Open Access
Add to BibSonomyAbstract
We experimentally demonstrated the transmission of 79.86-Gb/s discrete-Fourier-transform spread 32QAM discrete multi-tone (DFT-spread 32QAM-DMT) signal over 20-km standard single-mode fiber (SSMF) utilizing directly modulated laser (DML). The experimental results show DFT-spread effectively reduces Peak-to-Average Power Ratio (PAPR) of DMT signal, and also well overcomes narrowband interference and high frequencies power attenuation. We compared different types of training sequence (TS) symbols and found that the optimized TS for channel estimation is the symbol with digital BPSK/QPSK modulation format due to its best performance against optical link noise during channel estimation.
© 2014 Optical Society of America
1. Introduction
As the emergence of many new bandwidth hungry multimedia applications such as HDTV, video call and cloud computing, there are rapidly growing demands on the capacity of optical communication systems. Coherent detection has been introduced to meet such high capacity demands for backbone transmission networks. Similarly, the demands of bandwidth are also growing rapidly in shorter reach to satisfy the applications in broadband access networks. Intensity modulated direct detection (IM/DD) systems utilizing directly modulated laser (DML) is considered as the most practical candidate for the next generation low-cost optical access networks [1–9]. Advanced modulation formats including discrete multi-tone (DMT) [1–6], half cycle 16-ary quadrature-amplitude-modulation (16QAM) [7], carrier-less amplitude phase modulation (CAP) [1, 6, 8] and pulse amplitude modulation (PAM) [9] recently have been reported in short reach optical fiber communication systems. DMT is a multi-carrier scheme which derives from orthogonal-frequency-division-multiplexing (OFDM) as a baseband version providing real valued signal transmission. It inherits all advantages of OFDM signal, such as transparent to modulation formats with frequency domain equalization (FDE) and robust to chromatic dispersion and polarization mode dispersion [2–5]. DMT is the most competitive candidate for the short reach optical transmission systems. However, the Peak-to-Average Power Ratio (PAPR) problem undoubtedly degrades the transmission performances of DMT signal. In order to reduce the PAPR, discrete-Fourier-transform-spread (DFT-spread) without any distortion is introduced in the IM/DD systems utilizing DML. In the DFT-spread DMT transmission system, an additional pair of DFT/IDFT is added in the off-line transceiver compared to conventional DMT transmission system. The digital signal mapped from Pseudo-random binary sequence is converted to analog signal after that additional DFT in the transmitter. This does not affect the DMT signal symbols generation and transmission, while it is essential for the training sequence (TS) based FDE. With that additional DFT in the TS symbol generation, digital TS symbol will be transformed to analog TS symbol. This can help to control the PAPR of TS symbol, but it may lead to the channel estimation performance degradation as digital TS symbol is obviously robust to the interference of noise during channel estimation. Without that additional pair of DFT/IDFT for TS symbol, the PAPR of DMT signal may increase a little, while the overall performance will be very good with the strong digital TS symbol.
In this paper, we extend the experimental investigation of TS symbol scheme for high capacity DFT-spread DMT transmission in IM/DD systems. PAPR improvement is obtained with DFT-spread. In addition, DFT-spread is confirmed to be effective in overcoming power attenuation and narrowband interference. 79.86-Gb/s DFT-spread 32QAM-DMT signal demonstrated the best performance with binary-phase-shift-keying/quadrature-phase-shift-keying (BPSK/QPSK) digital TS symbol and the strongest robustness against optical noise in FDE. With the optimized TS symbol, no penalty is observed and the bit error ratio (BER) is well below the requirement of hard decision Forward Error Correction (HD-FEC) threshold (3.8 × 10−3) after 20-km standard single-mode fiber (SSMF) transmission.
2. Principle
Channel estimation is a critical procedure in FDE for DMT signal. With accurate channel estimation, the physical impairments can be compensated in equalization and the signal can be recovered. Figures 1(a) and 1(b) show the block diagrams of conventional and DFT-spread optical DMT, respectively. Here the FFT size for DMT signal generation and demodulation is N. L subcarriers in the positive frequency bins are used to convey 32-QAM data, L subcarriers in the negative frequency are filled with Hermitian symmetric data to generate DMT signal and the remaining subcarriers are reserved null for dc-bias and oversampling [5]. Relative to conventional DMT, the DFT-spread DMT transceiver has an extra pair of L-point DFT/IDFT in the DMT transmitter and receiver, respectively. In the TS based FDE for DMT, the digital signal processing (DSP) operations for TS and DMT signal symbols are ensured exactly consistent. In the conventional DMT transceiver, the DSP operations for TS and DMT symbols are completely identical and the modulation format for TS is chosen to be QPSK. While in the DFT-spread transceiver shown in Fig. 1(b), the DSP operations for TS are discussed due to that additional pair of DFT/IDFT in DMT transceiver. In the transmitter of DFT-spread DMT, the TS used in FDE in the receiver can be generated with/without an extra L-point DFT. The digital and analog TSs are classified according to the type of the data before DMT modulation.If the modulation formats of the data are BPSK/QPSK/16QAM, this type of TS is classified to digital TS. The TS should be regarded as analog TS when DFT-spread is applied in the TS generation. With that extra L-point DFT, the TS becomes analog signal in frequency domain and the PAPR of corresponding time domain signal after DMT modulation becomes lower. Without that extra L-point DFT, the TS is digital signal in frequency domain and is the same as TS of conventional DMT signal. Unfortunately, the PAPR of corresponding time domain signal is a little higher. We will discuss the performance of digital and analog TSs in channel response estimation. In the digital TS schemes, BPSK/QPSK/16QAM-TSs will be discussed. In analog TS schemes, we only choose the analog TS generated by extra L-point DFT with QPSK for discussion as the distribution characteristics of all kinds of analog TSs are similar. Apart from this, conventional DMT with QPSK TS is also discussed for comparison.
Figure 2 gives the probability distribution of amplitude of DMT signal: DFT-spread DMT with BPSK/QPSK/16QAM/analog TS symbol and conventional DMT. For the DFT-spread DMT, 6 high discrete peaks appearing after that extra L-point DFT is applied for 32-QAM data. From this point of view, DFT-spread DMT can be regarded as the combination of one high probability PAM-6 signal and one low probability analog signal, and it explains why the PAPR of DFT-spread DMT is lower than conventional DMT. The probability distributions of DFT-spread DMT are quite similar except for slight differences in their TSs. In Fig. 2(d), another two small discrete peaks like PAM-2 can be seen as extra L-point DFT is applied for QPSK data in TS. The probability distribution of conventional DMT in Fig. 2(e) likes Gaussian distribution. Cumulative distribution function (CCDF) denotes a probability distribution of the PAPR of current DMT symbol is over a certain threshold. During the CCDF calculation, the N and L are 8192 and 2048, respectively. Figure 3 gives the calculated CCDF curves for conventional DMT and DFT-spread DMT with different types of TS. The PAPRs of DFT-spread DMT with different types of TS are quite similar expect for some slight differences in the TS and outperform that of conventional DMT. This means the additional DFT/IDFT pair for TS symbol will not affect the PAPR of DFT-spread DMT. 3.4-dB PAPR improvement is attained at the probability of 2 × 10−4.
Fig. 2 Probability density function of (a) DFT_spread DMT with BPSK TS, (b) DFT_spread DMT with QPSK TS, (c) DFT_spread DMT with 16QAM TS, (d) DFT_spread DMT with analog TS, (e) Conventional DMT.
3. Experimental setup
The experimental setup for the 32QAM-DMT signal transmission utilizing DML in IM-DD system is given in Fig. 4. The DMT signal is generated off-line in Matlab and then uploaded into a Fujitsu DAC with 64-GSa/s sampling rate. The 3-dB bandwidth of the DAC is 12GHz and the attenuation of the DAC at 16GHz is 4.6dB. Table 1 summarizes the parameters for the DFT-spread DMT and conventional DMT. The net bit rate is 79.86Gbit/s. The DMT signal from DAC is amplified to 2.4V by an electrical amplifier (EA) with 20-dB gain. The optical carrier at 1295.43nm from a commercial distributed-feedback (DFB) DML is directly modulated by DMT signal. The DML with 10-GHz 3-dB bandwidth and 20-MHz linewidth is biased at 89mA to produce 9.8-dBm average output power. The output signal is injected into 20-km SSMF. After fiber transmission, optical to electrical (O/E) conversion is implemented via an Agilent analog receiver. After a low pass filter (LPF), the signal is captured by a real-time Tektronix oscilloscope with 50-GSa/s sampling rate and subsequently processed by off-line DSP. The required received optical power for the receiver is −16dBm for 10-Gb/s on-off keying (OOK) signal. The 3-dB bandwidth of the Agilent receiver is 14GHz.. The off-line DSP is listed in detail in Fig. 1. Figure 4(a) gives the optical spectra (0.02-nm resolution) of the optical carrier before and after modulated by conventional DMT and DFT-spread DMT signals. As the PAPR of DFT-spread DMT signal is lower, the average electrical power of DFT-spread DMT is larger than conventional DMT. From Fig. 4(a) we can see that the optical signal-to-noise ratio (OSNR) of DFT-spread DMT with QPSK TS is slightly larger than conventional DMT. The corresponding electrical spectra of these two different DMT signals with −2.73dBm received optical power in optical back-to-back (OBTB) are shown in Fig. 4(b) and the SNR of DFT-spread DMT with digital QPSK TS is larger than conventional DMT. In this paper, BER was obtained by direct error counting with 96 DMT symbols (95 × 2048 × 5 = 972800 bits).
Table 1. Parameters for the DFT-spread DMT and Conventional DMT Modulation Schemes
4. Experimental results and discussions
The estimated channel response of DFT-spread 32QAM-DMT signal with digital QPSK TS and analog TS is shown in Figs. 5(a) and 5(b), respectively. Compared to the channel response obtained with digital QPSK TS, both the amplitude and phase of the channel response acquired with analog TS exhibit high-frequency fluctuations as the analog TS is much more vulnerable than the digital TS to the noise in the optical fiber link. Thus, the performance of channel estimation with digital TS is much better than that with analog TS.
The error bits versus frequency for five types of DMT signals in OBTB with −2.73dBm received optical power are shown in Fig. 6. The corresponding constellations are given in Fig. 7. In the conventional DMT shown in Fig. 6(e) there are some frequency bins with relatively high error ratio, which are caused by narrowband interference in ADC and DAC. And also the error ratio increases when the frequency increases, which is induced by power attenuation in high frequency bins. After DFT-spread is applied in the DMT transceiver, the error bits become evenly distributed, as each QAM data symbol is carried onto all data subcarriers after an extra 2048-point DFT. DFT-spread DMT with BPSK/QPSK TS demonstrated the best performance among these DFT-spread DMT signals, which means the digital TS outperforms analog TS and the best digital TSs are BPSK and QPSK modulated.
Fig. 6 Error distributions of (a) DFT_spread DMT with BPSK TS, (b) DFT_spread DMT with QPSK TS, (c) DFT_spread DMT with 16QAM TS, (d) DFT_spread DMT with analog TS, and (e) conventional DMT.
Fig. 7 Constellations of (a) DFT_spread DMT with BPSK TS, (b) DFT_spread DMT with QPSK TS, (c) DFT_spread DMT with 16QAM TS, (d) DFT_spread DMT with analog TS, and (e) conventional DMT.
The measured BER versus received optical power is shown in Fig. 8. The BER performance of DFT-spread DMT with digital TSs is better than that of DFT-spread DMT with analog TSs and conventional DMT. DFT-spread DMT with BPSK/QPSK TS symbol demonstrated the best performance. No power penalty is observed for 79.86-Gb/s DFT-spread 32-QAM DMT signal after 20-km SSMF transmission and the BER is well below the requirement of HD-FEC threshold (3.8 × 10−3).
5. Conclusion
Transmission and reception of 79.86-Gb/s DFT-spread 32-QAM DMT signal is successfully demonstrated after 20-km SSMF. The optimized TS for DFT-spread DMT signal is digital BPSK/QPSK symbol as it shows the strongest robustness against optical noise during channel estimation. Our results show that PAPR is reduced and DFT-spread is robust to power attenuation and narrowband interference in the channel.
Acknowledgments
This work is partly supported by NNSF of China (No. 61377079, 61325002), “863” projects under grants 2012AA011303 and 2013AA010501.
References and links
1. L. Tao, Y. Ji, J. Liu, A. P. T. Lau, N. Chi, and C. Lu, “Advanced modulation formats for short reach optical communication systems,” IEEE Netw. 27(6), 6–13 (2013). [CrossRef]
2. H. Yang, S. C. J. Lee, E. Tangdiongga, C. Okonkwo, H. P. A. van den Boom, F. Breyer, S. Randel, and A. M. J. Koonen, “47.4 Gb/s transmission over 100 m graded-index plastic optical fiber based on rate-adaptive discrete multi-tone modulation,” J. Lightwave Technol. 28(4), 352–359 (2010). [CrossRef]
3. R. P. Giddings, E. Hugues-Salas, and J. M. Tang, “Experimental demonstration of record high 19.125 Gb/s real-time end-to-end dual-band optical OFDM transmission over 25 km SMF in a simple EML-based IMDD system,” Opt. Express 20(18), 20666–20679 (2012). [CrossRef] [PubMed]
4. M. Beltrán, Y. Shi, C. Okonkwo, R. Llorente, E. Tangdiongga, and T. Koonen, “In-home networks integrating high-capacity DMT data and DVB-T over large-core GI-POF,” Opt. Express 20(28), 29769–29775 (2012). [CrossRef] [PubMed]
5. F. Li, X. Li, L. Chen, Y. Xia, C. Ge, and Y. Chen, “High level QAM OFDM system using DML for low-cost short reach optical communications,” IEEE Photon. Technol. Lett. 26(9), 941–944 (2014). [CrossRef]
6. J. L. Wei, D. G. Cunningham, R. V. Penty, and I. H. White, “Study of 100 Gigabit Ethernet using carrierless amplitude/phase modulation and optical OFDM,” J. Lightwave Technol. 31(9), 1367–1373 (2013). [CrossRef]
7. A. S. Karar and J. C. Cartledge, “Generation and detection of a 56 Gb/s signal using a DML and half-cycle 16-QAM Nyquist-SCM,” IEEE Photon. Technol. Lett. 25(8), 757–760 (2013). [CrossRef]
8. L. Tao, Y. Wang, Y. Gao, A. P. T. Lau, N. Chi, and C. Lu, “40 Gb/s CAP32 system with DD-LMS equalizer for short reach optical transmissions,” IEEE Photon. Technol. Lett. 25(23), 2346–2349 (2013). [CrossRef]
9. W. Kobayashi, T. Fujisawa, S. Kanazawa, and H. Sanjoh, “25 Gbaud/s 4-PAM (50 Gb/s) modulation and 10-km SMF transmission with 1.3-μm InGaAlAs-based DML,” Electron. Lett. 50(4), 299–300 (2014). [CrossRef]
