throbber
FPO
`
`SLIM FILMS
`
`T
`
`oday’s network architects have already be-
`gun to build the Optical Internet using tech-
`nologies that can switch some or all of the
`light signals in one incoming optical fiber to
`one or several outgoing fibers. The new-gen-
`eration Optical Internet will serve as a high-
`speed mail delivery vehicle that will bring
`units of data called packets to a point nearby
`a recipient. There the time-consuming process of sorting out
`which packet goes where—the Internet equivalent of the local
`post office—will fall to electronic routers from companies
`such as Cisco Systems and Juniper Networks. Over time, even
`this task of switching individual packets may be taken over
`by routers that use photons, not electrons, for processing.
`The IP packet is the Internet’s basic unit of currency. In
`today’s networks, every e-mail message gets chopped into
`thousands of packets, which are switched over different path-
`ways and reassembled at a final destination using the Internet
`Protocol (IP). The routers, together with other networking
`equipment, convert data from lightwaves to electronic signals
`in order to read the packets and send them along to their fi-
`nal destination: a mail server, where they are reassembled
`into a coherent message before a synthesized voice proclaims,
`“You’ve got mail!” The key is that the routers all along the
`way can easily read the address of each IP packet.
`The lightwave network that does the heavy hauling may
`serve as only one stage in the evolution toward a truly all-
`optical network, however. When networks routinely shuttle
`
`OPTICAL MULTIPLEXER, a component of an optical packet router,
`sends four incoming wavelengths to two output ports.
`
`terabits (trillions of bits) a second, the conversion step into
`electronics may prove too costly both in time and money. In
`coming years, a router will routinely have to break down a
`data stream carrying 40 gigabits (billions of bits) a second
`over a single wavelength into 16 parallel electronic data
`streams, each transmitting 2.5 gigabits a second within the
`router. Moving a massive number of packets every second
`through multiple layers of electronics in a router can lead to
`congestion and degrade network performance.
`
`Light to Light
`
`In response to these problems, network engineers are con-
`
`templating a solution that will use lightwaves to process
`lightwaves and optical switches that can redirect packets at
`blindingly fast speeds. The technology for photonic routing
`switches is still in its infancy. Creation of a photonic packet-
`switched network will depend on overcoming multiple tech-
`nological hurdles equivalent to those that had to be ad-
`dressed by electronics engineers from the mid-1950s to the
`mid-1970s, going from individual capacitors and resistors sol-
`dered onto a circuit board to monolithic integrated circuits.
`Nevertheless, the quest has begun. The Defense Ad-
`vanced Research Projects Agency (DARPA) has funded pro-
`grams in optical packet switching at the University of Cali-
`fornia at Santa Barbara, Telcordia Technologies in Morris-
`town, N.J., Princeton University and Stanford University.
`Alcatel in France, the Technical University of Denmark and
`
`
`
`9696
`
`
`
`Scientific American January 2001Scientific American January 2001
`
`
`
`Routing Packets with LightRouting Packets with Light
`
`Copyright 2000 Scientific American, Inc.
`
`Ex. 1025
`CISCO SYSTEMS, INC. / Page 1 of 4
`
`

`

`The ultimate all-optical
`network will require
`dramatic advances in
`technologies that use
`one lightwave to
`imprint information
`on another
`
`by Daniel J. Blumenthal
`
`ROUTING
`
`PACKETS
`WITHLIGHT
`
`the University of Strathclyde in Scotland have also launched
`research efforts.
`A potential communications technology for photonic
`packet switching in future networks is known as All-Optical
`Label Swapping (AOLS). In AOLS, individual IP packets or
`groups of packets get tagged with an optical label. The IP
`packet is like an envelope with an address on the front
`(called the header) and contents inside (called the payload).
`AOLS attaches a label to the packet by placing it, say, in
`front of the header. The use of labels resembles packing into
`the same mailbag all letters going through the same set of
`major cities on the way to their final destination. Postal
`workers read only the label at each intermediate routing
`point, not each individual envelope.
`In a communications network, an optical router will look
`only at the smaller label and determine which output fiber or
`wavelength to send the packet to, instead of reading the
`header inside the packet. Current photonic technologies do
`not make this task easy. Optical components that perform the
`function of the integrated-circuit elements that read and
`process a label exist mostly as laboratory demonstrations that
`have not made their way into the commercial mainstream.
`Take something as simple as the buffers used for tempo-
`rary storage of the bits in a packet. In designing an electronic
`router, an engineer would hold the packet in a buffer that is
`similar to the dynamic random-access memory (DRAM) used
`in computers. Photons cannot be stored easily, however. So de-
`signers must then figure out ways to buffer the packet by com-
`
`plex optical circuits that do not hold the photons still but
`rather corral the pulses of light into a holding area similar to
`an automobile traffic circle. Other approaches use techniques
`that match how long it takes the pulses in a packet to move
`through a switch. They do so by sending the pulses through a
`predetermined length of fiber. The time it takes to traverse this
`“detour” fiber equals the time needed by the switch to strip off
`the packet label and read the routing information it contains.
`Optical buffers on laboratory benchtops are currently
`hundreds of times larger than the submicron-size dimensions
`of their electronic buffer counterparts. It is not known if the
`optical version of the dense conglomeration of transistors
`found in the Pentium chip will ever come to be a reality.
`Besides buffers, logic circuits that can process labels
`based on light controlling light have also been shown in the
`lab but are a long way from being engineered into a full-
`scale router. The first examples of photonic routers will
`therefore still rely partly on electronics. A packet that enters
`a first-generation AOLS router will have a small amount of
`its optical energy diverted down a separate pathway, where
`it gets converted by a photodetector into an electronic copy
`of the packet and label. The label goes to circuitry that reads
`its contents and computes the next pathway for the packet
`along the network. The electronic processor generates a new
`label, attaches it to the original packet, and sends a control
`signal to the switch to specify a particular wavelength or
`fiber along which the packet should travel [see box on next
`page]. Despite the need to make an optoelectronic conver-
`
`www.sciam.com
`
`Scientific American January 2001 97
`
`Copyright 2000 Scientific American, Inc.
`
`Ex. 1025
`CISCO SYSTEMS, INC. / Page 2 of 4
`
`

`

`Optical Router
`
`1 DEMULTIPLEXER
`
`2 OPTICAL SPLITTER
`
`3 LABEL ERASER
`
`4 LABEL WRITER
`
`5 WAVELENGTH
` CONVERTER
`
`INCOMING FIBER
`
`WAVELENGTHS
`
`PACKET
`
`PACKET
`
`LABEL
`
`ALL-OPTICAL LABEL SWAPPING is an emerging tech-
`nology for photonic switching of individual packets of
`data. A packet enters the switch, gets stripped off and
`read electronically (A) or with lightwave processing ele-
`ments (B). The processors decide on which wavelength
`and fiber a packet should exit the switch. A new label
`gets placed on the packet, which then gets switched to
`a new wavelength and an output fiber.
`
`PHOTODETECTOR
`
`SEMICONDUCTOR-
`OPTICAL
`AMPLIFIER
`
`CONTROL ELECTRONICS
`
`SEMICONDUCTOR-
`OPTICAL
`AMPLIFIER
`
`CONTROL PHOTONICS
`
`A
`
`B
`
`LASER
`
`COMBINER
`
`LAURIE GRACE
`
`sion for the tiny label, perhaps 20 bits
`in size, the optical switches can forgo
`electronic processing of the much larger
`packet, which may range from 40 to lit-
`erally thousands of bytes. The big pack-
`et moves through the switch at the
`higher-speed optical bit rate. Making
`the conversion for the entire packet
`might be prohibitive in cost as well as
`consuming power and space for elec-
`tronic equipment.
`Once the label has been processed,
`the packet finally arrives at the physical
`elements that switch it from the input to
`the output fiber. Optical switching ele-
`ments pose another major engineering
`challenge, because the switching com-
`ponents must be able to send a packet
`to a new fiber or wavelength in a nano-
`second or less. Several technologies have
`emerged that are capable of switching
`packets at the desired speeds. One ex-
`ample is the semiconductor-optical am-
`
`plifier, a device that uses stimulated
`emission of light as the basis for switch-
`ing, the same process that drives a laser.
`To switch a signal on an incoming
`fiber to any of many outgoing ones, the
`amplifier forms an optical bridge be-
`tween the desired input and output
`fibers. When a data-carrying optical sig-
`nal reaches the bridge, an electric cur-
`rent in a control signal injects electrons
`and “holes” (areas where electrons are
`absent) into the amplifier. Light entering
`the amplifier causes the electrons and
`holes to combine with one another, giv-
`ing off more photons that are exact
`copies of the optical signal trying to
`cross the bridge. Once the signal reaches
`a certain level of power, it moves from
`one side of the bridge to the other.
`When the control current is shut off,
`light at the amplifier input is absorbed
`and does not make it through to the
`output fiber. Although the control signal
`
`that opens the gateway is electronic, the
`stream of photons carrying packets
`races through the switch without get-
`ting converted to an electronic signal.
`The technology remains in early de-
`velopment. Investigators have devised
`fast switches with up to eight incoming
`and outgoing fibers each, although for
`marketable products they will need to
`create switches with hundreds or thou-
`sands of input and output ports.
`Semiconductor-optical amplifiers may
`also perform a vital role in realizing a
`technology that can transfer a stream of
`bits directly from one lightwave to an-
`other. This wavelength converter can
`switch packets to multiple output fibers
`by combining the technology with a de-
`vice called a wavelength multiplexer
`that, much like a prism, can separate
`light into its individual colors. The
`wavelength converter will determine
`how to route packets to avoid switching
`
`98
`
`Scientific American January 2001
`
`Routing Packets with Light
`
`Copyright 2000 Scientific American, Inc.
`
`Ex. 1025
`CISCO SYSTEMS, INC. / Page 3 of 4
`
`

`

`7 MULTIPLEXER
`
`6 BUFFER
`
`ELECTRONIC
`CONTROL
`SIGNAL
`
`OPTICAL
`CONTROL SIGNAL
`
`OUTPUT
`FIBERS
`
`1 Data on an incoming fiber get
`separated by wavelength (demulti-
`plexed) into different pathways in
`the router.
`
`2 An optical splitter makes a copy
`of the packet and sends it either to
`a control module, which may be ei-
`ther electronic (A) or photonic (B).
`The control module reads only the
`label, discarding the rest of the
`packet.
`
`3 The other copy of the packet goes
`to a label eraser, where the label is
`removed.
`
`4 A label writer then writes a new
`label on receipt of a signal from the
`control modules—denoting the next
`destination on the network where
`the packet is to be sent.
`
`5 Another signal from a control
`module directs the wavelength con-
`verter to send a packet from one
`wavelength to another. (The same
`process occurs for packets being
`channeled on the incoming blue,
`green and red wavelengths.)
`
`6 An optical buffer holds a packet
`until the control module commands
`it to send the packet through a
`multiplexer.
`
`7 A multiplexer directs the packet-
`containing wavelengths to one of
`the output fibers.
`
`conflicts. If two packets in different
`fibers both need to be transferred to the
`red wavelength in a given output fiber,
`the wavelength converter will route one
`of the packets to a green wavelength to
`ensure that they do not conflict.
`
`Interacting Wavelengths
`
`Experimental converters have em-
`
`ployed photonic integrated-circuit
`(PIC) technology, the optical equivalent
`of integrating electronic components on
`microchips. In a converter built with
`PICs, light from a laser moves along
`waveguides, similar to optical fibers. In
`some types of optical processing, a
`wavelength carrying a stream of bits in-
`teracts with another wavelength, im-
`printing information on it. The light-to-
`light transfer occurs when the stream of
`optical bits causes the waveguide to
`change the phase of the recipient light-
`
`wave, which then exits into an output
`fiber. A change in phase might represent
`a digital one and no change a zero.
`To make converters feasible, research-
`ers have devoted massive efforts to the
`development of tunable lasers that can
`be set to different wavelengths onto
`which bits from incoming packets can
`be modulated. Currently laboratory la-
`sers allow an incoming data stream to
`be switched to any of 80 wavelengths,
`and that number will grow to hundreds
`in the future.
`Ultimately even the decisions on how
`to route a packet may by necessity be-
`come all-optical. As packet speeds on a
`given wavelength exceed 160 gigabits a
`second, a typical packet will race through
`the switch in a nanosecond, whereas it
`would take 100 nanoseconds to process
`a label using electronics. In essence, the
`address label will have become larger
`than the envelope (packet) onto which it
`
`is affixed. Increasing the speed at which
`the label moves through the processor—
`to beyond 100 gigabits a second—might
`tax the limits of the electronic control
`circuitry (although new exotic electron-
`ic technologies that could operate at
`speeds above 100 gigabits per second
`are being tested).
`Anticipating this problem, a few labo-
`ratories have fashioned early prototypes
`of ultrahigh-speed optical logic gates
`that can be built from the same technol-
`ogy used to build light-controlled switch-
`es. These devices—developed at Prince-
`ton, the M.I.T. Lincoln Laboratory and
`British Telecom Labs, among others—
`utilize different configurations of semi-
`conductor-optical amplifiers or other op-
`tical materials to implement simple opti-
`cal Boolean logic gates (that is, AND,
`XOR and NOT) that can process light
`signals moving at speeds in excess of 250
`gigabits a second.
`These gates have made extremely sim-
`ple packet-routing decisions in the labo-
`ratory but cannot yet be scaled up to the
`number needed to make complex rout-
`ing decisions required for a commercial
`switch. Still, device integration and new
`optical switching technologies may one
`day make photonic control possible,
`marking the true advent of the all-opti-
`cal router. If such issues can be resolved,
`a packet might travel from New York
`City to Los Angeles through IP routers
`and never pass through electronics.
`
`SA
`
`DANIEL J. BLUMENTHAL is co-
`founder of Calient Networks, head-
`quartered in San Jose, Calif., which
`makes intelligent optical switches based
`on microelectromechanical devices. He
`is on a leave of absence from the Uni-
`versity of California, Santa Barbara,
`where he is associate professor in the
`department of electrical and computer
`engineering and associate director for
`the Center on Multidisciplinary Opti-
`cal Switching Technology (MOST),
`which is funded by the Defense Ad-
`vanced Research Projects Agency.
`
`Further Information
`Photonic Packet Switching Technolo-
`gies, Techniques and Systems. Special
`issue of Journal of Lightwave Technology
`(IEEE/OSA). Vol. 16, No. 12; December
`1998.
`Optical Networking Solutions for
`Next-Generation Internet Networks.
`Marco Listanti and Roberto Sabella in IEEE
`Communications Interactive; September
`2000. Available by subscription at www.
`comsoc.org/~ci/ on the World Wide Web.
`
`www.sciam.com
`
`Scientific American January 2001 99
`
`Copyright 2000 Scientific American, Inc.
`
`Ex. 1025
`CISCO SYSTEMS, INC. / Page 4 of 4
`
`

This document is available on Docket Alarm but you must sign up to view it.


Or .

Accessing this document will incur an additional charge of $.

After purchase, you can access this document again without charge.

Accept $ Charge
throbber

Still Working On It

This document is taking longer than usual to download. This can happen if we need to contact the court directly to obtain the document and their servers are running slowly.

Give it another minute or two to complete, and then try the refresh button.

throbber

A few More Minutes ... Still Working

It can take up to 5 minutes for us to download a document if the court servers are running slowly.

Thank you for your continued patience.

This document could not be displayed.

We could not find this document within its docket. Please go back to the docket page and check the link. If that does not work, go back to the docket and refresh it to pull the newest information.

Your account does not support viewing this document.

You need a Paid Account to view this document. Click here to change your account type.

Your account does not support viewing this document.

Set your membership status to view this document.

With a Docket Alarm membership, you'll get a whole lot more, including:

  • Up-to-date information for this case.
  • Email alerts whenever there is an update.
  • Full text search for other cases.
  • Get email alerts whenever a new case matches your search.

Become a Member

One Moment Please

The filing “” is large (MB) and is being downloaded.

Please refresh this page in a few minutes to see if the filing has been downloaded. The filing will also be emailed to you when the download completes.

Your document is on its way!

If you do not receive the document in five minutes, contact support at support@docketalarm.com.

Sealed Document

We are unable to display this document, it may be under a court ordered seal.

If you have proper credentials to access the file, you may proceed directly to the court's system using your government issued username and password.


Access Government Site

We are redirecting you
to a mobile optimized page.





Document Unreadable or Corrupt

Refresh this Document
Go to the Docket

We are unable to display this document.

Refresh this Document
Go to the Docket