Coherent optical OFDM (CO-OFDM) combined with orthogonal band multiplexing provides a scalable and flexible solution for achieving ultra high-speed rate. Among many CO-OFDM implementations, digital Fourier transform spread (DFT-S) CO-OFDM is proposed to mitigate fiber nonlinearity in long-haul transmission. In this paper, we first illustrate the principle of DFT-S OFDM. We then experimentally evaluate the performance of coherent optical DFT-S OFDM in a band-multiplexed transmission system. Compared with conventional clipping methods, DFT-S OFDM can reduce the OFDM peak-to-average power ratio (PAPR) value without suffering from the interference of the neighboring bands. With the benefit of much reduced PAPR, we successfully demonstrate 1.45 Tb/s DFT-S OFDM over 480 km SSMF transmission.
©2012 Optical Society of America
Coherent optical OFDM (CO-OFDM) has recently received much attention as a candidate for long haul transmissions [1–3]. Significant progresses have been witnessed in this research area in which ultra high speed transmissions over thousand kilometers fiber have been demonstrated during the last a few years [4–9]. Although the maximum bandwidth supported by the current electrical components is only of the order of several tens of gigahertz, the technique of “orthogonal-band-multiplexing” (OBM), can overcome the electrical bandwidth limitation, and realize ultra high-speed transmission per superchannel (wavelength) . This concept was first demonstrated in a 107 Gb/s CO-OFDM long-haul transmission [5,10]. Different from the traditional wavelength-division multiplexing (WDM) technique, OBM can construct a “superchannel” using a multiple-carrier comb from a single wavelength in a spectral efficient manner. Each carrier is used to convey a portion of the transmitted signal. The entire signal spectrum is continuous without any frequency guard band in between. The influence of guard band for OBM-OFDM is investigated in , which shows that the orthogonality of optical carrier can guarantee zero linear cross-talk penalty between the adjacent sub-bands.
It is well known that CO-OFDM possesses many advantages, such as extreme robustness against fiber chromatic dispersion (CD) and polarization mode dispersion (PMD), low digital computation complexity, and high spectral efficiency. Besides its main drawback of digital-to-analog converter (DAC) requirement, large peak-to-average power ratio (PAPR) in OFDM may cause problems for DAC (limited resolution) and RF amplifiers, and nonlinearity in transmission. To reduce the PAPR, one of the most simple and effective techniques is “clipping” [12,13]. Clipping performs very well in single band transmissions, which can effectively reduce the OFDM PAPR. However it also introduces the out-of-band spectral leakage . This leakage is problematic for the OBM method, where multiple bands are placed side by side. The leakage spectral components impair the neighboring bands, and lead to additional penalties.
Another effective PAPR reduction technique is called “discrete Fourier transform spread OFDM” (DFT-S OFDM), which has been well studied in wireless communications, and has been adopted into the next generation 4G mobile standard, known as long-term-evolution (LTE) . In optical communications, DFT-S OFDM was studied to mitigate the fiber nonlinearity in . In this paper, we evaluate the performance of DFT-S OFDM in both single and multi-band transmission systems. When the application of DFT-S OFDM is used in multi-band fiber transmissions, the crosstalk between multiple bands can be significantly eliminated compared with conventional clipping schemes . Furthermore, we experimentally demonstrate superchannel Tb/s DFT-S CO-OFDM transmission over 400 km single mode standard fiber (SSMF). Our demonstration shows that DFT-S CO-OFDM may be a promising technique for the future superchannel transmission over long distance.
2. The principle of DFT-S OFDM
Fig. 1(a) shows the diagram for generating DFT-S OFDM signal at the transmitter. In the experiment, four bands are used to carry the payload data. For DFT-S OFDM systems, instead of directly applying the IDFT to convert payload from time-domain to frequency-domain, the payload will first go through DFT spreading. The whole payload is first partitioned into four sets. A 16-point FFT is performed on each set. The four sets are then mapped into the center of the transmitter bandwidth. Note that due to the imperfect of the bias in the optical I/Q modulator and other factors, the DC part of the signal may be un-recoverable, so that the middle subcarrier has to be unfilled. By padding zeros on the higher frequency part, a 128- point FFT will convert the frequency domain signal to time domain. The following procedures in the transmitter are the same as the traditional OFDM processing, such as cyclic prefix insertion. Fig. 1(b) shows the digital signal processing for the DFT-S OFDM. When the OFDM signal is synchronized, a 128-point FFT is used to convert the signal into frequency domain. Four band signals are separated and processed individually. In each band, a 16-point IFFT is used to re-convert the signal. The processing of channel and phase noise estimation are the same as traditional OFDM. The phase noise information can be further shared between the multiple bands to obtain the most likely estimated results. Finally, the constellation is reconstructed and decision is extracted.
An important issue on DFT-S OFDM is its computational complexity, due to additional IFFF/FFT in the transmitter and receiver. Here we use radix-2 Cooley–Tukey algorithm to calculate the computational complexity. According to the well-known radix-2 Cooley–Tukey algorithm, to compute the N-point FFT with only complex multiplies (ignoring number of additions for simplicity). For the 4 bands of 16-point FFTs, the overall number of complex multiplications is 4 × 16/2 × log2(16) = 128. Whereas, the whole band of 128-point IFFT in the transmitter is 128/2 × log2(128) = 448. Therefore, the additional computational complexity caused by DFT-Spread operation is ~28.6% more than the conventional OFDM case. Similar conclusion can be drawn for the receiver, but the ratio may slightly vary dependent on the complexity of other operations, such as channel and phase estimation.
3. Experimental setup for evaluating DFT-S OFDM in multi-band system
Fig. 2 shows the experimental setup for 150 Gb/s DFT-S CO-OFDM system. The transmitted signal for both DFT-S OFDM and conventional OFDM scheme is generated off-line by MATLAB program with a data sequence of 215-1 PRBS and mapped to 4-QAM constellation. An arbitrary waveform generator (AWG) is used to produce the I/Q RF signals at 10 GS/s, which are subsequently fed into I and Q ports of an optical IQ modulator, respectively. The OFDM baseband signal is constructed with 66 subcarriers. The FFT length is 128. The middle 2 subcarriers are unfilled, where the wavelength of local oscillator is located. In addition, other 4 subcarriers are used to estimate the phase noise. The other subcarriers on the two sides of the four payload sets are zeros padded. The individual band per tone is limited to be less than 6.6 GHz, because the RF synthesizer used to drive the second optical modulator has only 20 GHz tuning range. Consequently, only 4 sets of 16-point DFT-S bands are used, spaced at 5.15625 GHz. 1/8 of the symbol period is used for cyclic prefix to assist in mitigating channel dispersion. The subsequent polarization multiplexing doubles the line rate per optical tone to 16.67 Gb/s calculated as
The multi-tone generator includes two optical intensity modulators (IM): The first stage generates 3 optical tones by driving an optical intensity modulator with 5.15625 GHz RF tone. Then the optical tones are fed into another IM, which is driven at 15.46875 GHz. The inset in Fig. 2 shows the generated optical ones. All the clock resources are phase-locked to the AWG using a 10 MHz reference clock. The signal bandwidth for each sub-band is equal to the tone spacing. Thus, there is no frequency guard-band among all the 9 sub-bands. The optical OFDM signal is then fed into a polarization splitter, with one branch delayed by one OFDM symbol period (14.4 ns) to emulate the polarization-multiplexing, resulting in a total line rate of 150 Gb/s. To evaluate the influence of crosstalk between multiple bands, back-to-back measurement is carried out. At the receiver side, a local oscillator is fed into polarization diversity optical hybrid to mix with the signal. The signal is detected by four pairs of balanced detectors. The four RF signals for the two IQ components are then input into a Tektronix oscillator scope and are acquired at 50 GS/s and processed off-line with a MATLAB program using 2x2 MIMO-OFDM models.
3. Experimental setup for evaluating DFT-S OFDM in multi-band system
Fig. 3 shows the RF spectra after the single-band detection. It can be seen that the DFT-S and conventional OFDM without clipping shows very clean baseband. However, when clipping ratio is set to 1.5, the out-band aliasing frequency component is enhanced by a few dBs. When the signal is transmitted via multiple bands, such spectral leakage will influence the adjacent bands.
Fig. 4 shows the BER performance as a function of OSNR in (a) single-band and (b) multi-band configuration. For both configurations, the DFT-S OFDM shows the same performance as without clipping, but greatly reduces PAPR value from 18.6 to 8.1 dB.
Table 1 shows the required OSNRs at the BER of 1x10−3 for DFT-S and conventional OFDM with various clipping ratio (CR). It can be seen that the reduced clipping level causes power penalty. For instance, CR of 1.5 requires additional 1.92 dB of OSNR compared to that without clipping. For 150 Gb/s CO-OFDM using multi-band transmission, the required OSNR is 9.54 dB higher due to the data rate increase from 16.67 to 150 Gb/s. In the cases of both no clipping and DFT-S, there is ~0.4 dB additional penalty for the multi-band signal. The multi-band detection penalties for CR of 2.5, 2, and 1.5 are 1.12, 1.26, and 1.86 dB respectively. Taking into consideration of experimental implementing penalty, the induced penalties for the three cases are about 0.7, 0.8, and 1.5 dB. These additional penalties are caused by the multi-band crosstalk. In contrast, DFT-S OFDM performs very well in the multi-band configuration while maintain much reduced PAPR.
4. 1.45 Tb/s DFT-S CO-OFDM superchannel transmission
Fig. 5(a) shows the system configuration for 1.45 Tb/s DFT-S CO-OFDM transmission system. Based on the previous experimental setup, a new optical carrier source generator is introduced here to support Tb/s transmission. The multi-tone generator includes three stages. The first two optical phase modulators are driven by a strong RF tone at 15.46875 GHz with a power of ~1 W. Then the optical tones are fed into a Waveshaper for filtering and reshaping. Fig. 5(b) shows the generated 29 optical tones after two cascaded phase modulators. Because no bias tuning is required for phase modulator, such configuration can provide very a stable carrier source . Then the produced carriers are fed into another optical intensity modulator driven at 5.15625 GHz. Fig. 5(c) shows all the generated optical tones after three cascaded optical modulators. The SNR of each tone is greater than 20dB. All the clock sources are synchronized to the AWG using a 10 MHz reference clock. There is no frequency guard-band among the 87 sub-bands. The optical OFDM signal is then fed into a polarization splitter, with one branch delayed by one OFDM symbol period (14.4 ns) to emulate the polarization multiplexing, resulting in a total line rate of 1.45 Tb/s. The signal is transmitted into a re-circulation loop with a span of 80 km standard single mode fiber. At the receiver side, a local oscillator is fed into polarization diversity optical hybrid to mix with the signal. The signal is detected by four pairs of balanced detectors. The four RF signals are then input into a Tektronix oscilloscope and are sampled at 50 GS/s and processed off-line.
Fig. 6(a) shows the BER curve as a function of OSNR in back-to-back transmission. To achieve BER of 1x10−3, the required OSNR is 28.3 dB. In , the required OSNR for 1.08 Tb/s is 27 dB. Our experimental result is exactly 1.3 dB away from , which is caused by the net rate increase. This means the DFT-S CO-OFDM provides the same performance as the traditional CO-OFDM for Tb/s back-to-back transmission. Fig. 6(b) shows the BER performance of all the sub-bands after 480 km transmission. In the measurement, every 3 tones are tested. As shown in the Fig. 6(b), all the measured sub-bands are under 1x10−3 level with some margin to the 7% FEC threshold.
We have experimentally evaluated the DFT-S OFDM in multi-band systems. Compared with conventional OFDM at various clipping levels, DFT-S OFDM contributes minimum crosstalk to the adjacent sub-bands, while achieving much reduced OFDM PAPR value. Furthermore, we experimentally investigated the performance of DFT-S CO-OFDM superchannel transmission. The back-to-back results give the same performances as the traditional CO-OFDM. We also successfully demonstrated a transmission of 1.45 Tb/s DFT-S CO-OFDM over 480 km SSMF. The demonstration shows that DFT-S CO-OFDM may be a promising technique for the future superchannel transmission over long distance.
This work was supported by the National Basic Research (973) Program of China (2010CB328300).
References and links
1. W. Shieh and C. Athaudage, “Coherent optical orthogonal frequency division multiplexing,” Electron. Lett. 42(10), 587–589 (2006). [CrossRef]
2. A. J. Lowery, L. Du, and J. Armstrong, “Orthogonal frequency division multiplexing for adaptive dispersion compensation in long haul WDM systems, ” Optical Fiber Communication Conference and Exposition and The National Fiber Optic Engineers Conference, Technical Digest (CD) (Optical Society of America, 2006), paper PDP39.
4. S. L. Jansen, I. Morita, T. C. Schenk, and H. Tanaka, “Long-haul transmission of16×52.5 Gbits/s polarization-division- multiplexed OFDM enabled by MIMO processing (Invited),” J. Opt. Netw. 7(2), 173–182 (2008). [CrossRef]
5. Q. Yang, Y. Tang, Y. Ma, and W. Shieh, “Experimental demonstration and numerical simulation of 107-Gb/s high spectral efficiency coherent optical OFDM,” J. Lightwave Technol. 27(3), 168–176 (2009). [CrossRef]
6. R. Dischler and F. Buchali, “Transmission of 1.2 Tb/s continuous waveband PDM-OFDM-FDM Signal with spectral efficiency of 3.3 bit/s/Hz over 400 km of SSMF,” in National Fiber Optic Engineers Conference, OSA Technical Digest (CD) (Optical Society of America, 2009), paper PDPC2.
7. Y. Ma, Q. Yang, Y. Tang, S. Chen, and W. Shieh, “1-Tb/s single-channel coherent optical OFDM transmission over 600-km SSMF fiber with subwavelength bandwidth access,” Opt. Express 17(11), 9421–9427 (2009). [CrossRef] [PubMed]
8. S. Chandrasekhar, X. Liu, B. Zhu, and D. W. Peckham, “Transmission of a 1.2-Tb/s 24-carrier no-guard-interval coherent OFDM superchannel over 7200-km of ultra-large-area fiber,” in 35th European Conference on Optical Communication, 2009 paper PD2.6.
9. J. Yu, Z. Dong, and N. Chi, “Transmission and coherent detection of 11.2 Tb/s (112x100gb/s) single source optical OFDM superchannel,” in Optical Fiber Communication Conference and Exposition 2011, paper PDPA6.
12. R. Gross and D. Veeneman, “SNR and spectral properties for a clipped DMT ADSL signal,” in Proc. of VTC’94, June 1994, 843–847.
13. X. Li, L. J. Cimini, and Jr., “Effect of clipping and filtering on the performance of OFDM,” IEEE Commun. Lett. 2(5), 131–133 (1998). [CrossRef]
14. D. Chanda, A. Sesay, and B. Davies, “Performance of clipped OFDM signal in fiber,” Electrical and Computer Engineering, 2004. Canadian Conference vol.4, 2401- 2404 May 2004.
15. S. McCanne, “Scalable multimedia communication using IP multicast and lightweight solutions for the 3G long-term evolution,” IEEE Commun. Mag. 44(3), 38–45 (2006).
16. Y. Tang, W. Shieh, and B. S. Krongold, “Fiber nonlinearity mitigation in 428-Gb/s multiband coherent optical OFDM systems,” in Optical Fiber Communication Conference, OSA Technical Digest (CD) (Optical Society of America, 2010), paper JThA6.
17. Q. Yang, Z. He, Z. Yang, S. Yu, and X. Yi, A. A.l Amin, and W. Shieh, “Coherent optical DFT-spread OFDM in band-multiplexed transmissions,” in 37th European Conference and Exposition on Optical Communications, OSA Technical Digest (CD) (Optical Society of America, 2011), paper We.8.A.6