SpringerSeriesin optical sciences 90/2 FoundedbyH.K.V.Lotsch Editor-in-Chief: W.T.Rhodes,Atlanta EditorialBoard: T.Asakura,Sapporo K.-H.Brenner,Mannheim T.W.Hänsch,Garching T.Kamiya,Tokyo F.Krausz,ViennaandGarching B.Monemar,Linköping H.Venghaus,Berlin H.Weber,Berlin H.Weinfurter,Munich Springer NewYork Berlin Heidelberg HongKong London Milan Paris Tokyo SpringerSeriesin optical sciences The Springer Series in Optical Sciences, under the leadership of Editor-in-Chief William T. Rhodes, GeorgiaInstituteofTechnology,USA,andGeorgiaTechLorraine,France,providesanexpandingselec- tionofresearchmonographsinallmajorareasofoptics:lasersandquantumoptics,ultrafastphenomena, opticalspectroscopytechniques,optoelectronics,quantuminformation,informationoptics,appliedlaser technology,industrialapplications,andothertopicsofcontemporaryinterest. Withthisbroadcoverageoftopics,theseriesisofusetoallresearchscientistsandengineerswhoneed up-to-datereferencebooks. Theeditorsencourageprospectiveauthorstocorrespondwiththeminadvanceofsubmittingamanu- script.SubmissionofmanuscriptsshouldbemadetotheEditor-in-ChieforoneoftheEditors.Seealso http://www.springer.de/phys/books/optical_science/ Editor-in-Chief WilliamT.Rhodes FerencKrausz GeorgiaInstituteofTechnology ViennaUniversityofTechnology SchoolofElectricalandComputerEngineering PhotonicsInstitute Atlanta,GA30332-0250,USA Gusshausstrasse27/387 E-mail:[email protected] 1040Wien,Austria E-mail:[email protected] and Max-Planck-InstitutfürQuantenoptik Hans-Kopfermann-Strasse1 EditorialBoard 85748Garching,Germany ToshimitsuAsakura Hokkai-GakuenUniversity BoMonemar FacultyofEngineering DepartmentofPhysics 1-1,Minami-26,Nishi11,Chuo-ku andMeasurementTechnology Sapporo,Hokkaido064-0926,Japan MaterialsScienceDivision E-mail:[email protected] LinköpingUniversity 58183Linköping,Sweden E-mail:[email protected] Karl-HeinzBrenner ChairofOptoelectronics HerbertVenghaus UniversityofMannheim InstituteofComputerEngineering Heinrich-Hertz-Institut B6,26 fürNachrichtentechnikBerlinGmbH 68131Mannheim,Germany Einsteinufer37 E-mail:[email protected] 10587Berlin,Germany E-mail:[email protected] TheodorW.Hänsch HorstWeber Max-Planck-InstitutfürQuantenoptik TechnischeUniversitätBerlin Hans-Kopfermann-Strasse1 OptischesInstitut 85748Garching,Germany Strassedes17.Juni135 E-mail:[email protected] 10623Berlin,Germany E-mail:[email protected] TakeshiKamiya MinistryofEducation,Culture,Sports HaraldWeinfurter ScienceandTechnology Ludwig-Maximilians-UniversitätMünchen NationalInstitutionforAcademicDegrees SektionPhysik 3-29-1Otsuka,Bunkyo-ku Schellingstrasse4/III Tokyo112-0012,Japan 80799München,Germany E-mail:[email protected] E-mail:[email protected] Mohammed N. Islam (Ed.) Raman Amplifiers for Telecommunications 2 Sub-Systems and Systems Foreword by Robert W. Lucky With286Figures MohammedN.Islam DepartmentofElectricalEngineeringandComputerScience UniversityofMichiganatAnnArbor 1110EECSBuilding 1301BealAvenue AnnArbor,MI48109-2122 [email protected] and XteraCommunications,Inc. 500WestBethanyDrive,Suite100 Allen,TX75013 USA [email protected] LibraryofCongressCataloging-in-PublicationData Ramanamplifiersfortelecommunications2:sub-systemsandsystems/editor,MohammedN.Islam. p.cm.–(Springerseriesinopticalsciences;v.90/2) Includesbibliographicalreferencesandindex. ISBN0-387-40656-5(hc.:alk.paper) 1.Fiberoptics. 2.Opticalcommunications. 3.Ramaneffect. 4.Opticalamplifiers. I.Islam,MohammedN. II.Series. TL5103.592.F52R352003 (cid:1) 621.38275–dc21 2003044945 ISBN0-387-40656-5 ISSN0342-4111 Printedonacid-freepaper. ©2004Springer-VerlagNewYork,Inc. Allrightsreserved.Thisworkmaynotbetranslatedorcopiedinwholeorinpartwithoutthewritten permissionofthepublisher(Springer-VerlagNewYork,Inc.,175FifthAvenue,NewYork,NY10010, USA),exceptforbriefexcerptsinconnectionwithreviewsorscholarlyanalysis.Useinconnectionwith anyformofinformationstorageandretrieval,electronicadaptation,computersoftware,orbysimilaror dissimilarmethodologynowknownorhereafterdevelopedisforbidden. Theuseinthispublicationoftradenames,trademarks,servicemarks,andsimilarterms,eveniftheyare notidentifiedassuch,isnottobetakenasanexpressionofopinionastowhetherornottheyaresubject toproprietaryrights. PrintedintheUnitedStatesofAmerica. 9 8 7 6 5 4 3 2 1 SPIN1094888090/2 www.springer-ny.com Springer-Verlag NewYork Berlin Heidelberg AmemberofBertelsmannSpringerScience+BusinessMediaGmbH To my loving wife, Nasreen Foreword IremembervividlythefirsttimethatIheardaboutthefiberamplifier.Atthattime, of course, it was the erbium-doped fiber amplifier, the predecessor of the Raman amplifierthatisthesubjectofthisbook. ItwasanearlymorninginaforgottenyearinMurrayHill,NewJerseyatoneof ourBellLabsmonthlyresearchstaffmeetings.Abouttwentydirectorsandexecutive directors of research organizations clustered around a long table in the imposing executiveconferenceroom.ArnoPenzias,thevicepresidentofresearch,presidedat thefootofthetable. Everyonewhoparticipatedinthoseresearchstaffmeetingswilllongremember theircultureandatmosphere.Arnowouldpickanarbitrarystartingpointsomewhere aroundthetable,andthedesignatedpersonwouldheadtothefrontofthetabletogive ashorttalkon“somethingnew”inhisorherresearcharea.Thisfirstspeakerwould invariablyfiddlehelplesslywiththecontrolsembeddedinthepodiumthatcontrolled the viewgraph projector, but eventually we would hear machinery grinding in the back room as a large hidden mirror moved into place. We would all wait quietly, arranging and choosing our own viewgraphs from the piles that lay on the table in frontofeveryparticipant. Therulesforthestaffmeetingwerethateachspeakerwasallowedsevenminutes andthreeviewgraphs.However,inspiteofArno’sbesteffortstoenforcethisregimen, everyonetooktoolongandusedtoomanyviewgraphs.Variousattemptsatusingloud timersandotherincentivesallfailed.Noonecouldgivearespectabletalkonaresearch topicforwhichtheyhadpassionatefeelingsinsevenminutes. Another rule was that anyone could forfeit his talk by simply saying, “I pass.” This forfeiture was always accepted without comment, but new directors always asked their friends about whether this would constitute a black mark against their performance.Nooneknewforsure,butrumorhaditthatitwasunwisetopassunless youweretrulydestituteofmaterial.Afterall,theimplicationwouldbethattherewas nothingnewinyourresearchorganizationforthelastmonth—notagoodindication ofyourmanagementskills. Withnoonepassing,andeveryonespeakingtoolong,thesestaffmeetingssome- timesseemedendless.Computerscientistswouldtalkaboutnewconstructsinsoft- viii Foreword ware,systemspeoplewouldtalkaboutnewtechniquesforspeechrecognition,physi- cistswouldtalkaboutsomenewlaser,chemistswouldshowdiagramsofneworganic materials,andsoforth.Itdidn’ttakelongforeachtalktoexceedtheunderstanding ofmostlistenersofwhateverspecialtywasbeingdiscussed.Ialwaysleftwithapro- foundsenseofthelimitationsofmyownknowledge,butwithanexhilaratinginkling intotheunfoldingofscience.Itwas,perhaps,thebestoftheoldresearchmodelin BellLabs,andinretrospectIcansaythatinthisnewcompetitiveworldImissthose oldscientific-stylemanagementmeetings. ItwasinsuchameetingthatIfirstheardaboutthefiberamplifier.Idon’tknow whetherIhadbeenpayingattention,butIwasimmediatelygalvanizedbytheimpli- cationsofthisnewdiscovery.Onewordcametomeandblazedacrossmymind:that wordwas“transparency.” Surprisingly, in my experience I am not always immediately enthusiastic about anewtechnologyuponinitialexposure.Onemightthinkthatthepotentialsofgreat breakthroughs are self-evident, but that does not seem to be the case. When I first heardabouttheinventionofthelaser,Ihadnopremonitionthatlaserswouldbecome the primary instrument of the world’s telecommunications traffic.When one of the inventorsofpublickeycryptographytoldmehisideaforhavingtwokeys,Iscoffed atthenaivetéofhisconcept.Irememberthinkingonfirsthearingaboutwhatisnow theprincipalalgorithmfordatacompressionthatIthoughtitwasonlyatheoretical exercise.Somytrackrecordforsuchinsightsisnotaltogethergood. However,withthefiberamplifierIwenttotheotherextreme.Iforesawadramatic revolution in communications. I spoke up at the staff meeting that morning to say that this invention would transform the architecture of communications networks. Thiswouldleadtotransparentnetworks,Isaid,andthatthiswouldnotnecessarily begoodforAT&T.Igotcarriedawaywiththisvision,andwentontosaythatpri- vatenetworkscouldhavetheirownwavelengthstravelingtransparentlythroughthe network,untouchedbythecommoncarrierinthemiddle.Oneprivatenetworkmight have“blue”light(figurativelyspeaking,ofcourse,becausewe’renottalkingabout visible wavelengths) whereas another would have “green.” I foresaw a plug on the wallthatpassedonlythechosenwavelength,whichwouldbeownedexclusivelyby thatparticularcustomer’snetwork.AT&Twouldthusbedeprivedoftheopportunity to process signals for value-added services.AT&T, in fact, wouldn’t have any idea whatwaspackedintothosewavelengths. Well,thathasn’texactlyhappened,buttoday’sopticalnetworksaremovingto- wards increased transparency, and Raman amplifiers will accelerate this trend.The advantagesoftransparencyarecompelling.Agreatmanyconstituentsignalscanbe amplifiedcheaplyinonefellswoop.Moreimportantly,thisamplificationisindepen- dentofthebitrates,protocols,waveforms,multiplexing,oranyotherparticularsof thetransmissionformat.Thedesignisn’t“lockedin”toanyspecificformat,andas thesedetailschange,theamplificationremainsaseffectiveasever.Inthecaseofthe Raman amplifier, the bandwidth is so enormous that adjectives seem inadequate to describeitspotentialforbulkamplification. Transparencyinthenetworkissoattractivethatprobablytheonlyreasonitisn’t done is that it is so difficult to achieve. One reason is, of course, the necessity for periodicallyunbundlingthesignaltoaddordropsubcomponents.Inthedigitalworld Foreword ix thishasusuallymeantacompletedemultiplexingandremultiplexingoftheoverall signal,anexpensiveoperation.Theopticalworldopensupthepossibilityofselective transparency for certain wavelengths whereas others are unpacked to do add-drop multiplexing. Sonetworktopologysetslimitsontransparency.Buttheotherreasontransparency ishardtoachieveistheimplicitaccumulationofimpairmentsasasignalincurssucces- siveamplifications.Itisironicthatthetelephonenetworkwasessentiallytransparent forthefirsthalf-centuryofitsexistence.Until1960thelong-haultransmissionsys- tems used analogue amplification to boost levels as the signal traversed the nation. Theinventionofthetriodevacuumtubeenabledthefirsttranscontinentaltransmis- sionsystemtobedeployedinthe1920s.Itwasamarvelousfeattobeabletosenda bandofsignals3000milesacrossthecountry,passingthroughmanyamplifiers,ac- cumulatingnoiseanddistortionalongtheway,butstillprovidingintelligiblespeech attheotherend.Someolderreaderswillrememberwhenlongdistancephonecalls soundedcracklyand“distant.”Now,ofcourse,itisimpossibletotellhowfaraway aconnectionis.Theyallsoundlocal,becauseofdigitaltransmission. Digitaltransmissionwasthetriumphofthe1960s.Thoughnowitseemsobvious, engineersfoundthephilosophyofdigitizationhardtograspforseveraldecadesafter theinventionofpulsecodemodulationbyReevesin1939.Thereisatrade-offhere: bandwidthagainstperfectibility.A3kHzvoicesignal,forexample,istransformedby ananalogue-to-digitalconverterintoa64,000kbpsstreamofbits,greatlyexpanding the necessary transmission bandwidth. However, this digital signal can be regener- atedperfectly,removingnoiseanddistortionperiodicallyasnecessary.Amiracleis achievedasthebitsarriveacrossthecountryinthesamepristineformaswhenthey left. So it was that all long distance transmission was converted to digital format. The introduction of the first lightwave transmission systems hurried this change, inasmuchaslightwavesystemsweredeemedtobe“intrinsicallydigital”becauseof theirnonlinearitiesandthelackofamplifiers.Noonecaredmuchatthetime—the early1980s—buttheentiredesignofthenetworkwaspredicatedonthetransmission of64kbpsvoicechannels.Themultiplexinghierarchy,theelectronicswitching,the synchronizationandtiming,andthetransmissionformatassumedthateverythingwas packaged into neat little voice channels.That, of course, was before the rise of the Internet. Nowopticalamplificationhasreversedthistrendofthelasthalf-centurytowards digitizationbaseduponahierarchyofvoicesignals.Itisn’tjustthatopticalamplifiers haveanenormousbandwidth.Theydosomethingthoseoldtriodevacuumtubescould neverdo:theyamplifywithoutsubstantiallyincreasingthenoiseanddistortionofthe signal.Ramanamplifiersareparticularlygoodinthisway.Moreover,becauseRaman amplificationisdistributedacrossthewholespanofthefiber,thesignallevelnever drops as low as it does when discrete amplifiers are employed. In a system using discrete amplifiers the signal level is at its lowest and most vulnerable right before thepointofamplification. Back at that research staff meeting I was concerned about the implications of transparencytothearchitectureofthenetwork.Atransparentnetworkis,bydefinition, a “dumb” network. It doesn’t do anything to the signal; it can’t, because it doesn’t x Foreword know what the signal is. As an AT&T employee, that sounded threatening. As an Internetuser,thatsoundedempowering.TheInternet,afterall,wasdesignedaround theso-calledend-to-endprinciple.InthearchitectureoftheInternet,intelligenceisat theperipheryofthenetwork,andthenetworkisasminimallyintrusiveasisnecessary toachieveinterconnection.Itisanextremelyimportantphilosophicalprinciplethat wasjustbeginningtobeunderstoodinthe1980s.Sincethentheargumenthasraged, andtheconceptofa“stupid”networkhasbeenputforwardbyanumberofInternet designersastheultimatedesiredobjective.Ifthatisso,thentheopticalamplifierhas madepossibletheultimatestupidnetwork. I can’t leave this foreword without mentioning another observation on perhaps a more personal level. Raman amplifiers epitomize for me the transformation of communications from a world of electrical circuits to one of quantum mechanical phenomena.Ofcourseyoucouldarguethattransistorsthemselvesdependonquan- tum mechanical principles, and surely the laser does, and so forth. But for many practicalanddesignpurposesthesedevicescouldbemodeledwithtraditionalcircuit equivalents.Sincethen,however,photonicshasincreasinglybecomeashowpieceof modernphysics.Theerbium-dopedfiberamplifierhadtobeunderstoodasaquantum interactionoflightwiththeerbiumatom.Ramanamplifiers,bycontrast,involvethe interactionoflightwithamaterialstructure.Wedescendevermoreintotherealmof quantumphenomena,intoaworldofsmallandimpressivemiracles. A number of my friends and associates at Bell Labs have contributed to this technologyandeventothisparticularbook.I’mveryproudoftheworkthattheyand their peers in academia and other industries have done in the creation of photonics technology. I’ve seen it grow around me and have taken vicarious pride in their accomplishments.SometimesItellpeoplethat,yes,Iknowtheinventorsofthisor thatgreattechnology,eventhoughImaynothaverealizedatthetimethesignificance oftheinvention.InthecaseofRamanamplifiersIrememberlearningaboutRaman effectsasoneoftheimpairmentstobeovercomeinopticaltransmission.Researchers in my organization were even then experimenting with Raman amplification, and althoughtherewasaglimmerofpotential,Ican’tsaythatIwasawareofwhattheir future might bring. Perhaps now its day has come, and that’s what this book is all about. RobertW.Lucky FairHaven,NewJersey March2003 Preface Technologies for fiber-optic telecommunications went through a major growth period—somemightevensayarevolution—roughlyduringtheyears1994to2000. Thisgrowthcameaboutduetotheconvergenceofseveralmarketdriversandtech- nologies. First were data traffic and the Internet, the key drivers of the demand for bandwidth. Prior to the explosion of data traffic and the Internet, voice traffic only grewatanaverageof4%ayear.TheInternet,ontheotherhand,grew100%ayear ormorestartingin1992andsustainedthisphenomenalgrowthrateatleastthrough about 2001. The second was the advent of the optical amplifier, which served the role in optical networks that the transistor had played in the electronics revolution. Theopticalamplifierwaskeybecauseitallowedthesimultaneousamplificationof a number of channels, as opposed to electronic regenerators that operated channel by channel. The third technology was wavelength-division-multiplexing (WDM), which made a single strand of fiber act as many virtual fibers. WDM has allowed thecapacityoffiberstobeincreasedbymorethantwoordersofmagnitudeoverthe pastfewyears,providingplentyofbandwidthtofuelthegrowthofdatatrafficand the Internet. WDM served the role in optical networks that integrated circuits had played in the electronics revolution. Just as the transistor permitted the revolution associatedwithintegratedcircuitsinelectronics,theopticalamplifierpermittedthe revolutionassociatedwithWDMinopticalnetworks.Becauseanumberofchannels couldbesimultaneouslyamplified,thecostofdeployingmorewavelengthsinWDM wasgatedbytheterminalendcostsratherthantheregeneratorcosts.Hencefarmore cost-effectivenetworksbecameavailablewiththecombinationofopticalamplifiers andWDM. Ramanamplificationhasbeenoneoftheopticalamplifiertechnologiesthathad a slow start, but then experienced a wide deployment with increasing performance needsofopticalnetworks.Itwouldbereasonabletoassumethatalmosteverynew or upgraded long-haul (∼300 to 600 km between regenerators) and ultra-long-haul (>600kmbetweenregenerators)willeventuallydeploysomeformofRamanampli- fication technology.Any deployment concerns about discrete or distributed Raman amplification have been outweighed by the performance improvements permitted withRamanamplification.Forexample,distributedRamanamplificationimproves