Experimental demonstration and numerical simulations of 16×10 Gb/s DWDM repeaterless transmission...

168 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 27, NO. 3, FEBRUARY 1, 2009

Experimental Demonstration and NumericalSimulation of 107-Gb/s High Spectral Efficiency

Coherent Optical OFDMQi Yang, Yan Tang, Yiran Ma, and William Shieh, Member, IEEE

Abstract—Orthogonal frequency-division multiplexing (OFDM)is a multicarrier modulation format in which the data are trans-mitted with a set of orthogonal subcarriers. Recently, thismodulation format has been actively explored in the field ofoptical communications to take advantages of its high spectralefficiency and resilience to chromatic and polarization disper-sion. However, to realize the optical OFDM at 100 Gb/s andbeyond requires extremely high electronic bandwidth for theelectronic signal processing elements. In this paper, we investigateorthogonal-band-multiplexed OFDM (OBM-OFDM) as a suitablemodulation and multiplexing scheme for achieving bandwidthscalable and spectral efficient long-haul transmission systems. TheOBM-OFDM signal can be implemented in either RF domain,or optical domain, or a combination of both domains. Using thescheme of OBM-OFDM, we show the successful transmissionof 107 Gb/s data rate over 1000-km standard single-mode fiber(SSMF) without optical dispersion compensation and withoutRaman amplification. The demonstrated OBM-OFDM system isrealized in optical domain which employs 2 2 MIMO-OFDMsignal processing and achieves high optical spectral efficiency of3.3 bit/s/Hz using 4-QAM encoding. Additionally, we performnumerical simulation of 107-Gb/s CO-OFDM transmission forboth single-channel and wavelength-division-multiplexed (WDM)systems. We find that the -factor of OBM-OFDM measuredusing uniform filling of OFDM subbands is in fact more conserva-tive, in particular, is 1.2 dB and 0.4 dB lower than using randomfilling for single-channel and WDM systems, respectively.

Index Terms—Anti-alias filtering, coherent communications,mixed-signal design, orthogonal frequency-division multiplexing(OFDM), polarization diversity, quadrature amplitude modula-tion (QAM), wavelength-division multiplexing (WDM).

I. INTRODUCTION

O RTHOGONAL frequency-division multiplexing(OFDM) has been extensively studied to combat RF

microwave multipath fading and has emerged as the leading

Manuscript received July 01, 2008; revised September 28, 2008. Current ver-sion published February 13, 2009.

Q. Yang, Y. Tang, and Y. Ma are with Victoria Research Labora-tory, National ICT Australia, Department of Electrical and ElectronicEngineering, University of Melbourne, Melbourne, VIC, 3010, Aus-tralia (e-mail: [email protected]; [email protected];[email protected]; [email protected]).

W. Shieh is with ARC Special Research Centre for Ultra-Broadband Infor-mation Networks, Department of Electrical and Electronics Engineering, Uni-versity of Melbourne, Melbourne, VIC 3010, Australia (e-mail: [email protected]).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/JLT.2008.2007134

modulation technology for the wireless and wire-line systemsin RF domain. Recently, an optical equivalent of RF OFDMcalled coherent optical OFDM (CO-OFDM) has been proposed[1] and has become a promising technique for high spectral ef-ficiency and dispersion resilient transmission [2], [3]. As the IPtraffic continues to grow at a rapid pace, the 100-Gb/s Ethernetis being considered as the next-generation transport standardfor IP networks [4]. As the data rate approaches 100 Gb/sand beyond, the electrical bandwidth required for CO-OFDMwould be at least 15 GHz [5] and is not cost-effective toimplement even with the best commercial digital-to-analogconverters (DACs) and analog-to-digital converters (ADCs) insilicon integrated circuit (IC) [6]. To overcome this electricalbandwidth bottleneck, we propose and demonstrate the conceptof OBM-OFDM to divide the entire OFDM spectrum intomultiple orthogonal bands. Due to the inter-band orthogonality,the multiple OFDM bands with zero or small guard bands canbe multiplexed and de-multiplexed without inter-band interfer-ence. With this scheme, transmission of 107-Gb/s CO-OFDMsignal over 1000 km (10 100 km) standard single-modefiber (SSMF) has been realized using only erbium-doped fiberamplifiers (EDFAs) and without a need for optical dispersioncompensation. Although several transmission experiments at100 Gb/s and above have been demonstrated at longer distancesrelying on a dispersion-compensation module and Raman am-plification (RA) in each span [7], [8], our work has achieved the1000-km transmission without optical dispersion compensationand without RA beyond 100 Gb/s. The 107-Gb/s OBM-OFDMcan be also considered as 5 21.4 Gb/s WDM channels withoutfrequency guard band, occupying 32-GHz optical bandwidth,implying a high spectral efficiency of 3.3 bit/s/Hz using only4-QAM encoding. To the best of our knowledge, we have alsoachieved a record back-to-back OSNR sensitivity of 15.8 dB at107 Gb/s for a BER of .

By multiplexing and demulitpelxing multiple OFDM bands,OBM-OFDM has the following advantages: 1) high spectral ef-ficiency can be achieved by allowing for zero or small guardband; 2) OBM-OFDM offers the flexibility of demodulating twoOFDM subbands simultaneously with just one FFT whereasthree (I)FFTs would be otherwise needed for the same pur-pose; 3) OBM-OFDM can be readily partitioned with electricalanti-alias filters, and subsequently processed with lower-speedDAC/ADCs [6]; and 4) the required cyclic prefix length is short-ened due to the subbanding of the overall spectrum.

In this paper, we extend and improve over the previous107-Gb/s OBM-OFDM transmission experiment [5], and

0733-8724/$25.00 © 2009 IEEE

YANG et al.: EXPERIMENTAL DEMONSTRATION AND NUMERICAL SIMULATION OF 107-Gb/s HIGH SPECTRAL EFFICIENCY COHERENT OPTICAL OFDM 169

Fig. 1. Conceptual diagram of OBM-OFDM. Anti-alias filters I and II corre-spond to two detection approaches illustrated in Section II.

substantiate the experiment with numerical simulation. Thispaper is organized as follows. In Section II, the principle ofOBM-OFDM is presented where a complete OFDM spectrumis partitioned into multiple orthogonal bands. In Section III, twoimplementations of OBM-OFDM in RF domain and opticaldomain are illustrated. The RF implementation can be realizedin a mixed-signal CMOS ASIC design. A detailed descriptionof OBM-OFDM experimental setup is described in Section IV.In Section V, we discuss the experimental results with a focuson the OSNR sensitivity and nonlinearity performance. InSection VI, the influence of frequency guard band is investi-gated by varying the amount of the guard band up to ten timesof the subcarrier spacing. Insignificant penalty is observedwhen the guard band equals to multiple times of the subcarrierspacing, namely, when the condition of the orthogonality issatisfied. To corroborate the experimental results, numericalsimulation is conducted to investigate 107-Gb/s CO-OFDMtransmission in both single-channel and WDM systems inSection VI. Finally, in Section VII we draw the conclusions.

II. PRINCIPLE OF OBM-OFDM

The basic principle of OBM-OFDM is to partition theOFDM into multiple subbands, while maintaining their orthog-onal property. As shown in Fig. 1, the entire OFDM spectrumcomprises OFDM bands, each with the subcarrier spacing of

, and band frequency guard spacing of . The subcarrierspacing is identical for each band due to using the samesampling clock within one circuit. The orthogonal conditionbetween the different bands is given by

(1)

Namely, the guard band is multiple ( times) of subcarrierspacing. In doing so, the orthogonality condition is satisfied forany two subcarriers inside the complete OFDM spectrum. Forinstance, the subcarrier in band 1 is orthogonal to anothersubcarrier in different OFDM band (band 2). Especially,when equals to 1 in (1), the OFDM bands can be multi-plexed/de-multiplexed even without guard band, despite the factthat they originate from different bands. We call this method ofsubdividing OFDM spectrum into multiple orthogonal bands“orthogonal-band-multiplexed OFDM” (OBM-OFDM). Anidentical bandwidth-scalable and spectral-efficient multiplexingscheme for CO-OFDM has been first proposed in [9], whereit is called cross-channel OFDM (XC-OFDM). We adopt theterm of OBM-OFDM to stress the bandwidth reduction throughsubbanding the OFDM spectrum.

Using such a scheme, each OFDM subband can be de-mul-tiplexed using an anti-alias filter slightly wider than the signalband. To detect OBM-OFDM, two approaches can be used, asshown in Fig. 1. First, the receiver laser is tuned to the centerof each band. Each band is detected separately by using an“anti-alias filter I” that low-passes only one-band RF signal.Second, the local laser is tuned to the center of the guard band.Two bands are detected by using an “anti-alias filter II” thatlow-passes two-band RF signal simultaneously. In either case,the inter-band interference can be avoided because of the or-thogonality between the neighboring bands, despite the leakageof the subcarriers from neighboring bands.

III. IMPLEMENTATION OF OBM-OFDM

OBM-OFDM can be implemented in either electrical or op-tical domain. As mentioned earlier, OBM-OFDM is particu-larly suitable to be realized with mixed-signal ICs to resolveADC/DAC bandwidth bottleneck, while the optical realizationof OBM-OFDM serves as an alternative to the other spectrallyefficient multiplexing schemes including coherent WDM [10],all-optical OFDM [11] and electrooptically subcarrier -multi-plexed OFDM [12].

A. Electrical Implementation of OBM-OFDM

Fig. 2 shows the conceptual diagrams for implementing theOBM-OFDM using mixed-signal circuits. In Fig. 2(a), eachOFDM baseband transmitter is implemented using digital ICdesign. The subsequent up-conversion, band-filtering, andRF amplification can be implemented in RF IC design. Theoutput of the OFDM baseband transmitter is filtered throughan anti-alias filter and up-convert to appropriate RF band withthe center frequency from to using an IQ modulatoror a complex multiplexer, the structure of which is shown inFig. 2(c). The range of to is centered around zero, givenby

(2)

where is the center frequency of the th OFDM band,is the band spacing, and is the maximum band number. Theoutput of each IQ modulator is a complex value that has real andimaginary parts as shown in Fig. 2(c). These complex signalsare further summed up at the output, namely, real and imagi-nary parts are added up in separate parallel paths. The combinedcomplex OFDM signal is used to drive an optical IQ modulatorto be up-converted to optical domain [9], [13]. At the receive endas shown in Fig. 2(b), the incoming signal is split into multiplesub bands and down-converted to baseband using IQ demod-ulators. Anti-alias filters should be used to remove unwantedhigh frequency components at the output of the demodulators.In such a way, the DAC/ADC only needs to operate at the band-width of each OFDM band, which is approximately scaled downby a factor equal to the number of subbands from the originalcomplete OFDM spectrum. For instance, the bandwidth of 107Gb/s data rate with QPSK modulation and polarization multi-plexing is around 35 GHz. If the number of subbands is five,each OFDM band will only need to cover about 7 GHz opticalbandwidth. The electrical bandwidth required is 3.5 GHz, or


Fig. 2. Schematic of OBM-OFDM implementation in mixed-signal circuits for (a) the transmitter, (b) the receiver, and (c) the IQ modulator/demodulator. Boththe output from the transmitter in (a) and the input to the receiver in (b) are complex signals with real and imaginary components.

half of the OFDM band spectrum if direct conversion is usedat transmit and receive ends. The DAC/ADC with bandwidthof 3.5 GHz can be implemented in today’s technology [6], andusing a wider bandwidth for each OFDM band will reduce thenumber of the OFDM bands further down to two or three.

B. Optical Implementation of OBM-OFDM

The OBM-OFDM could be realized using either subcarriermultiplexing [3] or wavelength multiplexing to patch multipleorthogonal bands into a complete OFDM spectrum (Fig. 1). TheOBM-OFDM can be also optically implemented by transmit-ting OFDM data through many WDM channel and locking allthe lasers to the common optical standard such as an opticalcomb [14]. In doing so, the orthogonality condition is satisfiedfor all subcarriers across the entire WDM channels. This formof OFDM transmission is called cross-channel OFDM (XC-OFDM) [9]. An optical filter or an electrical filter with band-width slightly larger than the channel bandwidth can be usedto select the desired channel. Consequently, no frequency guardband is necessary between neighbouring WDM channels.

IV. EXPERIMENTAL SETUP AND DESCRIPTION

Although the electronic OBM-OFDM is more cost-effectivesolution, the related research work will involve expensivehigh-speed mixed signal design, foundry run, and chip testing.We choose optical multiplexing to obtain OBM-OFDM for

Fig. 3. Experimental setup for 107 Gb/s OBM-OFDM systems.

proof-of-concept demonstration at 107 Gb/s. Fig. 3 shows theexperimental setup for 107 Gb/s CO-OFDM transmission. Itis stressed that this experiment is carried out recently and themeasurement data shown in this paper is completely different


Fig. 4. Multiple tones generated by two cascaded intensity modulators.

than the prior publication [5]. The detail of the experimentalsetup will be described in the ensuing subsections.

A. CO-OFDM Transmitter

The 107-Gb/s OBM-OFDM signal is generated by multi-plexing 5 OFDM (sub)bands. In each band, 21.4-Gb/s OFDMsignals are transmitted with both polarizations. The multifre-quency optical source spaced at 6406.25 MHz is generatedby cascading two intensity modulators (IM). The guard bandequals to just one subcarrier spacing [ in (1)]. Fig. 4shows the multiple tones generated by this cascaded archi-tecture using two IMs which is different than previous setupusing one IM and one phase modulator (PM) [15]. The newsetup enables better flatness across the five bands, and lessleakage spectrum out of the intended five bands. Only themiddle five tones with large and even power are used forperformance evaluation. The transmitted signal is generatedoff-line by MATLAB program with a length of PRBSand mapped to 4-QAM constellation. The digital time domainsignal is formed after IFFT operation. The total number ofOFDM subcarriers is 128, and guard interval is 1/8 of theobservation window. The middle 82 subcarriers out of 128are filled, from which 4 pilot subcarriers are used for phaseestimation. The I and Q components of the time-domain signalare uploaded onto Tektronix Arbitrary Waveform Generator(AWG), which provides the analog signals at 10 GS/s for bothI and Q parts. Fig. 5(a) shows the electrical spectrum of the I/Qchannel at the output of AWG. It can be seen that the aliasingcomponents of OFDM signal are present above 6 GHz. Whencombining multiple OFDM subbands, such aliasing frequencycomponents will degrade the signals in the adjacent bands.A 3-GHz low-pass electrical filter is used to eliminate thealiasing OFDM components. Fig. 5(b) shows the electricalspectrum after low-pass filtering, where the aliasing spectrumcomponents are removed.

The AWG is phase locked to the synthesizer through 10-MHzreference. The optical I/Q modulator comprising two MZMswith 90 phase shift is used to directly impress the baseband

Fig. 5. Electrical spectrum for (a) directly at the output of the AWG and (b)after 3-GHz anti-aliasing filters.

OFDM signal onto five optical tones. The modulator is biased atnull point to completely suppress the optical carrier and performlinear RF-to-optical up-conversion [13]. The optical output ofthe I/Q modulator consists of five-band OBM-OFDM signals.Each band is filled with the same data at 10.7-Gb/s data rateand is consequently called “uniform filling” in this paper. Toimprove the spectrum efficiency, 2 2 MIMO-OFDM is em-ployed, that is, two OFDM transmitters are needed to send twoindependent data into each polarization, which are then detectedby two OFDM receivers one for each polarization.

A cost-effective method is adopted to emulate the two trans-mitters as shown below: the single-polarization optical OFDMsignal at the output of the I/Q modulator is first evenly spittedinto two polarization branches with a polarization-beam splitter(PBS), with one branch delayed by one OFDM symbol period,


Fig. 6. (a) RF spectrum for the 107-Gb/s signal using a polarization diversitycoherent receiver. The band numbers are depicted next to the correspondingbands. (b) The RF spectrum at the receiver after the 3.8-GHz anti-alias filter.

i.e., 14.4 ns in this experiment, which equals to one OFDMsymbol length. The two polarization branches are subsequentlycombined, emulating two independent transmitters, one on eachpolarization, resulting in a composite data rate of 21.4 Gb/s. Thetwo polarization components are completely independent due tothe delay of 14.4 ns for each OFDM symbol.

B. Fiber Link

A multispan fiber link is emulated with a recirculation loop,which consists of 100-km SSMF fiber and an EDFA to compen-sate the link loss. Neither optical dispersion compensation norRaman amplification is used for the transmission.

C. CO-OFDM Receiver

The signal is coupled out of the loop and received with apolarization diversity coherent receiver [16], [17] comprising apolarization beam splitter, a local laser, two hybrids, and fourbalanced receivers. The complete OFDM spectrum, comprisedof five subbands, is shown in Fig. 6(a). The entire bandwidth for107-Gb/s OFDM signal is only 32 GHz, which is about 5 GHztighter than that shown in the previous experiment [5]. This isbecause: 1) the guard band is reduced from 8 to 1 times of thesubcarrier spacing and 2) the number of the pilot subcarriers isreduced from 8 to 4.

The local laser is tuned to the center of each band, and the RFsignals from the four balanced detectors is firstly passed throughthe anti-aliasing low-pass filters with a bandwidth of 3.8 GHz,such that only a small portion of the frequency components fromother bands is passed through, which can be easily removedduring OFDM signal processing. The performance of each bandis measured independently. The detected RF signals are thensampled with a Tektronix time domain-sampling scope (TDS)at 20 GS/s. The sampled data is processed with a MATLAB pro-gram to perform 2 2 MIMO-OFDM processing. The receiver

signal processing [16], [17] involves: 11) FFT window synchro-nization using Schmidl format to identify the start of the OFDMsymbol; 2) software estimation and compensation of the fre-quency offset; 3) channel estimation in terms of Jones MatrixH; 4) phase estimation for each OFDM symbol; and 5) constel-lation construction for each carrier and BER computation.

The channel matrix is estimated by sending 30 OFDMsymbols using alternative polarization launch. The total numberof OFDM symbol evaluated is 500 symbols. Mathematically,the transmitter information symbol of the two polarizations inthe forms of Jones vector are given by

(3)

where and are transmitted OFDM information symbols fortwo polarizations, Assume the fiber transmission Jones Matrix

is

(4)

Ignoring the additive noise, the two received OFDM symbol isthus given by

(5a)

or equivalently

(5b)

From (5b), the transmitted information symbols can be recov-ered from the received signals by inverting H

(6)

The training symbols are generated by filling the odd symbolswith known random data, while nulling the even symbols.After the polarization multiplexing emulator, the training sym-bols form a pattern of alternative polarization launch for twoconsecutive OFDM symbols. Using odd training symbols, theassociated channel estimation can be expressed as

(7)

and using even training symbols as

(8)

It can be seen from (7) and (8) that, by using alternative polar-ization training symbol, the full channel estimation of can beobtained. Then, using the inverse of this matrix in (6) and the


TABLE I(a) BER DISTRIBUTION AT BACK-TO-BACK AT AN OSNR OF 17.0 dB. (b) BER

DISTRIBUTION AFTER 1000 km TRANSMISSION AT AN OSNR OF 18.9 dB

received information symbols, the transmitted symbols in thetwo polarizations can be estimated. The estimated transmittedsymbols will be demapped to the closest constellation points torecover the transmitted digital bits.

V. MEASUREMENT AND DISCUSSION

Fig. 6(a) shows the RF spectrum after 1000-km transmis-sion measured with the polarization diversity coherent receivershown. No frequency guard band is used in our trans-mission measurement. It can be seen that five OFDM bands eachwith 6.4 GHz bandwidth are closely patched together and the en-tire OFDM spectrum occupied is about 32 GHz. The out-bandcomponents are due to the multifrequency source generationnot tightly bounded at five tones. This artifact will not exist inthe real application using either subcarrier multiplexing or op-tical multiplexing OBM-OFDM. Fig. 6(b) shows the detectedelectrical spectrum after using a 3.8 GHz electrical anti-aliasfilter for one-band detection. The anti-alias filter is critical forOBM-OFDM implementation. As is shown in Fig. 6(a), withoutelectrical anti-alias filter, the electrical spectrum will be as broadas 16 GHz, indicating that at least 32 GS/s ADC has to be used.However, the filtered spectrum in Fig. 6(b) can be easily sam-pled with 20 GS/s or even at a lower speed of 10 GS/s. Ad-ditionally, despite the fact that there are some spurious compo-nents from the neighboring band that is leaked at the edge of the3.8-GHz filter, since they are orthogonal subcarriers to the inter-ested OFDM subcarriers at the center, they do not contribute tothe interference degradation. Compared with our previous work[5], the spectrum of 107-Gb/s CO-OFDM is reduced from 37 to32 GHz by using zero guard band and less pilot subcarriers inthis experiment.

Tables I(a) and (b) shows the performance of five bands atboth back-to-back and 1000-km transmission. It can be seen thatboth polarizations in each band can be recovered successfully,and this is done without the need for a polarization controllerat receive. The difference of BER in each entry is attributedto the tone power imbalance and instability as well as the re-ceiver imbalance for two polarizations. The nonlinearity toler-ance of OBM-OFDM transmission is first measured by varyingthe launch power into the optical links. Fig. 7 shows the mea-sured as a function of launch power at 1000 km. It can be seenthat the optimal power for 107-Gb/s OBM-OFDM is around

1 dBm.Fig. 8 shows the BER sensitivity performance for the entire

107 Gb/s CO-OFDM signal at the back-to-back and 1000-km

Fig. 7. System performance of 107 Gb/s OBM-OFDM signal as a function ofthe launch power at the reach of 1000 km.

Fig. 8. BER sensitivity of 107 Gb/s CO-OFDM signal at the back-to-back and1000-km transmission.

transmission with the launch power of 1 dBm. The BER iscounted across all five bands and two polarizations. It can beseen that the OSNR required for a BER of is, respectively,15.8 and 16.8 dB for back-to-back and 1000-km transmission.The back-to-back 15.8 dB of OSNR sensitivity at 107 Gb/s isabout 1.2 dB better than that reported in the previous publica-tion [5]. This improvement is attributed to three factors: 1) thefive tones generated in this paper have much better flatness thanthose in [5]; 2) four pilot subcarriers are used instead 8 in [5];and 3) the balance between the two arms of polarization diver-sity coherent receiver is improved. To the best of our knowledge,the OSNR of 15.8 dB for a BER of is the record sensitivityfor 107-Gb/s transmission. Fig. 9 shows the system perfor-mance of the 107-Gb/s CO-OFDM signal as a function of reachup to 1000 km. It can be seen that the reduces from 17.2 to12.5 dB when reach increases from back-to-back to 1000 kmtransmission. By improving the balance between our polariza-tion diversity coherent receivers, the disparity between the twopolarizations has been improved from 1.5 dB in [5] to 0.49 dBin this report.

VI. GUARD BAND INFLUENCE ON OBM-OFDM

As discussed in Section II, the orthogonality condition is sat-isfied when the guard band is integer times of subcar-rier spacing . Here, we show a generalized study of the influ-ence of guard band to the system performance. We first verifythrough experiment the validity of the orthogonality conditionthat minimizes the inter-band interference (IBI).

The experiment setup is the same as Fig. 3. Due to the IBIthe subcarriers at the two edges of each band bear the largest


Fig. 9. �-factor of 107-Gb/s CO-OFDM signal as a function of transmissiondistance.

Fig. 10. SNR sensitivity performance of two edge subcarriers at (a) back-to-back transmission and (b) 1000-km transmission. The guard band frequency isnormalized to the subcarrier spacing.

inter-band penalty. Fig. 10(a) and (b) shows the SNR sensitivityperformance of the “edge subcarriers” (the first and the last sub-carrier of the band) as a function of the guard band normal-ized to the subcarrier spacing, at the back-to-back and 1000-kmtransmission with 1 dBm launch power. For simplicity, onlyone polarization is presented. The system signal-to-noise ratio(SNR) oscillates as the guard spacing increases with a step sizeof half of the subcarrier spacing. It is shown in theory that ICIinterference due to frequency spacing is a sinc function [18].The SNR oscillation eventually stabilizes to a constant valueof around 10.5 dB. By comparing with the stabilized constant,the system penalty as a function of the guard band can be in-vestigated. Fig. 10 shows the system penalty for both polariza-tions, when the guard band equals to a multiple of the subcar-rier spacing, the penalty almost decreases to zero at 1000-kmtransmission, validating the orthogonal condition.

Fig. 11. Conceptual diagram of simulated CO-OFDM system.

VII. SIMULATION OF 107-Gb/s OBM-OFDM TRANSMISSION

In the experiment shown in Sections IV and V, five OFDMbands with the same data are multiplexed (uniform filling) togenerate an OBM-OFDM signal. There is a concern whetherthis uniform filling will underestimate the nonlinearity such thatour experimental may overestimate the CO-OFDM transmis-sion performance. Therefore, it is important to compare the per-formance difference between the uniform filling used in the ex-periment and the scenario for which each band is filled with in-dependent data (random filling). Furthermore, while the experi-ment is more convincing and will be the ultimate method of val-idation, the numerical simulation is more flexible and can carryout many more measurement scenarios with relatively low cost.In this section, we intend to conduct the simulation to comparethe nonlinearity performance difference between the uniformlyfilled OBM-OFDM and randomly filled OBM-OFDM, in bothsingle-channel and WDM environments.

The basic CO-OFDM system with polarization multiplexedtransmission is shown in Fig. 11. We configure the simulationcondition in the way such that it closely mirrors our experi-mental setup. Two optical OFDM signals are generated inde-pendently and combined with a polarization beam combiner.Multiple WDM channels with CO-OFDM modulation formatare launched into the optical link. The optical link consists ofmultispan SSMF fibers and an EDFA to compensate for the loss.The output signal after transmission is detected using an opticalcoherent receiver which serves as an optical-to-RF OFDM di-rect down-converter. In the OFDM receiver, the signal is firstsampled using an ADC and demodulated by performing FFTto recover the data. We apply commonly used parameters forour simulation: 100-km span distance, fiber chromatic disper-sion of 16 ps/nm/km, 0.2-dB/km loss, and a nonlinear coeffi-cient of m W. The fiber span loss is compensatedby an EDFA with a gain of 20 dB and noise figure of 6 dB. Thelinewidths of the transmit and receive lasers are assumed to be100 kHz. Eight WDM channels are spaced at 50 GHz. The sameOFDM parameters with that in experiment are used: OFDMsymbol period of 12.8 ns. The number of OFDM subcarriers foreach band is 128 (640 for five bands) and the middle 82 subcar-riers out of 128 are filled. The guard interval is set to one eighthof the observation period, and 4-QAM encoding is used for each


Fig. 12. System � performance of 107 Gb/s OBM-OFDM signal as a func-tion of fiber launch power after 1000 km for both single-channel and WDMCO-OFDM transmission.

subcarrier. The bandwidth of each CO-OFDM signal is 32 GHz.The relative phase shift between subcarriers and channel esti-mation are calculated by using training sequences. The phasedrift from the laser phase noise is also estimated and compen-sated using pilot subcarriers described in [19]. For the uniformfilling, the user data is mapped onto 82 subcarriers and the samedata block is repeatedly used to fill out the rest 328 subcarriers.In this way, the transmitted OFDM signal similar to that used inthe experiment can be simulated, that is, the five bands are filledwith the identical data. For the random filling, the data mappedonto each subcarrier has no correlation [w4].

Fig. 12 shows the Q factor as a function of launch power underthe uniform filling and the random filling after 1000 km of fiberfor single-channel and WDM transmission. The single-channeloptimal launch power is, respectively, 0 and 1 dBm for uniformfilling and random filling, whereas the optimal launch powerfor WDM systems is both 0 dBm for both uniform filling andrandom filling. We can see that if the power is increased beyondthese optimal values, the nonlinearity of optical fiber will de-grade the performance of system, resulting in a reduced . Onthe other hand, if the input power is reduced from these optimalvalues, the optical signal noise ratio (OSNR) of the receivedsignal becomes lower, also resulting in a reduced . For bothsingle-channel and WDM transmission, uniform filling showsworse nonlinearity performance. The reason is that the uniformfilling increases the peak-to-average power ratio (PAPR) dueto the correlation between the bands, which will severely de-grade the signal at the high launch power. The phase walk-offand de-correlation between neighbouring bands arising from thefiber dispersion helps reduce the nonlinearity after long distancefiber transmission. However, there is still residual differenceafter 1000 km transmission between uniform filling and randomfilling.

The slight system performance underestimation is observedby using the uniform filling in comparison with random fillingdue to enhanced nonlinearity in the uniform filling. For thesingle-channel transmission, the optimum launch power dif-ference between uniform filling and random filling is 1 dBand the optimum difference is about 1.4 dB. In the case ofWDM, two filling schemes have the same optimum launchpower of 1 dBm and have the difference of 0.4 dB. It

can be seen that at the fixed launch power of 1 dBm for thesingle-channel scenario, the difference between the randomfilling and uniform filling is less than 0.5 dB, which agrees withthe discussion in [5]. Most importantly, the simulation signifiesthat for single channel transmission, the true (random filled)107-Gb/s OBM-OFDM system will have better system perfor-mance than what we have observed in Sections IV and V. Theconclusion also applies to the WDM transmission. Therefore,OBM-OFDM with the uniform filling provides a conservativeestimation of true OBM-OFDM system. This observation isimportant because, to completely de-correlate the neighboringband data, multiple AWGs and band mixer have to be used,which in turn leads to a much more expensive proposition forconstructing the experimental setup. It is also noted that atthe launch power of 1 dBm, the experimental of 12.5 dB(Fig. 7) is about 3.3 dB below the simulated for the uniformlyfilled single channel at the reach of 1000 km. This differenceis attributed to the imperfection of CO-OFDM transmitterand receiver, and OFDM signal processing involving nonidealphase estimation and channel estimation.

VIII. CONCLUSION

We have proposed and elucidated the principle of orthog-onal-band-multiplexed OFDM (OBM-OFDM) to subdivide theentire OFDM spectrum into multiple orthogonal bands. As a re-sult, the DAC/ADCs do not need to operate at extremely highsampling rate. We show the proof-of-concept transmission ex-periment through optical realization of OBM-OFDM. A nu-merical simulation for both single-channel and WDM transmis-sion is conducted to verify the effectiveness of the demonstratedOBM-OFDM experiment system. To the best of our knowl-edge, we have achieved a record back-to-back OSNR sensitivityof 15.8 dB at the data rate of 107 Gb/s for a BER of .The demonstrated system employs 2 2 MIMO-OFDM signalprocessing and achieves high electrical spectral efficiency of3.3 SSMF bit/s/Hz.

REFERENCES

[1] W. Shieh and C. Athaudage, “Coherent optical orthogonal frequencydivision multiplexing,” Electron. Lett., vol. 42, pp. 587–589, 2006.

[2] W. Shieh, X. Yi, and Y. Tang, “Transmission experiment of multi-gi-gabit coherent optical OFDM systems over 1000 km SSMF fiber,” Elec-tron. Lett., vol. 43, pp. 183–185, 2007.

[3] S. L. Jansen, I. Morita, N. Takeda, and H. Tanaka, “20-Gb/s OFDMtransmission over 4,160-km SSMF enabled by RF-Pilot tone phasenoise compensation,” presented at the Optical Fiber Commun. Conf.,Anaheim, CA, 2007, Paper PDP15, unpublished.

[4] M. Duelk, “Next generation 100 Gb/s Ethernet,” presented at theECOC, 2005, Tu3.1.2, unpublished.

[5] W. Shieh, Q. Yang, and Y. Ma, “107 Gb/s coherent optical OFDMtransmission over 1000-km SSMF fiber using orthogonal band multi-plexing,” Opt. Exp., vol. 16, pp. 6378–6386, 2008.

[6] H. Sun, K.-T. Wu, and K. Roberts, “Real-time measurements of a 40Gb/s coherent system,” Opt. Exp., vol. 16, pp. 873–879, 2008.

[7] P. J. Winzer, G. Raybon, and M. Duelk, “107-Gb/s optical ETDM trans-mitter for 100 G Ethernet transport,” presented at the Eur. Conf. OpticalCommun., Glasgow, Scotland, 2005, Paper Th4.1.1, unpublished.

[8] C. R. S. Fludger, T. Duthel, D. van den Borne, C. Schulien, E.-D.Schmidt, T. Wuth, E. de Man, G. D. Khoe, and H. de Waardt, “10� 111Gbit/s, 50 GHz spaced, POLMUX-RZ-DQPSK transmission over 2375km employing coherent equalisation,” presented at the Optical FiberCommun. Conf., Anaheim, CA, 2007, Paper PDP22, unpublished.

[9] W. Shieh, H. Bao, and Y. Tang, “Coherent optical OFDM: Theory anddesign,” Opt. Exp., vol. 16, pp. 841–859, 2008.


[10] A. D. Ellis and F. C. G. Gunning, “Spectral density enhancement usingcoherent WDM,” IEEE Photon. Technol. Lett., vol. 17, pp. 504–506,2005.

[11] A. Sano, E. Yoshida, H. Masuda, T. Kobayashi, E. Yamada, Y.Miyamoto, F. Inuzuka, Y. Hibino, Y. Takatori, K. Hagimoto, T.Yamada, and Y. Sakamaki, “30 � 100-Gb/s all-optical OFDM trans-mission over 1300 km SMF with 10 ROADM nodes,” presented atthe Eur. Conf. Opt. Commun., Berlin, Germany, 2007, Paper PD 1.7,unpublished.

[12] T. Kobayashi, A. Sano, E. Yamada, Y. Miyamoto, H. Takara, and A.Takada, “Electro-optically subcarrier multiplexed 110 Gb/s OFDMsignal transmission over 80 km SMF without dispersion compensa-tion,” Electron. Lett., vol. 44, pp. 225–226, 2008.

[13] Y. Tang, W. Shieh, X. Yi, and R. Evans, “Optimum design forRF-to-optical up-converter in coherent optical OFDM systems,” IEEEPhoton. Technol. Lett., vol. 19, pp. 483–485, 2007.

[14] B. R. Washburn, S. A. Diddams, N. R. Newbury, J. W. Nicholson, M.F. Yan, and C. G. Jorgensen, “Phaselocked, erbium-fiber-laser-basedfrequency comb in the near infrared,” Opt. Lett., vol. 29, pp. 250–252,2004.

[15] Y. Wang, Z. Pan, C. Yu, T. Luo, A. Sahin, and A. E. Willner, “A multi-wavelength optical source based on supercontinuum generation usingphase and intensity modulation at the line-spacing rate,” presented atthe Eur. Conf. Opt. Commun., Rimini, Italy, 2003, Paper Th3.2.4.

[16] W. Shieh, “Coherent optical MIMO-OFDM for optical fibre commu-nication systems,” in Proc. Workshop 5, Eur. Conf. Opt. Commun.,Berlin, Germany, 2007.

[17] S. L. Jansen, I. Morita, and H. Tanaka, “16 � 52.5-Gb/s, 50-GHzspaced, POLMUX-CO-OFDM transmission over 4,160 km of SSMFenabled by MIMO processing KDDI R&D Laboratories,” presented atthe Eur. Conf. Opt. Commun., Berlin, Germany, 2007, Paper PD1.3.

[18] J. Armstrong, “Analysis of new and existing methods of reducing in-tercarrier interference due to carrier frequency offset in OFDM,” IEEETrans. Commun., vol. 47, no. 3, pp. 365–369, Mar. 1999.

[19] X. Yi, W. Shieh, and Y. Ma, “Phase noise effects on high spectral effi-ciency coherent optical OFDM transmission,” J. Lightw. Technol., vol.26, pp. 1309–1316, 2008.

Qi Yang was born in Wuhan, China, in 1981. He received the B.E. and M.E.degrees in electronic and information engineering from Huazhong Universityof Science and Technology, Wuhan, China. He is currently working toward thePh.D. degree in electrical and electronic engineering at the University of Mel-bourne, Melbourne, Australia.

In 2006, he was with Huagong Genuine Optics Technology Company, Ltd. Hewas involved in optical CDMA module and system design. His main researchinterests are optical OFDM, coherent detection, and high-speed system design.

Yan Tang was born in Jiangsu, China. She received the B.S. degree in electricaland electronic engineering from Southeast University, Jiangsu, China, in 2000and the M.S. degree in telecommunication engineering from the University ofMelbourne, Melbourne, Australia, in 2005, where she is currently working to-ward the Ph.D. degree in electrical and electronic engineering.

Her research interests are signal processing and coherent optical communi-cations with a focus on the optimum design for coherent optical OFDM trans-mitters and receivers.

Yiran Ma received the B.Eng. degree in telecommunication engineering fromthe Harbin Institute of Technology, Harbin, China, in 2005 and the M.S. de-gree in telecommunication engineering from the University of Melbourne, Mel-bourne, Australia, where he is currently working toward the Ph.D. degree.

In 2005, he was an intern for six months at the China Academy of Space Tech-nology, where he developed algorithms of real-time encryption and decryptionfor communication networks. His current research interests include coherentoptical system design, signal processing of MIMO-OFDM, and nonlinearity inoptical long-haul systems.

William Shieh (S’96-M’96) received the M.S. degree in electrical engi-neering-communications and the Ph.D. degree in physics from the Universityof Southern California, Los Angles, in 1994 and 1996, respectively.

From 1996 to 1998, he worked as a member of Technical Staff with the JetPropulsion Laboratory, Pasadena, CA. From 1998 to 2000, he was a Memberof the Technical Staff with Bell Labs, Lucent Technologies, Holmdel, NJ.From 2000 to 2003, he was a Technical Manager with Dorsal Networks,Columbia, MD. Since 2004, he has been with the Department of Electricaland Electronic Engineering, University of Melbourne, Melbourne, Australia.His current research interests include OFDM techniques in both wireless andoptical communications, coherent optical communication systems, and opticalpacket switching. He has published more than 100 journal and conferencepapers and has submitted 14 U.S. patents (nine issued) covering areas ofpolarization controller, wavelength stabilization in WDM systems, and Ramanamplifier-based systems and subsystems.

Experimental demonstration and numerical simulations of 16×10 Gb/s DWDM repeaterless transmission...

Documents

Transcript of Experimental demonstration and numerical simulations of 16×10 Gb/s DWDM repeaterless transmission...