3-7 highly efficient optical communication technologies · transmission capacity (section 2) and...

$: 3-7 Highly Efficient Optical Communication Technologies · transmission capacity (Section 2) and high-density optical signal processing technology on scales smaller than the diffraction$
61MIYAZAKI Tetsuya and NARUSE Makoto

1 Introduction

In response to expanding needs in infor-mation transmission, conventional opticalcommunication technologies have seenincreased bit rates per wavelength channel—to2.5 Gb/s, 10 Gb/s, and 40 Gb/s—in addition toincreases in the extent of wavelength multi-plexing［1］. However, with increasing bit rates,waveform distortion due to dispersion isbecoming a serious problem. The extent ofmultiplexing is also limited by the amplifica-tion band of the repeater amplifiers, such thattwo or more optical amplifiers with differentamplification bands must be used in parallel.This requirement causes a number of prob-lems, including increased power consumptionin the repeater system. In addition, anothersignificant problem is seen in the ultimateimplementation of functionality in the opticaldomain, as with an optical data router. Howev-er, the diffraction limit of light (approximately

1μm in communication wavelengths, which isapproximately 100 times the line width of aVLSI circuit) represents a serious barrier inoverall system integration, and many are hop-ing to see a breakthrough in this area. In short,demand is now high for the development ofoptical technologies that will facilitate a fun-damental, widespread improvement in effi-ciency (e.g., in the efficiency of frequencyuse, energy use, and size). In light of these cir-cumstances, this article describes the latesttechnological movements in optical signalprocessing aimed at achieving high efficiency.In particular, we will focus on multi-leveltransmission technology to enhance opticaltransmission capacity (Section 2) and high-density optical signal processing technologyon scales smaller than the diffraction limit oflight, based on near-field optical interactions(Section 3).

3-7 Highly Efficient Optical CommunicationTechnologies

3-7 – Multi-level Optical Transmission and Ultra-highDensity Optical Signal Processing –

MIYAZAKI Tetsuya and NARUSE Makoto

Highly efficient optical communication technologies are becoming important since the con-ventional strategies, such as denser wavelength multiplexing or higher time-domain multiplex-ing, are approaching their physical limits. Power-efficiency and volume-efficiency are additionalimportant demands required in optical communications. In this paper, we show our recent devel-opment in high-efficient optical communications such as multi-level transmission technologiesand high-density optical signal processing performed in a scale smaller than the diffraction limitof light.

KeywordsHighly efficient optical communication, Multi-level, Synchronous detection, Nanophotonics

62 Journal of the National Institute of Information and Communications Technology Vol.53 No.2 2006

2 Time-domain high-efficiencyoptical signal processing tech-nology (multi-level transmissionmethod)

The multi-level transmission method hasgenerated high expectations as a means ofhighly efficient optical transmission, with thepotential to increase the capacity of informa-tion transmission by a factor of two to four ormore［2］-［8］. Optical fiber communication is anextremely underdeveloped technology interms of frequency use relative to wirelesscommunication, which is already reaching theShannon limit. Thus, optical communicationpresents the possibility of significant furtherdevelopment, including improved efficiencyin information transmission and signal-pro-cessing technologies. Many methods of multi-level optical modulation have been reported todate. One example is seen in ASK-DPSK(APSK)［2］, which combines Amplitude ShiftKeying (ASK) and Differential Phase ShiftKeying (DPSK); another method is known asDifferential Quadrature Phase Shift Keying(DQPSK)［3］, which uses the optical phaseonly. However, the ASK-DPSK method,which combines optical phase modulation andoptical intensity modulation, presents theproblem of waveform distortion caused byintermodulation in optical phase modulation(PM) caused by the optical amplitude modula-tion (AM) component. On the other hand, theDQPSK method requires a signal encoderentailing complicated operations in the electri-cal signal stage［3］, which causes problems inrealizing high-speed transmission of 40 Gb/sor greater. In addition, Differential Phase ShiftKeying for multi-level transmission of 8 PSKor more is difficult for reasons similar to thoselimiting corresponding transmission in wire-less communication. To the best of our knowl-edge, any established method for stable real-time multi-level optical phase synchronousmodulation/demodulation with phase noisetolerance has not been reported. We have pro-posed the inverse optical pulse hybrid modula-tion method, which superposes optical phase

modulation on inverse optical pulse signals, ina multi-level transmission system with lowintermodulation. This section reports on thesatisfactory intermodulation suppressionoffered by this method. Here we also discussthe Pilot Carrier aided self-homodyne detec-tion method, which features superior phasenoise resistance and stability. Overall, thismethod is aimed at the realization of ultra-high-speed multi-level transmission.

2.1 Inverse optical pulse hybrid mod-ulation method［5］

Figure 1 shows a timing chart for the pro-posed inverse optical pulse hybrid modulationmethod. Introducing pulsed light into aninverse RZ modulator turns the “1” code inthe data signal for optical amplitude modula-tion (“1010011” in Fig. 1) into a dip (darkoptical pulse) and the “0” code into a highlevel, converting the input light into inverseRZ optical pulses with a duty ratio of 50 per-cent or greater (Fig. 1 (b)). The inverse RZoptical pulses are then further superposed withphase modulation (“πππ0ππ0” in Fig. 1). Theconventional optical phase/amplitude hybridmodulation method (Fig. 1 (a)) purposelydegrades the extinction ratio in order to super-pose phase modulation on low-level opticalpulses corresponding to the “0” code. Thus,delicate adjustment of the extinction ratio isrequired to balance the quality of the two sig-nal components—optical phase and opticalamplitude—in addition to accurate matching

Fig.1 Timing chart for inverse opticalpulse hybrid modulation method


of the time slots for amplitude modulation andphase modulation. On the other hand, inverseRZ optical pulses feature a duty ratio of50 percent or more, such that synchronizationof the time slots for phase modulation andamplitude modulation is not necessarilyrequired (Fig. 1 (b)). The receiver unit dividesthe optical signal into two and sends the splitsignals into the optical amplitude detectionunit and the optical phase detection unit.These units separately demodulate the signal.Figure 2 shows the waveforms of the receivedsignal with the proposed method (a) and withthe conventional method (b), for 20-Gb/s opti-cal phase/amplitude hybrid modulation(2 bits/symbol: doubled information transmis-sion efficiency; 20 ps/div). The figure con-firms that the proposed method can signifi-cantly reduce the waveform distortion causedby intermodulation from the optical amplitudecomponent to the optical phase modulationdata component. We have also confirmed thatcombining DQPSK modulation and theinverse RZ optical pulses effectively suppress-es the intermodulation caused by the ampli-tude modulation for 30-Gb/s, 3-bits/symbol(tripled information transmission efficiency)transmission and produces error-free charac-

teristics (i.e., with a bit error rate (BER) lessthan10－9)［5］.

2.2 Pilot Carrier multi-level opticalphase synchronous detectionmethod［6］［7］

Figure 3 shows a conceptual illustration ofthe proposed multi-level optical phase syn-chronous detection method for 2-bits/symboltransmission (or QPSK, for Quadri-PhaseShift Keying). Under the conventional opticalphase synchronous detection method, thetransmitter leaves a slight component from thecarrier, using an incomplete phase modulation,and the receiver traces this slight residualcomponent of the carrier using an opticalphase-locked loop (or “optical PLL”). Thus,extremely stringent phase noise performanceis required, and it is therefore difficult toobtain stable optical phase synchronous detec-tion［8］. For this reason, we allowed transmit-ter input light (linearly polarized at45 degrees) into a phase modulator that onlymodulates the phase of the light in the TMdirection. In this manner, the multi-level phasemodulated optical signal (TM) and unmodu-lated CW light (TE), which acts as a Pilot Car-rier, are generated together. We constructedthe receiver with a module consisting of anLN waveguide with an integrated polarizedbeam splitter, half-wave plate, optical phaseadjuster, and multiplexer (Fig. 4). The polar-ized beam splitter performs homodyne detec-tion by rotating the polarization of only thePilot-Carrier component 90 degrees. Althoughthe receiving system requires polarization con-trol, the phase control system needs only toadjust the relative phase in a narrow band thatincludes the polarization dispersion fluctua-tion. As a result, this method significantlyreduces the requirements with respect to thespectral line width of the light source relativeto conventional optical PLL, which uses alocal oscillator as a light source. Compared tothe Differential Phase Shift Keying method,this approach does not make use of the previ-ous bit in operation. This method thereforedoes not require an encoding/decoding unit,

Fig.2 Waveforms observed in phasedetection for 20-Gb/s opticalphase/amplitude hybrid modula-tion


which would involve complicated processeswithin the electronic circuits, in turn causingproblems in ultra-high-speed transmission.Balanced homodyne detection in the receiveris also expected to suppress intermodulationcaused by the amplitude modulation compo-nent and amplitude noise.

Figure 5 shows the RF frequency charac-teristics of the Common Mode Rejection Ratio(CMRR) for synchronous homodyne detectionof the prototype LN integration module. Whenthe frequencies of the received signal were1 GHz and 10 GHz, we obtained superiorCMRR of 30 dB and 15 dB, respectively. Fig-ure 6 shows the dependence of receiver sensi-tivity on spectral line width (full width at halfmaximum) for the light source, as defined byan error rate of 10－9 at the time of 20-Gb/s(2 bits/symbol) QPSK synchronous homodynedetection. In past experimental reports onQPSK phase synchronous detection based ona local oscillator in the receiver system［4］［8］,the spectral line width of the light source is on

the order of several hundred kHz in all cases(to the left of the dotted line in the figure).However, the proposed method only slightlydegrades receiver sensitivity and produces anextremely fine receiver waveform for a lightsource with a line width of 30 MHz (indicatedas B in the figure), even compared with a casein which line width is 300 kHz (indicated as Ain the figure). This is because the proposedmethod uses the LN module and performsautonomous homodyne detection of the PilotCarrier, which carries a phase noise identicalto that of the signal component. This proce-dure cancels out the phase noise.

Fig.3 Pilot Carrier self-homodyne detection method

Fig.4 LN integration module

Fig.5 RF frequency characteristics of theCommon Mode Rejection Ratio insynchronous detection


3 Highly efficient signal process-ing technology in the spatialdomain (ultra-high-density opti-cal signal processing)

In addition to increased efficiency in thetime domain as discussed in Section 2, anoth-er important problem remains for next-genera-tion optical systems in increasing efficiency inthe spatial domain. The diffraction limit oflight (approximately 1μm in the communica-tion wavelength) is seriously restricting sys-tem integration. To overcome such physicallimitation of conventional optical technolo-gies, we have focused on nanophotonics,based on near-field optical interactions, whichis free from the diffraction of light. We inves-tigated basic problems in high-efficiency opti-cal signal processing in the spatial domain(i.e., high-density optical signal processing)regarding the functionalities required, such asoptical table lookup processing in optical datarouters and optical CDM, based on the physi-cal principles that are uniquely available innanophotonics.

3.1 Local near-field optical interactionand global signal processing

Here we take the optical table lookup pro-cessing required in an optical data router as anexample. Optical table lookup is a matchingoperation between the destination address of

the input packet and the routing table in therouter. Let us consider the basics of its opticalimplementations based on near-field opticalinteractions［9］. We have found that a tablelookup or a matching operation requires thefollowing two basic mechanisms.(1) Input data items are generally constructed

with two or more bits, and we must obtainevaluation results for all of these bits. Inother words, it is not sufficient to clarifymatching or mismatching for each bit;therefore it is necessary to provide a mech-anism of global evaluation among theentirety of the bits (i.e., global summa-tion).

(2) Input data should be compared with asmany data entries in the routing table aspossible. In other words, a data broadcastmechanism is required. Even if the basicprocesses described in (1) may be integrat-ed in a nano-scale, the entire system—con-taining many of these integrated mod-ules—must work as a massively parallelprocessing system. Here, propagating light is, in fact,

extremely suitable for such global functionali-ties as are described in (1) and (2) above, andhas played an important role in conventionaloptical signal processing. In other words, thepropagating character of this conventionallightwave naturally matches the global func-tionalities by means of lenses or optical wave-guides. However, the diffraction of light posesa serious problem; the size of the entire sys-tem generally becomes impractically large. Onthe other hand, one of the ultimate characteris-tics of optical near-fields is its physical locali-ty; it does not propagate. Therefore, one of thebasic problems in the optical signal processingin the nanometer-scale is to achieve “globalfunctionality” based on “physically local”mechanisms. Thus, we will show an exampleof an architecture that takes advantage of thetwo characteristics of optical near-fields: (1)the resonance energy transfer between closelyseparated quantum dots, and (2) the fact thatthis energy transfer is forbidden in conven-tional propagating light.

Fig.6 Dependence of receiver sensitivityon spectral line width in QPSK syn-chronous homodyne detection


3.2 Global summation of dataWe can transfer signals (excitons) to a par-

ticular quantum dot using the near-field opti-cal interaction between quantum dots. Forexample, there is a resonance energy levelbetween a quantum dot of size a and a quan-tum dot of size √2a. The signal (exciton) gen-erated in the smaller dot can move to the larg-er dot through this resonance level［10］. Thus,if the system features a structure in which thesmaller dots surround the larger dots, asshown in Fig. 7, the excitons in the smallerdots move to the larger dots. In this manner,the global summation operation Σxi is imple-mented in nano-scale by an appropriate con-figuration of the quantum dots of appropriatesizes.

Here, the details of the exciton transfermechanism are as follows. As an example,between quantum dot QDA of size a and quan-tum dot QDB of size √2a, resonance energylevels ((1,1,1) for QDA and (2,1,1) for QDB)are present, and the exciton of QDA moves toQDB through this level and transitions to thelower level, (1,1,1), of QDB. The energy dissi-pation involved in this process is attributedonly to the sublevel relaxation in QDB; There-fore this mechanism is also significantly ener-gy-efficient compared both to electric VLSIcircuits and to existing optical technologies.

The lower level of QDB is used for the outputsignal. Here, when the lower level of QDB isoccupied by a signal from other dots, energytransition is forbidden due to Pauli’s exclusionprinciple. In this case, the so-called opticalnutation occurs meaning that the exciton goesback and forth between QDA and the upperlevel of QDB, enabling the exciton to move tothe lower level of QDB in the end. Such par-tially bidirectional signal transition also con-tributes to the global summation. Figure 7 pre-sents an experiment based on this principleusing CuCl quantum dots. The figure indicatesthe output signal (384 nm) for a combinationof input light at three different frequencies(381.3 nm, 376 nm, and 325 nm). The figureconfirms an output signal level in accordancewith the number of input signals and showsthat the system is integrated on a scale smallerthan the diffraction limit［11］.

3.3 Data broadcastingInput data need to be provided for many

data entries in the lookup table. If individualdata items require individual waveguides, thephysical volume necessary for the wiring willform a serious interconnection bottleneck.Here, by taking note of the broadcast struc-ture, we can make use of the fact that the sig-nal-transfer mechanism using optical near-

Fig.7 Global summation of data based on optical near-field interactions


field interactions is forbidden in propagatinglight. For example, propagating light cannotexcite the upper level (2,1,1) of QDB (forbid-den dipole). Thus, if we select light at a fre-quency that does not influence operation with-in the device, we can simultaneously providedata to two or more independent circuitblocks. In other words, if the system isdesigned to use light energy without overlap atenergy level Ωint (relating to the internal oper-ation) and energy levels Ωin and Ωout (relatingto input and output), multiple bits can bebroadcast by using propagating light at differ-ent frequencies. Figure 8 shows an example ofan experiment based on this principle usingthe same CuCl quantum dots described earlier.The figure shows the multiple optical switches(indicated by the closed square, the closed cir-cle, and the closed diamond) in the sample areoperated simultaneously with lights which areuniformly irradiated within the entire system;meaning that individual addressing is notrequired for each of the switches. In otherwords, the principle of broadcast interconnectswas confirmed［12］.

The summation and broadcast processesdescribed above are the basic fundamentalprocesses for optical table lookup operations.However, it covers a wide range of computa-tions since the summation and broadcast leads

to the so-called memory-based architecture,which covers a variety of signal processingapplications. In their practical realization, sta-ble operation at room temperature and opera-tion in the communication wavelength bandwill be important. In this respect, we havedemonstrated the operation of these principlesbased on semiconductor quantum dots［13］andwill continue to pursue further development inthis area.

4 Future developments

For the 100-Gb Ethernet, planned forintroduction around 2010, there is some con-cern that current optical amplifier bandwidthswill prove insufficient. In this context much isexpected of the highly efficient multi-leveloptical transmission method, as a decisivetechnology that can both increase transmissioncapacity and suppress power consumptionwithin a network system. In addition, dynami-cally changing traffic must be accommodatedefficiently while maintaining the desired sig-nal quality. Thus, we are eager to makeprogress on this highly efficient adaptive mod-ulation method, which will enable adaptiveconfiguration of information transfer efficien-cy (at 2 to 4 bits/symbol or more), whenapplied in combination with the highly effi-

Fig.8 Broadcast of data based on the near-field optical interaction


cient optical modulator NICT has been devel-oping to date (See Article 3-6,) as well aswith private-sector technologies such as high-performance FEC. The nanophotonics tech-nology constituting the basics of ultra-high-density optical signal processing can also befused with advanced practical technologiessuch as those relating to quantum dots andmetal nanostructures. The relevant researchand development is rapidly beginning world-wide in key areas of optical technology. Aim-ing at early acquisition of next-generation keytechnologies, we hope to continue to promoteresearch and development as we solidify the

basic technologies involved.

Acknowledgments

We offer our sincere thanks to ProfessorMotoichi Ohtsu of the University of Tokyo,Dr. Tadashi Kawazoe of the Japan Science andTechnology Agency, Dr. Yuichi Matsushima,Vice President of the NICT, and Dr. FumitoKubota, Executive Director of NICT, NewGeneration Network Research Center, as wellas the members of the relevant groups thattook part in discussions on our daily activities.

References01 “An Illustrated : All about Recent Photonic Network Technologies”, Chapter-1, NICT Ultrafast Photonic

Network Group, Nippon Jitsugyo Publishing, ISBN4-534-03914-X, 2005. (in Japanese)

02 K. Sekine, N. Kikuchi, S. Sasaki, S. Hayase, C. Hasegawa, and T. Sugawara, “Proposal and Demonstra-

tion of 10-Gsymbol/sec 16-ary (40 Gbit/s) Optical Modulation / Demodulation Scheme”, in ECOC ’04,

We3.4.5, 2004.

03 R. A. Griffin and A. C. Carter, “Optical differential quadrature phase-shift key (oDQPSK) for high capaci-

ty optical transmission”, in OFC’02, WX6, pp. 367-368, 2002.

04 D.-S. Ly-Gagnon, K. Katoh, and K. Kikuchi, “Unrepeatered optical transmission of 20 Gbit/s quadrature

phase-shift keying signals over 210 km using homodyne phase-diversity receiver and digital signal pro-

cessing”, Electron. Lett., Vol. 41, No. 4, pp. 206-207.2005.

05 T. Miyazaki and F. Kubota, “Superposition of DQPSK over inverse-RZ for 3-bit/symbol modulation/

demodulation”, IEEE Photonics Technology Letters, Vol. 16, No. 12, pp.2643-2645, 2004.

06 T. Miyazaki and F. Kubota, “PSK self-homodyne detection using a pilot carrier for multi-bit/symbol trans-

mission with inverse-RZ signal”, IEEE Photonics Technology Letters, Vol. 17, No. 6, pp.1334-1336. 2005.

07 T. Miyazaki, “Linewidth-Tolerant QPSK Homodyne Transmission Using a Polarization-Multiplexed Pilot

Carrier”,IEEE Photonics Technology Letters, Vol. 18, No. 2, pp. 388-390, 2006.

08 S. Norimatsu, K. Iwashita, and K. Noguchi, “An 8 Gb/s QPSK optical homodyne detection experiment

using external-cavity laser diodes”, IEEE Photonics Technology Letters, Vol. 4, No. 7, pp.765–767, 1992.

09 M. Naruse, T. Miyazaki, T. Kawazoe, K. Kobayashi, S. Sangu, F. Kubota, and M. Ohtsu, “Nanophotonic

computing based on optical near-field interactions between quantum dots”, IEICE Transaction on Elec-

tronics Special Section on Nanophotonics, Vol. E88-C, No. 9, pp. 1817-1823, 2005.

10 T. Kawazoe, K. Kobayashi, J. Lim, Y. Narita, and M. Ohtsu, “Direct Observation of Optically Forbidden

Energy Transfer between CuCl Quantum Cubes via Near-Field Optical Spectroscopy”, Physical Review

Letters, Vol. 88, pp. 067404-1-4, 2002.

11 M. Naruse, T. Miyazaki, F. Kubota, T. Kawazoe, K. Kobayashi, S. Sangu, and M. Ohtsu, “Nanometric

summation architecture using optical near-field interaction between quantum dots”, Optics Letters,

Vol. 30, No. 2, pp. 201-203, 2005.


12 M. Naruse, T. Kawazoe, S. Sangu, K. Kobayashi, and M. Ohtsu, “Optical interconnects based on optical

far- and near-field interactions for high-density data broadcasting”, Optics Express, Vol. 14, pp. 306-

313, 2006.

13 T. Kawazoe, K. Kobayashi, K. Akahane, N. Yamamoto, and M. Ohtsu, “Demonstration of nanophotonic

NOT-gate using near-field optically coupled quantum dots”, Applied Physics B, to appear.

MIYAZAKI Tetsuya, Dr. Eng.

Group leader, Ultrafast Photonic Net-work Group, New Generation NetworkResearch Center (former: Groupleader, Ultrafast Photonic NetworkGroup, Information and Network Sys-tems Department)

Optical Communications Technology

NARUSE Makoto, Dr. Eng.

Senior Researcher, Ultrafast PhotonicNetwork Group, New Generation Net-work Research Center (former: SeniorResearcher, Ultrafast Photonic Net-work Group, Information and NetworkSystems Department)

Optical Systems, Nanophotonics

3-7 highly efficient optical communication technologies · transmission capacity (section 2) and...

Documents