Reception of Dual-Spatial-Mode CO-OFDM Signal Over a Two-Mode Fiber

634 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 30, NO. 4, FEBRUARY 15, 2012

Reception of Dual-Spatial-Mode CO-OFDMSignal Over a Two-Mode Fiber

An Li, Abdullah Al Amin, Member, IEEE, Xi Chen, Simin Chen, Guanjun Gao, andWilliam Shieh, Senior Member, IEEE, Fellow, OSA

Abstract—We demonstrate the use of mechanical grating basedmode converters to achieve two forms of dual-spatial-mode trans-mission: and , and dual modes. High modalextinction ratio ( dB) is shown for the mode converter withina 10-nm wavelength range. We first present 107-Gb/s coherentoptical OFDM (CO-OFDM) transmission over a 4.5-km two-modefiber using and modes where the mode separation isperformed optically. We then show 58.8-Gb/s CO-OFDM trans-mission using dual modes where the mode separation isachieved via 4 4 electronic MIMO processing.

Index Terms—Coherent communications, few-mode fiber, fiberoptics components, mode converter, OFDM.

I. INTRODUCTION

T HE emerging bandwidth-rich internet applications havelead to relentless demand for high-speed optical trans-

port. Recently, there has been increasing interest looking at few-mode fiber (FMF) as the next generation transmission mediafor achieving the capacity beyond that of the conventional stan-dard single-mode-fiber (SSMF) [1]–[5]. One of the advantagesof the FMF, e.g., two-mode fiber (TMF) is that it enables dou-bling or tripling of the fiber capacity without a need for expo-nential increase of SNR compared to SSMF fiber, throughmulti-plexing of the spatial modes [2]–[5]. By adding another degreeof freedom—here the spatial modes with respect to the wave-length, polarization and higher-order modulation, the informa-tion data coding could also be more efficient [6]. Comparedwith the previous approaches based on conventional multimodefibers (MMF) [7]–[10], FMF has the advantage of better modeselectivity and easier management of the mode impairments.Recently, a number of groups have demonstrated mode-divisionmultiplexing (MDM) of and modes [2], [3], two de-generate modes [4], and even all three

Manuscript received July 15, 2011; revised November 04, 2011; acceptedNovember 28, 2011. Date of publication December 08, 2011; date of currentversion February 01, 2012.A. Li and W. Shieh are with the Centre for Energy-Efficient Telecommunica-

tions (CEET), Department of Electrical and Electronics Engineering, Universityof Melbourne, Melbourne, VIC 3010, Australia (e-mail: [email protected]).A. A. Amin, X. Chen, and S. Chen are with the Department of Electrical

and Electronics Engineering, University of Melbourne, Melbourne, VIC 3010,Australia.G. Gao is with the Department of Electrical and Electronics Engineering,

University of Melbourne, Melbourne, VIC 3010, Australia, and also with theState Laboratory of Information Photonics and Optical Communications, Bei-jing University of Posts and Telecommunications, Beijing, China.Color versions of one or more of the figures in this paper are available online

at http://ieeexplore.ieee.org.Digital Object Identifier 10.1109/JLT.2011.2178817

modes [5] over FMF or TMF fibers.The advance into few-mode transmission requires brand newresearch in a wide range of topics from device to system level,including TMF design, TMF compatible component design andTMF transmission, among others. In mode-multiplexed trans-mission systems, mode selective components play a critical role.All the mode selective devices proposed in [2]–[5] fall into twomajor categories: free-space based and fiber based. The formeruses phase masks that are either based on liquid-crystal-on-sil-icon (LCoS) spatial light modulator (SLM) [4] or specially-fabricated glass plate [5]. However, free-space components areoften bulky. On the contrary, fiber based mode selective devicehas the advantage of compactness and easiness of integration.The coupler proposed in [11] could be a promising solution infuture MDM systems even though fabrication of this coupler in-volves sophisticated fiber etching, fusion and tapering process.A simple and tunable mechanical pressure-induced long periodfiber grating (LPFG) was proved to be an efficient mode con-verter (MC) [12], and its application has been demonstrated inboth 107-Gb/s mode-multiplexed OFDM transmission system[2] and 2 10 Gb/s NRZ MDM system [3]. In this paper, weextend our previous work in [2] to dual mode

transmission using the same all-fiber basedMC.We firstpresent 107-Gb/s coherent optical OFDM (CO-OFDM) trans-mission over a 4.5-km two-mode fiber using andmodes where the mode separation is performed optically. Wethen show 58.8-Gb/s CO-OFDM transmission using dualmodes where the mode separation is achieved via 4 4 elec-tronic MIMO processing.

II. MODE SELECTIVE COMPONENT DESIGN

A. LPFG Based Mode Converter (MC)

The main purpose of the MC is to convert optical signalsfrom to mode, or vice versa. There are variousmethods that can realize mode conversion suchas microbending [12], periodic mechanical pressure [13], andrefractive index modulation induced by lasers [14] on a FMFor TMF fiber. The resonant coupling happens when the gratingpitch equals to the beating length ,where and are the propagation constants of theand modes [15]. Fig. 1 shows the physical design of theMC. The TMF we use is a 4.5-km Ge-doped step-index fiberwith a core diameter of 11.9 m and nominal refractive indexstep of mode cutoff wavelength is2323 nm and loss of 0.26 dB/km. The measured modal groupdelay is 3.0 ns/km and mode beat length is around 524 m.The large differential-mode-delay (DMD) results in very small

0733-8724/$26.00 © 2011 IEEE

LI et al.: RECEPTION OF DUAL-SPATIAL-MODE CO-OFDM SIGNAL OVER A TWO-MODE FIBER 635

Fig. 1. Schematic diagram of a LPFG based mode converter. Thegroove pitch and pressure can be adjusted for optimization for certain wave-length or conversion ratio. The deformation of fiber core is assumed to be s-bendarc shape with radius r.

Fig. 2. Normalized power of mode converted from mode versuseffective coupling length for a LPFG-based mode converter under the gratingpitch m, at wavelength of nm. The three curves corre-spond to the core deformation radius of , 0.1, and 0.2 m, respectively.

modal mixing in our TMF due to large mismatch of modaleffective indices [16]. To simplify the analysis here we onlyconsider the deformation effect formed in a mode converter.The simulation is based on beam propagation method [17] andthe core deformation is assumed to be an s-bend arc shapeas shown in Fig. 1. The coupling efficiency depends on thecoupling length with various core deformation defined by theradius of the s-bend arc ( , 0.1 and 0.2 m) as is shownin Fig. 2. We find out that the optimum coupling length isinversely proportional to the core deformation radius r, and fora 0.2 m deformation the optimum coupling length is 8.1 mm,which assures that our MC can be very compact. Fig. 3 showsthe wavelength dependence of coupling efficiency at a couplinglength of 8.1 mm ( ridges) and core deformation of 0.2m. Fig. 4 shows the extinction ration (ER) versus wavelengthunder the same coupling length and core deformation. Theextinction ratio is defined as the power ratio betweenmode and mode after mode conversion. It can be seenthat theoretically the ER can be very high around 1550 nmand maintain above 20 dB for more than 10-nm wavelengthrange. In light of the simulation result, we fabricated 4 metalgratings with 20 evenly-spaced grooves on a polished surface.The groove pitch is controlled to be m. All fourMCs are made as follows: first we place a 0.9-mm jacketedTMF onto an aluminum slab with tape. The TMF and gratingare then mounted between a 3–axis stage and L-shape steel. Theangle between grating/fiber and applied pressure can be con-trolled by the stage and the position of fiber. The angle betweengrating/fiber determines the effective pitch

Fig. 3. Normalized power of mode versus wavelength for a LPFGbased -mode converter with core deformation m. Sim.: Simulation;Exp.: Experiment.

Fig. 4. Extinction ratio versus wavelength for a LPFG based -mode converterwith, core deformation m. Sim.: Simulation; Exp.: Experiment.

where is the original pitch size of the grating and is theangle. After appropriate adjustment of the stage and position ofthe fiber, maximum coupling ratio and ER for all MCs occur ataround 1550 nm. The measured and simulated ERs are shownin Fig. 4, and it is confirmed that ER can be maintained beyond20 dB for a 13-nm wavelength range. The best ERs of 26.8,22.8, 24.6, and 24.3 dB for MCs1-4 occur at 1551 nm.

B. Free Space Mode Combiner/Splitter

The mode combiner/splitter is comprised of two 2-axis preci-sion stages, one beamsplitter (BS) and three collimating lensesas shown in Fig. 5. The signal is polarization multiplexed andmode converted before entering into the mode combiner. Thetwo input TMFs that carry either or modes gen-erated by MCs are connected with the two input ports of themode combiner, whose position can be manually aligned bythe precision stages. Output port of the mode combiner is fixedusing a fiber collimator and connected with the 4.5-km trans-mission fiber. The input signal is first collimated to a spot sizeof diameter of 2 mm by one of the movable collimating lenswith numerical aperture and effective focal length

mm. The collimated beam is subsequently passingthrough the BS in either transmission or reflection direction, andfinally focused onto the core of the output fiber by another lensinside the packaged collimator. The BS is polarization insensi-tive with less than 5% difference in transmission for s- and p-po-larization at 1550 nm. The input and output of TMFs are connec-torized before mounted onto the stage with a FC-type adapter.


Fig. 5. Schematic diagram of a free-space mode combiner. BS: beamsplitter.The precision stages have freedom of two-axes (X and Y. Z is the light propa-gation axis). The beams are collimated before entering the BS to minimize thedivergence and distortion.

The connectors are specially designed with an adjustable key sothat the fiber can be axially rotated. By adjusting the key of theconnectors, the orientation of the two modes can be ma-nipulated to be orthogonal (90 ) to each other. An IR camera isplaced in the unused path of the BS to monitor the orientationand orthogonality of the two modes, as shown in Fig. 5.The loss in the reflection/transmission path of the BS is esti-mated to be about 3.5/4.5 dB, and the loss of the focusing systemdue to misalignment and Fresnel reflection is around 1 dB. The1-dB power difference between the two paths is balanced usingSMF attenuator before mode conversion. It is also possible toupgrade to a 3 1 mode combiner by introducing another set ofprecision stage and collimating lens together with one more BS,which we will apply in the future for triple-mode transmissionexperiment . The mode splitter has thesame structure as the combiner except that it is operated in theopposite direction.

III. MODE TRANSMISSION SETUP AND RESULTS

The 107-Gb/s dual mode coherent OFDMtransmission setup [2] is shown in Fig. 6. Four transmitterscorresponding to the two modes ( and , each withtwo polarizations x and y) are implemented as follows: first, theOFDM signal is generated off-line with MATLAB program.The FFT size is 64, with middle 40 subcarriers filled. Cyclicprefix (CP) is set to be 1/8 of the observation window. TheOFDM waveform consists of 500 symbols, while initial 20symbols with alternative polarization launch are used as trainingsymbol (TS) for channel estimation. After IFFT operation, thereal and imaginary components of the time-domain signalare uploaded onto a Tektronix Arbitrary Waveform Generator(AWG). Then we generate three tones spaced at 6.563 GHzfrom an external cavity laser (ECL) operating at 1549.3 nmwavelength and an intensity modulator (IM) driven by a syn-thesizer. The optical OFDM signal is then modulated on eachtone by using the AWG to drive a nested Mach–Zehnder mod-ulator. The orthogonally multiplexed 3-band OFDM signal isthen divided and recombined on orthogonal polarizations with

one symbol delay to emulate polarization multiplexing. Thesampling rate of AWG is set at 10 GSa/s, and consequently theOFDM symbol length is 7.2 ns. The raw data rate is 150 Gb/sand net data rate after overheads (7% FEC, 4% TS, 12.5% CPand 5 discarded subcarriers around DC) is 107 Gb/s. We launcha constant optical power of 5.5 dBm into the SMF before MC1for TMF transmission. A mode stripper (MS) is used aftercenter splicing to strip off unwanted mode, ensuring pure

mode launch into TMF. The MS is realized by tightlybending the bare TMF fiber over 7-mm posts of about 20 roundsthat provides a rejection ratio of dB against modeas shown in [2]. The first MC has a nominal 50:50 conversionratio on modes, with both modes carrying theabove described polarization multiplexed signal. It is estimatedthat after 4.5-km TMF fiber, modal delay betweenmodes is 13.5 ns. This amount of modal delay is nearly twotimes of the OFDM symbol length. Therefore, the two modesare completely de-correlated after 4.5 km TMF transmission,validating the independent reception of the signal onor mode. At the receiver, the two modes are detectedindividually as follows: (1) For mode, a mode stripper(MS2) is used to strip off mode, with the remainingmode coupled into the SMF fiber. The signal is subsequentlydetected by coherent optical receiver. (2) For mode, asecond mode converter (MC2) with nominal 100% conversionratio is used to convert the mode into . Any residual

mode that is not converted to and mode thatis converted to is stripped off by a subsequent MS. Theconverted mode (now ) is coupled into the SMF,subsequently fed into the coherent optical receiver.The received signal is converted from optical to electrical do-

main and sampled by a 50 GSa/s digital oscilloscope. The re-ceived signal is processed using 2 2 MIMO-CO-OFDM pro-gram. The digital signal processing (DSP) consists of five steps:1) timing synchronization; 2) frequency offset compensation;3) channel estimation and phase estimation; 4) subcarrier map-ping and constellation recovery, and 5) BER andQ computation.With this setup, the received power of mode is dBmand of mode is dBm, and therefore the end-to-endlosses are 6 and 10.8 dB for and modes, respectively.The higher loss for the mode can be attributed to the exces-sive loss ofMSs (0.2–0.4 dB), and loss ofMCs ( dB,

dB) combined with the fact that there exists randomcoupling between the degenerate modes inside the 4.5-kmTMF fiber, whereas only one of the two orientations can belaunched or detected due to the asymmetry of our MC. Fig. 7(a)and (b) show respectively the high-resolution (0.01 nm) op-tical spectra of received signal for and modes. Goodconstellations in all 3 bands and both modes are confirmed, asshown in Fig. 7(c) and (d). We could not measure any errorsout of 100 590 bits for any given combination out of 12 combi-nations of signal states (3 bands, 2 polarizations and 2 modes).Overall Q factor for all bands, polarizations and modes are sum-marized in the Table I. The relative difference of 2 dB in av-erage Q for the and modes can be attributed to theincreased crosstalk in case, as the ER of the MC becomesnon-ideal under random perturbation between two spatialmodes. The results demonstrate the feasibility of the use of TMF


Fig. 6. Experimental setup for 107-Gb/s dual-mode dual polarization transmission over 4.5-km TMF fiber. ‘X’ indicates controlled coupling between modesof SMF and TMF by center splicing.

Fig. 7. (a) Optical spectrum and (c) constellations of LP mode. (b) Opticalspectrum and (d) constellations for LP mode after 4.5-km transmission inthree bands.

in dual-mode and dual-polarization transmission to increase thefiber capacity and/or SE. In this experiment, we assume neg-ligible mode coupling between modes due to thehigh modal mismatch and relatively short distance, which hasbeen experimentally proved in [16]. The modes are detected in-dividually without the need of MIMO processing which has ad-vantage of reduced complexity. The major limiting factors fora TMF-based transmission system include (i) the ER of MCsand rejection ratio of MSs; (ii) the polarization- and mode- de-pendence of the MCs and free-space components, which intro-duce mode dependant loss (MDL), polarization dependant loss(PDL) and crosstalk (discrepancy in ER); (iii) the intrinsic lossof the fiber since a TMF amplifier is not readily available. Forlong-haul transmission, MIMO processing is necessary as therewill be significant accumulation of mode coupling. In the fol-lowing section, we demonstrate dual mode transmissionby using MIMO DSP.

TABLE IMEASURED Q FACTOR (IN DB) FOR A 107-GB/S, 3-BAND CO-OFDMDUAL-MODE, DUAL-POLARIZATION TRANSMISSION ON 4.5-KM TMF

IV. DUAL MODE TRANSMISSION SETUP AND RESULT

In Section III, to detect , manual spatial control isused. Similar to polarization mode division multiplexing, thepreferred method of mode detection is to employ spatialmode diversity. Fig. 8 shows the experimental setup of the4 4 coherent MIMO-OFDM system [2] based on TMFswhere spatial diversity is employed at both transmit andreceive. Four transmitters are emulated by polarization andmode multiplexing as follows: first, the transmitted signal isgenerated off-line with MATLAB program. The total numberof OFDM subcarriers is 128, and cyclic prefix (CP) is set to1/16 of the observation window. Middle 68 subcarriers out of128 are filled excluding center 3 subcarriers that are nullifiedfor RF-pilot [18]. 50 OFDM symbols are sent for evaluation,out of which 4 symbols are used for channel estimation. Thedigital time-domain signal is formed after IFFT operation anduploaded to the AWG. The operating wavelength of ECL is1550.92 nm. The OFDM symbol length is 13.6 ns in this case.The OFDM signal is then divided and recombined on orthog-onal polarizations using polarization beam combiner (PBC)and polarization maintaining fiber (PMF) with one symbol


Fig. 8. Experimental setup of a coherent 4 4-MIMO system over a two-mode fiber (TMF). PBC/PBS: polarization beam combiner/splitter; MC: mode converter;MS: mode stripper; WSS: wavelength selective switch, emulated by a Finisar waveshaper.

delay to emulate polarization division multiplexing (PDM).After the PDM emulation, the signal is again split with a 3-dBSMF coupler and one of the two branches is subsequentlydelayed by two OFDM symbol length (27.2 ns), enabling themode division multiplexing (MDM) emulation. Out of the 4training symbols, only one symbol is filled, with all other 3symbols remain unfilled. After 1 symbol delay for PDM and2 symbol delay for MDM emulation, TS in all 4 tributariesbecome orthogonal. These two delayed MDM signals arethen converted from fundamental mode of SMF tothe and modes of the TMF by utilizing the MCand mode combiner described in Section II. After TMF trans-mission the free-space mode splitter decomposes the signalinto two orthogonal modes, each converted back to thefundamental mode in SMFs. Any residual componentscan be filtered by mode converting to and then strippedoff with the subsequent MSs. In this experiment the MS arerealized by tightly winding the 0.9-mm jacketed TMF fiberover 8-mm posts of more than 10 rounds. After this the twoPDM tributaries are further split to orthogonal polarizations byPBS and mixed with 4 LO branches using 3-dB couplers into4 photodetectors (PDs) with trans-impedance amplifiers. Atreceiver, we choose heterodyne coherent detection so that only4 PDs and one digital oscilloscope with 4 channels of ADC aresufficient to realize the 4 4 MIMO reception.The optical power in the system is monitored at four nodes

(A-D) as depicted in Fig. 8. The end-to-end loss is estimated forall the possible TX/RX port combinations, as shown in Table II.The relatively high power loss in our system is mainly attributedto the loss of the beamsplitter (3.5–4.5 dB), misalignment/re-turn loss ( dB) and mode converter (1.4–2.4 dB). Neverthe-less, the received power is well above the sensitivity of PDs.Because of the relatively low differential mode delay (DMD)between the two LP modes, the modal coupling happens evenfor a short TMF fiber of 4.5 km due to random perturbation.

The 4 4 MIMO-CO-OFDM could evaluate the channel ma-trix in the presence of polarization as well as mode coupling.One of the estimated 4 4 channel matrices for the twomodes after 4.5-km TMF transmission is illustrated in Fig. 9.A smooth channel response for a narrow bandwidth is con-firmed which shows slow frequency selectivity. Using polyno-mial curve fit of the phase angle of the diagonal components( and ), the second order coefficient is cal-culated to be 2 10 which corresponds to the accumulatedchromatic dispersion (CD) of 28.9 ps/nm/km. The successful re-ception of all data symbols also confirms the slow time varyingof channel within the OFDM symbol length scale. The receivedspectrum of the signal is shown in Fig. 10. A 5.3-GHz band-width is occupied by the OFDM signal including the RF pilotregion. Because of heterodyning detection, signal-to-signal in-termixing products is present around LO frequency. To avoidthis penalty in the low frequency region, 5-GHz guard bandis kept. The net data rate for all polarizations and modes is58.8 Gb/s (128-point FFT, 68 subcarriers filled, 8 QAM) takinginto consideration CP and TS overheads. The measured BERversus OSNR for the signal transmission is shown in Fig. 11.The required OSNR for BER of is found to be 17.9 dBon average for b2b with less than 1-dB discrepancy betweenthe best/worst tributaries, and 18.8 dB after 4.5-km transmis-sion similarly with less than 1-dB variation between tributaries.The discrepancy in polarizations may attribute to the PDL ofthe free-space system as well as possible slightly launch powerimbalance in the two polarizations. The discrepancy in modescould be due to the power imbalance and misalignment. Theclean constellation in the inset confirms the effectiveness of ourMDM CO-OFDM system. The major limiting factors of thetransmission distance of our 4 4 mode multiplexed systemare: (i) the crosstalk from residual mode (the ER of MCsand rejection ratio of MSs); (ii) the orthogonality and rejec-tion ratio of the two modes in mode combiner/splitters;


Fig. 9. Measured channel matrix parameters for the two modes after4.5-km TMF transmission. (a) Real and imaginary components and (b) phaseof the 4 4 fiber channel matrix. The subcarrier indices are mapped from low tohigh frequency. The subcarrier index difference of 1 corresponds to frequencyspacing of 78 MHz.

(iii) PDL, MDL and the end-to-end loss; Nevertheless, for a4.5-km fiber we have sufficient margin as can be seen fromFig. 11(a). It has been demonstrated in our recent experiment[19] that a 35.3 Gb/s Dual- mode 4 4 MIMO-OFDMtransmission was successfully achieved over 26-km TMF usingthe same setup. Compared with the 107-Gb/s transmission resultusing mode in Section III, in this dual modetransmission we implement 4 4 MIMO processing to over-come modal coupling. Because heterodyning technique is used

Fig. 10. Received RF spectrum of one of the four tributaries after 4.5-km TMFtransmission. The 5-GHz guard band is intentionally made to avoid intermixingproduct of signal in heterodyning.

Fig. 11. (a) Measured OSNR versus BER performance for 8-QAM 4 4MIMO-OFDM system for back-to-back (B2B) and 4.5-km TMF transmissionconfiguration. (b) Constellation of received signal after 4.5-km TMF transmis-sion with dB.

TABLE IIEND-TO-END LOSS OF THE EXPERIMENTAL SETUP

to save the need for another real-time oscilloscope, a narrow-band optical filter is required for multi-band OFDM signal de-tection to eliminate the image noise. Due to the unavailability ofsuch a filter we only show single-band transmission (5.3 GHz)in this experiment, but in principle multiband OFDM transmis-sion can be achieved either using a tunable optical filter in het-erodyne detection, or using intradyne detection setup with twosynchronized real-time oscilloscopes.

V. CONCLUSION

We have demonstrated two forms of dual-spatial-modetransmission: LP and LP , and dual LP modes. Highmodal extinction ratio ( dB) is shown for the modeconverters within a 13-nm wavelength range. We first present


107-Gb/s coherent optical OFDM (CO-OFDM) transmissionover a 4.5-km two-mode fiber using LP and LP modeswhere the mode separation is performed optically. We thenshow 58.8-Gb CO-OFDM transmission using dual LP modeswhere the mode separation is achieved via 4 4 electronicMIMO processing.

REFERENCES

[1] W. Shieh, “High spectral efficiency coherent optical OFDM for 1 Tb/sethernet transport,” in Proc. OFC, 2009, paper OWW1.

[2] A. Li, A. A. Amin, X. Chen, and W. Shieh, “Transmission of 107-Gb/smode and polarization multiplexed CO-OFDM signal over a two-modefiber,” Opt. Exp., vol. 19, pp. 8808–8814, 2011.

[3] N. Hanzawa, K. Saitoh, T. Sakamoto, T. Matsui, S. Tomita, and M.Koshiba, “Demonstration of mode-division multiplexing transmissionover 10 km two-mode fiber with mode coupler,” in Proc. OFC 2011,paper OWA4.

[4] M. Salsi, C. Koebele, D. Sperti, P. Tran, P. Brindel, H. Mardoyan, S.Bigo, A. Boutin, F. Verluise, P. Sillard, M. Bigot-Astruc, L. Provost, F.Cerou, and G. Charlet, “Transmission at 2 100 Gb/s, over two modesof 40 km-long prototype few-mode fiber, using LCOS basedmodemul-tiplexer and demultiplexer,” in Proc. OFC 2011, paper PDPB9.

[5] R. Ryf, S. Randel, A. Gnauck, C. Bolle, R. Essiambre, P. Winzer, D.Peckham, A. McCurdy, and R. Lingle, “Space-division multiplexingover 10 km of three-mode fiber using coherent 6 6 MIMO pro-cessing,” in Proc. OFC 2011, paper PDPB10.

[6] I. B. Djordjevic, M. Arabaci, L. Xu, and T. Wang, “Spatial-domain-based multidimensional modulation for multi-Tb/s serial optical trans-mission,” Opt. Exp., vol. 19, pp. 6845–6857, 2011.

[7] S. Berdague and P. Facq, “Mode division multiplexing in opticalfibers,” Applied Optics, vol. 21, pp. 1950–1955, 1982.

[8] S. H. Murshid, A. Chakravarty, and R. Biswas, “Simultaneous trans-mission of two channels operating at the same wavelength in standardmultimode fibers,” in Proc. CLEO/QELS 2008, paper JWA107.

[9] B. C. Thomsen, “MIMO enabled 40 Gb/s transmission using modedivision multiplexing in multimode fiber,” in Proc. OFC2010, paperOThM6.

[10] B. Franz, D. Suikat, R. Dischler, F. Buchali, and H. Buelow, “Highspeed OFDM data transmission over 5 km GI-multimode fiber usingspatial multiplexing with 2 4 MIMO processing,” in Proc. ECOC2010, paper Tu3.C.4.

[11] K. Y. Song, I. K. Hwang, S. H. Yun, and B. Y. Kim, “High performancefused-type mode-selective coupler using elliptical core two-mode fiberat 1550 nm,” IEEE Photon. Technol. Lett., vol. 14, no. 4, pp. 501–503,Apr. 2002.

[12] I. Hwang, S. Yun, and B. Kim, “Long-period fiber gratings based onperiodic microbends,” Opt. Lett., vol. 24, pp. 1263–1265, 1999.

[13] S. Savin et al., “Tunable mechanically induced long-period fiber grat-ings,” Opt. Lett., vol. 25, pp. 710–712, 2000.

[14] D. D. Davis, T. K. Gaylord, E. N. Glytsis, S. G. Kosinski, S. C. Met-tler, and A. M. Vengsarkar, “Long-period fibre grating fabrication withfocused CO laser pulses,” Electron. Lett., vol. 34, pp. 302–303, 1998.

[15] A. Yariv, “Coupled-mode theory for guided-wave optics,” IEEE J.Quantum Electron, vol. QE-9, no. 9, pp. 919–933, Sep. 1973.

[16] F. Yaman, N. Bai, Y. Huang, M. Huang, B. Zhu, T. Wang, and G. Li,“10 112 Gb/s PDM-QPSK transmission over 5032 km in few-modefibers,” Opt. Exp., vol. 18, pp. 21342–21349, 2010.

[17] J. VanRoey, J. van derDonk, and P. Lagasse, “Beam-propagationmethod: Analysis and assessment,” J. Opt. Soc. Amer., vol. 71, pp.803–810, 1981.

[18] S. Jansen, I. Morita, N. Takeda, and H. Tanaka, “20-Gb/s OFDM trans-mission over 4 160-km SSMF enabled by RF-Pilot tone phase noisecompensation,” in Proc. OFC 2007, paper PDP15.

[19] A. A. Amin, A. Li, S. Chen, X. Chen, G. Gao, and W. Shieh, “Dual-LP11 mode 4 4MIMO-OFDM transmission over a two-mode fiber,”Opt. Exp., vol. 19, pp. 16672–16679, 2011.

An Li was born in Wuhan, China. He received the B.E. degree in optical infor-mation science and technology and M. E. degree in physical electronics fromHuazhong University of Science and Technology, Wuhan, China, in 2003 and2006, respectively. He is currently working toward the Ph.D. degree at the De-partment of Electrical and Electronic Engineering, University of Melbourne,Melbourne, Australia.From 2006 to 2007, he was with Fiberxon Inc. (Source Photonics) as a Hard-

ware Engineer focusing on the R&D of G(E)PON optical transceivers. His cur-rent research interests include mode division multiplexing, few mode fiber-based transmission systems, wavelet analysis, and energy-efficient transmissionsystems.

Abdullah Al Amin (S’01–M’05) was born in Dhaka, Bangladesh, in 1974. Hereceived the B.E., M.E., and Ph.D. degrees in electronic engineering from theUniversity of Tokyo, Tokyo, Japan, in 1999, 2001, and 2005, respectively.From 2005 to 2007, he was a Postdoctoral Research Fellow with the Research

Center for Advanced Science and Technology, University of Tokyo, where hewas responsible for developing an “optical-router” prototype as part of NEDO-funded Photonic Network Project. From 2007 to 2009, he has been with KDDIR&D Laboratories, Inc., Saitama, Japan, conducting research on high-speed op-tical transmission using advanced modulation formats. He joined the Universityof Melbourne, Melbourne, Australia, as a Research Fellow in 2010, and his cur-rent research focus is on coherent optical OFDM and few-mode fiber basedtransmission systems.

Xi Chen was born in Hunan, China. She received the B.E. degree in com-munication engineering from National University of Defense Technology,Changsha, China, in 2008. She is currently working toward the Ph.D. degreein electrical and electronic engineering at the University of Melbourne,Melbourne, Australia.Her current research interests include optical OFDM systems, fiber nonlinear

effects, and few-mode fiber (FMF)-based communication systems.

SiminChenwas born in Fujian, China. He received the B.S. andM.S. degrees inelectrical and electronic engineering from Sichuan University, Chengdu, China.He is currently working toward the Ph.D. degree at the Department of Electricaland Electronic Engineering, University of Melbourne, Melbourne, Australia.From 2006 to 2007, he was with Fiberxon Inc. (acquired by MRV in 2007),

working on optical transceivers and FPGAs. His research interests are signalprocessing and coherent optical communication systems with a focus on thereal-time implementation of coherent optical OFDM transmitter and receiver.

Guanjun Gao received the B.E. degree in communication engineering fromBeijing University of Posts and Telecommunications (BUPT), Beijing, China,in 2006, where he is currently working toward the Ph.D. degree.He is currently a Visiting Student with the University of Melbourne, Mel-

bourne, Australia. His current research interests include optical OFDM systems,few-mode fiber (FMF)-based communication systems, and cross-layer design ofoptical networks.

William Shieh (S’96–M’96) received the M.S. degree in electrical engineeringand Ph.D. degree in physics from the University of Southern California,Los Angeles, in 1994 and 1996, respectively.Since 2004, he has been with the Department of Electrical and Electronic En-

gineering, University of Melbourne, Melbourne, Australia. His current researchinterests include OFDM techniques in both wireless and optical communica-tions, coherent optical communication systems, and optical packet switching.He has published more than 110 journal and conference papers, and submitted14 U.S. patents (nine issued) covering areas of polarization controller, wave-length stabilization in WDM systems, and Raman amplifier-based systems andsubsystems.Dr. Shieh is a Fellow of the Optical Society of America.

Reception of Dual-Spatial-Mode CO-OFDM Signal Over a Two-Mode Fiber

Documents

Transcript of Reception of Dual-Spatial-Mode CO-OFDM Signal Over a Two-Mode Fiber