-
`
`work efficiency.“_ In the ring architecture, an
`
`OADM can be introduced to make efficient use of
`
`network capacity, network protection, wavelength
`routing, and many more features.
`In the mesh
`architecture, an OXC could provide the scalabili—
`ty, modularity, and transparency required in the
`network.
`In addition, photonic networks based
`on OADMS and OXCS will provide openness and
`transparency in future networks to accommodate
`various client signals with different bit rates and
`formats (e.g., SONET, SDH, ATM, and IP) efiicient~
`ly and to forward the client signals transparently
`to end users.
`
`This paper provides a perspective of the lat-
`est optical path layer technologies. Some key"
`advancements in the OADM architectures using
`
`acousto-optic tunable filters (AOTF) will be
`described along with the concept of optical path
`protection. Also, optical path cross connect
`architectures will be discussed along with the key
`
`features required for practical use of this technology.
`
`2. Optical Add Drop Multiplexer (OADM)
`OADM technology is used to cost effectively
`
`access part ofthe bandwidth in the optical domain
`being passed through the in~linc amplifiers with
`
`46
`
`FUJITSU Sci. Tech. J.,35,1,pp.46~55(Ju|y 1999)
`
`Cisco Systems, Inc.
`Exhibit 1047, Page 1
`
`

`
`T Cltilaarwza at al.: Photonic Networleing Using Optical Add Drop Multiplexers and Optical Cr0ssV~Connects
`
`the minimum amount of electronics. OADM can
`
`be used in the static as well as dynamic mode.
`Table 1 shows the migration scenarios of OADM.
`In passive OADM, the add and drop wavelengths
`are fixed beforehand.
`In dynamic mode, the
`OADM can be set to any wavelength. after instal-
`lation. Passive OADM is currently being used in
`networks with WDM systems. The technologies
`used to accomplish passive OADM are thin~f'1lm
`interference filters, fiber gratings, and planar
`waveguides. The optical characteristics such as
`the insertion loss and the inter-band and intra-
`band crosstalk are Well understood for each of
`
`these technologies when used in a passive OADM‘
`application. Dynamic OADM has the advantages
`of better cost~effectiveness and flexibility than
`passive OADM because it can select any wave-
`length by provisioning on demand without
`changing its physical configuration. A smooth
`
`migration from passive to totally reconfigurable
`and dynamic OADM will be necessary. Dynamic
`OADM is classified into two generations. The sec~
`0nd generation is mainly applied in a linear
`configuration without an optical path protection
`function.
`The path protection function is
`supported by electrical ADMS. Finally, the third
`generation will be applied in a ring configuration
`
`Table 1
`Migration of OADM.
`
`1
`
`Passive
`OADM
`
`Dynamic reconfigurable
`OADM
`
`Fixed
`
`Setlable by provisioning
`
`Manual
`change of
`fiber
`connections
`
`Automatic change of
`connections by optical SW
`with provisioning
`
`SONET/SDH APS
`
`Optical layer
`APS
`Linear, ring
`OBLSR/OBPSR
`Linear
`Optical SW, AOTF
`Tunable wavelength LD
`
`Linear
`Fiber grating
`Dielectric filler
`
`FUJITSU Sci. Tech. J.,35, ‘lr,(Ju|y 1999)
`
`to provide optical layer path protection based on
`‘the 4—fiber Bi-directional Line Switched Ring
`(BLSR) and other protection schemes.”
`Regarding the architecture of dynamic
`OADM configurations, there are two types. One
`is the SW type with a back—to~back inultiplexerl
`demultiplcxer, and the other is the AOTF type.
`Figure 1 shows these configurations. One of the
`technical difficulties in using the SW type for
`OADM is that for n channels in a WDM system,
`
`an n X n optical switch will be required on the
`drop and add sides to accomplish a dynamic ca-
`pability. This can be extremely expensive and
`cumbersome. Other problems such as channel
`passband narrowing due to concatenation of mul-
`tiplexers/dcmultiplexers for a channel spacing of
`0.8 nm or less can also create maj or problems in a
`long—distance network.” The AOTF type holds a
`lot of promise for providing a cost—effective solu-
`tion for a static as well as dynamic OADM and
`presents no passband narrowing problem. We
`have therefore been developing dynamic OADM
`systems using AOTF.
`
`2.1 Acousto optic tunable filter (AOTF)
`
`configuration
`Figure 2 shows the device configuration of
`an AOTF developed by Fujitsu.” The device is
`fabricated on lithium niobate (LiNb03) and is com-
`
`.
`
`posed of an inter—digital transducer (IDT), optical
`waveguide, thin-film surface acoustic wave (SAW)
`
`AOTF Type
`
`TL :Tunable aser
`
`SW'and AOTF types of OADM architectures.
`
`Cisco Systems, Inc.
`Exhibit 1047, Page 2
`
`

`
`T. Chileama et al.: Photonic Networking Using Optical Add Drop Mzr.lf.iplexers and Optical Cross-Connects
`
`guide, and polarization beam splitters (PBSS). The
`incident light is propagated over the optical
`waveguide and divided into perpendicular com-
`
`ponents (TE/TM) by the first PBS. An acoustic
`wave is generated by applying an RF signal to the
`IDT. This acoustic wave travels through the SAW
`guide and causes a periodic modulation of the re-
`fractive index of the optical waveguide. This
`change of refractive index induces TE-TM or
`
`Tllled thin-film SAW guide
`
`)2, 2A, 15’
`Optical signal
`output
`Drop slgnal
`M, 13
`
`Control signal (RF)
`(170480 MHz)
`SAW: Surface acouslic wave
`
`‘
`_
`‘
`(lnler~dIgllal transducer)
`
`'
`
`Figure 2
`Configuration of AOTF.
`
`TM-TE conversion for only the drop wavelength.
`The drop wavelength corresponds to the applied
`RF frequency and becomes perpendicular to the
`incident light. The second PBS is then used to
`separate the drop wavelength from the incident
`light. An AOTF can not only drop a single
`wavelength but also multiple wavelengths sirnul~
`taneously. By changing the number of RF‘ signals
`and their frequencies, we can control the number
`
`and frequencies of the drop Wavelengths. There
`are no moving parts in the AOTF, and it offers
`high-speed wavelength tuning that can be done
`sequentially or randomly based on the applied RF
`frequency. Although the insertion loss of AOTFS
`
`has been relatively high and the sidelobe suppres-
`sion has been poor, recent advancements have
`significantly improved both of these characteris-
`tics and allow optimal network efficiencies in the
`photonic layer unit (SAU).
`Figure 3 shows WDM signal separation us-
`ing an AOTF. By using two AOTFS, we were able
`
`lcaloutput(5dB/div.)Oplioaloutput(5dB/div.)
`
`
`
`
`Opticaloutput(5dB/div.)
`
`-
`Opt
`
`1561
`1545
`1529
`Wavelength (nm)
`
`1 561
`1545
`1 529
`Wavelength (nm)
`
`1561
`1545
`1529
`Wavelength (nm)
`
`FUJITSU Sci. Tech. J.,35, l,(J‘uly1999)
`
`1529
`
`1561
`1 545
`Wavelength (nm)
`
`17:ZM.5
`Frequency (MHZ)
`
`175
`
`4.
`
`2
`'2[D'13
`El.4
`3CL..
`D0
`To
`.9..
`G.
`O
`
`RFoutput(5dB/div.)
`
`(170 ~18O MHZ)
`
`F lgure 3
`Wavelength selection by AOTF.
`
`Cisco Systems, Inc.
`Exhibit 1047, Page 3
`
`

`
`T. Chikmna er ail; Photonic Networking Using Optical Add Drop Multiplzexers and Optical Cross—Conneczs
`
`to extract any wavelength among 32 wavelengths
`separated with a 0.8 nm channel spacing. The
`first AOTF dropped five wavelengths simulta-
`neously and passed the remaining 27 to the
`through port. Then, the second AOTF extracted
`the desired wavelength from the five dropped ones.
`This method provides sufficient adjacent channel
`crosstalk suppression.
`
`2.2 OADM design considerations
`To maximize the effectiveness of an OADM,
`the functions listed in Table 2 have to be consid-
`
`ered. Our system was designed according to the
`following five main considerations:
`1)
`In the case of an OADM node with AOTFS,
`an AOTF can be used instead of an n X n
`
`switch to retrieve individual or multiple
`channels. Also, transponders (0/E, E/O,
`optical modulators) with tunable lasers can
`
`be used on the add side to provide the dy—
`namic capability. In this case, the OADM can
`support random selection ofwavelengths.
`Our system can support a drop-and~c0ntin-
`
`ue or broadcast feature by using a signal tap
`component within the OADM node. This
`
`.
`Table 2
`Functions of OADM.
`
`Wavelength
`MUXIDMUX
`
`-Maximum number of wavelengths:
`16, 32, 64, 128...
`A
`-Number of add/drop wavelengths
`
`Wavelength
`cross-connect
`(A SA: K slot
`assignmem)
`
`lntcr~office IF
`
`»Add wavelength: Fixed/Seilable
`-Drop wavelength: Fixed/Settable
`-Through wavelength: Fixecllsellable
`(it 8|: ;\. Slot interchange)
`-Broadcast (Drop and Continue)
`-Transmission fiber: SMF/DSF/NZ-DSF,
`span, number of spans
`-Inter-«working for maintaining survivability:
`SONETISDH APS, optical SNR
`
`_
`lntra~office IF
`
`400192/c, OC48/c. Asynchronous signal,
`G—Elhernet, 100 BaseF
`-lntenworking for maintaining survivability
`
`Management
`and control
`
`-Wavelength control, Performance monitoring,
`Output level control, wavelength path trace
`-Transferring supervisory channel:
`OSC (optical supervisory channel)
`
`FUJITSU Sci. Tech. J.,35, ’l,(July 1999)
`
`function will be useful for the configuration
`ofdual~hubbing in a multiple ring connection.
`Our system can support a function of optical
`layer protection such as the 4-fiber Bi-
`directional Line Switched Ring (BLSR).
`Figure 4 shows the OADM node configura—
`tion of an optical self-healing ring. This
`configuration consists of four fibers for two
`bidirectional lines, dual nodes (work and
`protection node) for each direction, optical
`span SWs, and optical ring SWS. Protection
`line and optical span SWs are used during
`OADM equipment failures and fiber breaks
`on the work side. Optical ring SWs are pre-
`pared to loop back during fiber breaks on both
`the work line and the protection line. In ad-
`dition, this system can monitor parameters ”
`of optical signals such as the optical power,
`wavelength, number ofwavelengths, and op-
`
`tical SNR using a built-in optical spectrum
`analyzer unit (SAU).5’
`‘
`It is clear that by using AOTF-based OADM,
`multi—channel access can be easily and cost
`effectively accomplished. Multi-channel
`
`drops can be fed into a simple splitter/cou-
`pler, after which multi-channel tributary
`interfaces can be used to feed the signals into
`a subtending ring.
`
`_
`Qnlical
`“W9. SW
`
`Optical
`1+1 span SW
`
`g
`
`H
`
`Optical
`1+1 span SW Optical
`ring SW
`
`Proleclion
`
`’ OADM";Work 3
`OADM
`: rolectio
`
`Figure 4
`OADM node coniiguralion for optical self-healing ring.
`
`Cisco Systems, Inc.
`Exhibit 1047, Page 4
`
`

`
`T. Chikanzrz et £11.: P/Lotonic Networking Using Optical Add Drop ultiplexers and Optical Cross-Connects
`
`5) Our system can support a Bellcore standard
`1510 nm OSC channel for retrieval of alarms
`
`and other supervisory information, These
`OADM nodes can support express pass-
`through of the OSC channel.
`
`2.3 Prototype OADM system
`Figure 5 shows the configuration of a proto-
`type OADM system using an AOTF, and Table 3
`lists its specifications. The OADM system con-
`
`sists of an OADM shelf, Tributary shelf, and
`Wavelength bank shelf. This system can accom»
`modate 32 wavelengths at 10 Gb/s with a 0.8 nm
`channel spacing (line capacity: 320 Gb/s) and add/
`drop any four wavelengths.‘ The AOTF in the
`OADM shelf divides input WDM signals into drop
`and through signals. The drop signals are passed
`through the AOTFS in the Tributary shelf to ex-
`
`l DTl}’52:’:]
`
`OADM shelf
`
`Wavelength
`bank shelf
`SAU
`
`Splifier
`
`Cgmbiner
`AOTF AOTF
`
`Mod.|—"‘ fivlocr.
`
`___L..
`Add signal
`
`Drop signal
`
`Figure 5
`OADM configuration using AOTFS.
`
`Table 3
`OADM system speciflcalions.
`
`.
`
`Architecture
`
`Line capacity
`and wavelength
`Acldldrop capacity
`wavelength
`
`Channel spacing
`Optical path rates
`
`AOTF-based free
`wavelength add/drop
`320 Gbls
`10 G X 32 w
`
`40 Gbls
`10 G><4w'
`100 GHZ
`10 Gb/S. 2.5 Gbls
`
`Protection
`
`——-—
`
`Wavelength
`conversion
`
`Compatible with
`FLASHWAVEISZOG
`
`8 w (maxi)
`32 w (in future)
`lTU~T grid
`Transparent
`
`Upgrade lo optical
`B LSR
`
`tract the desired wavelengths, which are then re-
`ceived by each electrical node. In the add process,
`the wavelength bank is used instead ‘of tunable
`LDs. In the Wavelength bank shelf, LDs having
`the same wavelengths as the wavelengths used
`in the line are prepared in advance. These wave-
`lengths are combined and provided to each optical
`external modulator. The modulators are driven
`
`by the add signal received from the O/E. The de—
`sired Wavelength is selected after modulation by
`an AOTF and launched out of the line as the add
`
`signal. As a result, the wavelength of the add sig-
`nal is converted to the desired Wavelength in the
`OADM.
`
`The prototype QADM system is shown in
`Figure 6. This system was demonstrated at
`Supercomin’98 in Atlanta as the I world’s first
`totally reconfigurable, dynamic OADM.
`
`3. Optical Cross-connect System
`To realize ‘efficiency and transparency in the
`optical network, wavelength grooming and rout-
`
`ing functions for each client signal andoptical path
`supervising functions such as performance moni-
`toring and path tracking must be provided. The
`key element for providing these functions is the
`optical cross-connect (OXC) system.”
`
`,
`Figure 6
`Prototype OADM system at Supercomm’98 in Atlanta.
`
`FUJITSU Sci. Tech. J.,35, l,(Ju|y1B99)
`
`Cisco Systems, Inc.
`Exhibit 1047, Page 5
`
`

`
`T Clzikmna. et :11; Phntonic Networking Using Optical Add Drop Multiplexers and Optical Cr'oss~Connects
`
`3.1 Key technologies
`Some technical issues have to be considered
`
`inrconnection with the development of the OX0.
`We investigated these issues from the viewpoint
`of the optical switch architecture and the super-
`vision of the optical path (especially, path
`tracking).
`v
`
`3.1.1 Optical switch architecture
`The optical switch is a key element for real-
`izing an OXC node.
`If the insertion loss of the
`optical switch is large, optical amplifiers (OAS)
`must compensate for the loss inside the node. If
`the loss Variation at each switch port is too large,
`the optical receivers (ORS) at the termination and
`regenerating nodes cannot receive the optical sig-
`nals because the dynamic range of the ORS is
`exceeded. The extinction ratio of the optical switch
`
`depends on switch-specific characteristics (the
`device that is used and the configuration of the
`
`S13
`
`S23
`
`S21
`
`S41
`
`(b) Proposed structure (Pl-LOSS switch)
`
`® :Cross © :Bar
`Si]: Control point for connecting input #i to output 125]
`Pi-LOSS : Path-independent insertion loss
`
`Figure 7
`Optical switch architecture (4 x 4).
`
`FUJITSU Sci. Tech. J.,35, 1,(July 1999)
`
`switch element). On the other hand, the crosstalk
`
`in the optical Inatrixswitch depends mainly on
`the architecture of the switch. If the crosstalk of
`
`the optical switch is large, optical signals will be
`disturbed by other optical signals. For these rea-
`sons, the optical switch requires an architecture
`which can suppress the degradation of transmis-
`sion quality (insertion loss, loss variation, and
`crosstalk).
`In the case of a conventional crossbar
`
`switch,” the number of switch elements in each
`optical path is different. In the 4 X 4 switch, the
`best path passes through only one switch element
`(from input #4 to output #4), While the worst path
`passes through seven switch elements (from in-
`put #1 to output #1) (Figure 7(51)). As a result,
`the insertion loss, its variation, and crosstalk be-
`come larger as the switch capacity increases.
`Therefore, we have proposed a PLLOSS (path—in-
`dependent insertion loss) optical switch as a
`solution for these problems (Figure ’7(b))."”
`In
`the PLLOSS switch architecture, all optical sig~
`nals pass through the same number of switch
`elements, which means that the insertion loss is
`constant and is about half of the maximum loss of
`the conventional crossbar switch.
`
`We developed the 8 x 8 optical switch using
`the PI—LOSS topology for the OXC node. The 8 X 8
`PI-LOSS switches are silica-based, thermo—optic
`matrix switches. Each switch element for the Pl-
`LOSS switch is
`constructed from double
`
`Mach-Zehnder Interferometers (MZIS). The in-
`sertion loss and insertion loss variation of the
`
`PLLOSS switch were less than 6 dB and 1 dB,
`respectively. The loss variation of 1 dB includes
`not only the variation between path routes, but
`also the Wavelength and polarization dependency
`loss. The total crosstalk from other channels was
`
`_
`less than -37 dB.
`3.1.2 Optical path tracking
`An optical path tracking method which is in~
`dependent of the client signals is also required
`for maintaining the transparency of the optical
`network. Optical path tracking using a pilot tone
`
`51
`
`Cisco Systems, Inc.
`Exhibit 1047, Page 6
`
`

`
`T. Chikanm. el al.: Phmfonic Networking Using Optical Add Drop Multiplexers and Optical Cross~Connects
`
`(PT) is an effective tracking method and is cur-
`rently being discussed in the lTU-’I‘.~‘”"“’ In this
`method, the PT signals modulate a sub-carrier
`that is superimposed onto an optical signal that
`acts as a main carrier. The PT signals that mod-
`ulate this sub-carrier are monitored at each node,
`which enables the identification (ID) of the opti-
`
`cal paths to be established, which in turn makes
`it possible to check the optical path connection.
`A block diagram for a PT circuit we manu-
`factured for trials is shown in Figure 8. The PT
`signal is a 4 kb/s base-band signal with 142 bits
`per frame. This frame contains the ID of an opti—
`cal path, the ID of the OXC, and the status
`
`information. The PT signal is modulated using
`quadrature phase shift keying (QPSK), since
`
`QPSK has good transmission characteristics and
`
`Optical signal 10
`(with PT)
`
`Optical
`modulator
`
`Bias
`current ib
`
`Sub-carrier
`oscillator
`
`Optical signal
`(with PT)
`
`l
`
`PT modulation
`current ipt
`QPSK modulator
`
`‘
`
`M PT;;eherator
`
`(a)_ PT sender (with it CNV)
`
`Selector
`
`AGC
`Amplifier
`
`| P
`
`PD
`.
`-
`. PD
`
`l:
`
`QPSK demodulator
`V
`
`PT detector
`
`(b) Pt receiver
`
`a good performance with respect to frequency
`availability. The sub—carrier frequency used in this
`
`I
`experiment was 150 kHz.
`The PT sender was mounted on a board in-
`
`cluding an OR module and an optical sender (OS)
`module with a laser diode (LD) and an optical ex-
`ternal modulator. The LD was directly driven by
`the QPSK modulated PT signal current superim-
`posed on a bias current. The output power of the
`LD was modulated in the optical modulator by the
`main signal.‘ The sender can adjust the sub—carri~
`er amplitude. We varied the sub-carrier amplitude
`modulation index, m, which is the ratio of the main
`
`optical signal to the sub—ca.rrier amplitude.
`The PT receiver was mounted on a board
`
`which had four receiving ports. The phase lock
`
`loop (PLL) technique was used in the receiver,
`which was useful for keeping the bandwidth of
`the electric receiver very narrow. Consequently,
`
`the receiver delivered a good performance with
`respect to the signal-to-noise ratio (SNR).
`
`3.2 Prototype system
`Figure 9 shows the configuration of the
`prototype OXC node, and Table 4 lists its specifi-
`
`cations. The ‘prototype system mainly consists of
`optical switches (OSWS), Wavelength multiplex-
`ers (WMUXS) and deinultiplexers (WDMXs), and
`optical pieamplifiers (Pre—OAs) and post-ampli-
`
`lntenoffice
`
`(Shelf—1)
`
`_
`(shegpz)
`
`inter-office
`(She|f—1) PONS
`
`(Sh
`
`Q03 monitor
`9 X
`t
`|ntra~office ;
`Av
`Doris
`PT senc er
`Shelf-3
`
`.
`;
`
`(She|f—4)
`PT receiver
`—}‘.l4>lntra—office
`;
`pods
`__.___. X8
`
`NE controller
`
`Figure 8
`Block diagram of PT sender and receiver.
`
`Figure 9
`Configuration, of prototype OXC system.
`
`52
`
`FUJITSU Sci. Tech.‘J.,35, 1,(July 1999)
`
`Cisco Systems, Inc.
`Exhibit 1047, Page 7
`
`

`
`T. Clzikamcz at al: Photonic Networking Using Optical Add Drop Multiplexers and Optical Cross—Conn,ects
`
`fiers (Post-OAS). In this system, we used our pro-
`posed PLLOSS switches for the OSWS. The PT
`senders were connected to the input
`intra-
`
`office ports, and the PT receivers were arranged
`at the output of the OSWS. The optical signals
`from each input port were routed by the OSWS to
`the appropriate output ports according to the con-
`trol signal from the network element (NE)
`controller. We also monitored quality of service
`(Q08) factors such as the optical power, wave-
`lengths, and optical SNR for each optical path at
`the input inter-office ports by using an optical
`spectrum analyzer. The performance information
`was sent to the NE controller.
`A
`
`Eight wavelengths can be multiplexed in a
`single fiber, which leads to a wavelength spacing
`‘of 1.6 nm. The OXC node accommodates three
`
`input ports and three output ports plus one intra- ‘
`office port. This means that the OXC node
`switches 24 optical paths (three ports x eight
`wavelengths). The bit transmission rates per op-
`tical path are 10 and 2.5 Gb/s. The WMUXS and
`WDMXS are array waveguide grating (AWG) mul-
`
`.
`V
`tiplexers and demultiplexers.
`The prototype OXC node was constructed
`using various kinds of boards mounted in a single
`
`Table 4
`OXC system specifications.
`
`Optical path
`Multiplexed 3.5
`lnput/Output ports
`-
`
`Wavelength path (WP)
`8 (@115 nm, lTU-T grid)
`3 (lntenand intra~office ports: 2 and 1)
`@Upgradeable'to eight ports
`
`‘ Signal speed
`Operation and
`supervision
`
`Optical devices
`
`10 G (OC492) and 2.5 G (0048)
`~ NE based control and management
`- Path trace using pilol lone signal
`- Quality monitor using OSA
`- 8 x 8 Pl~LOSS optical switch
`(Silioabased TOSW)
`- WDMX and WMUX: AWG
`
`Equipment size
`
`1 bay (6.5-foot rack)
`
`AWG: Arrayed-waveguide grating
`OSA: Optical spectrum analyzer
`
`FUJITSU Sci. Tech. J.,35,l,(July1999)
`
`cabinet. The cabinet contained an inter~ofiice port
`interface shelf (shelf-1), an OSW shelf (shelf-2),
`an intra—office port interface shelf (shelf-8), and
`an optical path supervision shelf (shelf—4). By us-
`ing eight OSW boards and another cabinet having
`two inter-office port interface shelves and one in-
`tra-office port
`interface shelf,
`the OXC can
`accommodate 64 optical paths.
`
`3.3 Evaluation results
`
`We measured the bit error rate (BER) of the
`
`system using the evaluation setup shown in
`Figure 10. For the performance test, four 10 Gb/
`s and four 2.5 Gb/s optical signals were input to
`
`the intra-oflice ports of OXC-X, relayed to OXC—Y,
`and then received at the output intra-office ports
`of OXC-Z. Figure 11(a) shows the BER Versus
`the received power for the case of the 10 Gb/s op-
`tical signals. The tests showed a power penalty of
`less than 1.2 dB at a BER of 10'“. The BER pen-
`alty is caused by the amplified spontaneous
`emission (ASE) noise of the OAS and coherent
`crosstalk in the OSWS.
`
`The quality of the PT signal and main opti-
`cal signal are in a trade-off relationship. The
`
`ERD‘
`2.5G 10G
`
`PPG : Pulse pattern generator
`ERD : Error detector
`O
`1 Measure point pilot tone
`
`b
`
`E/O
`
`2.5 G
`
`Figure 10
`Evaluation setup for OXC.
`
`Cisco Systems, Inc.
`Exhibit 1047, Page 8
`
`

`
`T. Chile-rtma. at 411.: Photonic Networking Using Optical Add Drop Multiplexers and Optical CrossAConnects
`
`Operation system
`
`10'9
`
`10-10
`
`.
`
`1041 *
`
`-16
`-18
`-22 -20
`Received power (dBm)
`(a) 10 G signals
`
`Received power (dBm)
`.(b) 2.5 G signals with PT
`
`Figure 11
`BER characteristics.
`
`larger the modulation index, m, the more the BER
`of the main signal is degraded, but the more the
`BER of the PT signal is improved. We therefore
`examined the relationship between in and the
`BER of the main optical signal. P’I‘ signals were
`superimposed on the 2.5 Gb/s optical signals in
`OXC—X. Then, the BERS of the main optical sig~
`nal were measured for the non-overmodulated
`
`signal and the overinodulated signal with an m of
`8.2% and 14.6%, respectively. Measurements
`showed. that the power penalty at a BER of 10“
`was 0.5 dB for an m_ of 8.2%, and 1.5 dB for an m
`of 14.6% (Figure 1100)). The PT signals were
`correctly detected by the PT receiver in OXC—Y.
`We suppose that a PT with an m of 10% will not
`interfere with the main signal very much, and we
`confirmed the optical path connection by using the
`PT scheme.
`
`The prototype system was successfully
`demonstrated with an operation system in
`SupercoInm’98. Figure 12 shows a photograph
`of the prototype system at the event.
`
`{June 9 «11, 1998)
`
`Figure 12
`Prototype OX_C system at Supercomm’98 in Atlanta.
`
`4. Summary
`We are just at the beginning of photonic net-
`
`working, which will enable the transfer _ of
`extremely large amounts oftraffic over optical f"-
`‘ ber and provide the foundation for efficiency,
`flexibility, and reliability in the next~generation
`backbone networks. The key network elements of
`OADM and OXC using advanced optical devices
`such as the AOTF and PLLOSS SW are now be-
`
`to provide
`available
`ing made
`management ofthc optical path layer.
`
`efficient
`
`References
`
`A
`
`1) K. Sato, S. Okamoto, and H. Hadaina:
`Network Performance and Integrity En—
`hancement with Optical Path Layer
`Technologies.
`IEEE J~SAC, SAC-12, 1,
`pp.159—170 (Jan. 1994).
`G—R~2979-CORE: Common Generic Require-
`ments fo1"Opti(:al Add-Drop Multiplexers
`(OADMS) and Optical Terminal Multiplexers
`(0TMs). Issue 1, Bellcore, Apr. 1998.
`II. Miyata, I-I. Onaka, K. Otsuka, and T.
`Chikama: Bandwidth and ripple require-
`ments for cascaded optical (de)multiplexers
`in multiwavelength optical networks. Proc.
`OFC’97, TuE3, Feb. 1997.
`Taniguchi, Y.
`T. Nakazawa, M. Doi,
`Takasu, and M. Seino: Ti: LiNbO3 AOTF
`for 0.8 nm channel-spaced WDM systems.
`
`FUJITSU Sci. Tech. J.,35, 1,(July 1999)
`
`Cisco Systems, Inc.
`Exhibit 1047, Page 9
`
`

`
`T Clzikruna et al.: Photonic Networking Using Optical Add Drop Multiplexers and Optical Cross-Cmmects
`
`OFC’98, PD1, Feb. 1998.
`K. Otsuka, T. Maki, Y. Sampei, Y. Tachikawa,
`
`N. Fukushima, and T. Chikama: A high-per-
`formance optical spectrum monitor with
`high-speed measuring time for WDM optical
`networks. Proc. ECOC’97, pp.147-150, Sep.
`1997.
`
`M. Koga, A. Watanabe, S. Okamoto, K. Sato,
`H. Takahashi, and M. Okuno: Optical Path
`Cross—connect Demonstrator Designed to
`Achieve 320 Gbit/s. Proc. ECOC’96, ThC.3.1,
`Sep. 1996.
`P. Granestrand, L. Thylen, B. Stoltz, K.
`
`Bergvall, W. Doldissen, H. Heidrich, and D.
`Hoffmann: Strictly nonblocking 8 X 8 inte-
`grated optic switch matrix in Ti: LiNbO-3.
`Proc. OFC’86,WAA3, Feb. 1986.
`‘
`
`8)
`
`T. Simoe, K. Hajikano, and K. Murakami:
`A path-independent insertion loss optical
`space switching network‘ Tech. Dig. ISS’87,
`4, 012.2, pp.999-1003 (1987).
`Y. Hamazumi and M. Koga: Transmission
`Capacity of Optical Path Overhead Transfer
`
`Scheme Using Pilot Tone for Optical Path
`Network.
`IEEE J Lightwave Technol., 15,
`12, pp.2197-2205 (Dec. 1997).
`Proposed Overhead Channel Realization for
`Optical Layers: ITU—T SG15, Delayed Con-
`tribution, D. 68, NTT, Geneva, April 1997.
`F. Heisxnann, M.’I‘. Fatehi, S. K. Korotky, and
`J. J. Veselliaz Signal Tracking and Perfor-
`mance Monitoring in Multi~Wavelength
`Optical Networks. Proc. ECOG’96, WeB.2.2,
`Sep. 1996.
`'
`
`Satoshi Kuroyanagi received the B.E.
`degree
`in Electrical Engineering
`from.Nagoya institute of Technology,
`Nagoya, Japan in 1986. He joined»
`Fujitsu Laboratories Ltd., Kawasaki,
`Japan in 1986 and has been engaged
`in research of photonic switching sys-
`tems and optical cross-connect sys-
`tems. He is a member of the institute of
`Electronics, information and Communi-
`cation Engineers (IEICE) of Japan.
`
`Email : kuroya@flab.fuji’(su.co.jp
`
`Terumi Chikama received the BS.,
`M.S., and Ph.D degrees in Physics from
`the University of Tokyo, Tokyo, Japan
`in 1977, ‘I979, and 1982, respectively.
`Since 1982 he has been with Fujitsu
`Laboratories Ltd., Kawasaki, Japan,
`where he has been engaged in research
`and development of high-speed optical
`transmission systems, wavelength
`division multiplexing systems, and pho-
`lonic networking. He is a member of
`the iEEE and the Institute of Electronics,
`information and
`Communication Engineers (IEICE) of Japan.
`
`E-mail : chikama@flab.fujiisu.co.jp
`
`Hiroshi Onaka received the B.S.
`degree in Electrical Engineering from
`KANAGAWA iNST|TUTE OF TECH-
`NOLOGY, Kanagawa, Japan in 1982.
`From 1982 to 1984 he was with the
`same university as a research associ-
`ate. Since, 1985 he has been with
`Fujitsu Laboratories Ltd., Kawasaki,
`Japan, where he has been engaged in
`research and development of coherent
`Iightwave transmission and optical
`wavelength division multiplexing transmission systems. He is a
`member oithe Institute of Electronics, Information and Commu-
`nication Engineers (IEICE) of Japan and the Japan Society of
`Applied Physics (JSAP).
`
`E-mall : onaka@flab.fujitsu.co.jp
`
`FUJITSU Sci. Tech. J.,35, 1,(.Ju|y1999)
`
`Cisco Systems, Inc.
`Exhibit 1047, Page 10