We experimentally demonstrated a 256-ary quadrature amplitude modulation (256QAM) direct-detection optical orthogonal frequency division multiplexing (DDO-OFDM) transmission system utilizing a cost-effective directly modulated laser (DML). Intra-symbol frequency-domain averaging (ISFA) is applied to suppress in-band noise while the channel response estimation and Discrete Fourier Transform-spread (DFT-spread) is used to reduce the peak-to-average power ratio (PAPR) of the transmitted OFDM signal. The bit-error ratio (BER) of 15-Gbit/s 256QAM-OFDM signal has been measured after 20-km SSMF transmission that is less than 7% forward-error-correction (FEC) threshold of 3.8 × 10−3 as the launch power into fiber is set at 6dBm. For 11.85-Gbit/s 256QAM-OFDM signal, with the aid of ISFA-based channel estimation and PAPR reduction enabled by DFT-spread, the BER after 20-km SSMF transmission can be improved from 6.4 × 10−3 to 6.8 × 10−4 when the received optical power is −6dBm.
© 2014 Optical Society of America
It is well known that orthogonal frequency division multiplexing (OFDM) with cyclic prefix (CP) is robust to chromatic dispersion (CD) and polarization mode dispersion (PMD) inherently existing in fiber-optics transmission system. It can be potentially applied for high spectral efficiency (SE) and flexible networks [1–14]. The direct-detection optical OFDM (DDO-OFDM) system has a simple and cost-effective configuration and tends to be applied in next generation passive optical networks (NG-PON2) [4, 15]. Directly modulated laser (DML) has the advantages of lower cost, compact size, lower power consumption, and larger output power [1–5] and it is considered as a competitive candidate for Capital Expenditures (CapEx) control in PON. Meanwhile, to fulfill the requirement of the NG-PON2, the bit rate up to tens of Gbit/s or even 40Gbit/s [6, 15] are necessary. In order to realize high-bit-rate PONs with DML, the most straightforward solution is to increase the SE with high level modulation formats. Both 64-ary quadrature amplitude modulation (64QAM) and 128QAM have been demonstrated in the DDO-OFDM access networks . As far as we concerned, 128QAM is the highest modulation format ever reported in the real-time DDO-OFDM system with 25-km standard single-mode fiber (SSMF) or 500-m multi-mode fiber (MMF) transmission . In optical high level QAM OFDM systems, PAPR reduction techniques are important challenges in order to increase their tolerance to optical transmitter and ðber nonlinearity impairments [9, 14]. Constant envelope (CE) OFDM (CE-OFDM) technique was proposed to be applied in the optical link to alleviate the nonlinearity impairments. In this technique, one additional pair of DFT/IDFT and phase modulation/demodulation should be realized in digital signal processing (DSP). The bandwidth should be broadened to obtain good performance. Compared to CE-OFDM, Discrete Fourier Transform-spread OFDM (DFT-spread OFDM) can be realized without phase modulation/demodulation in DSP. At the same time, the bandwidth can be maintained the same. Thus DFT-spread OFDM is much practical in the short reach NG-PON2 where the ðber nonlinearity impairments are negligible.
In this paper, a simple DDO-OFDM system with 256-QAM modulation format is demonstrated proof-of-concept utilizing a cost-effective DML. The maximum bit rate can reach 15Gbit/s after 20-km SSMF transmission under the 7% hard-decision forward-error-correction (HD-FEC) threshold of 3.8 × 10−3. The intra-symbol frequency-domain averaging (ISFA) [12, 13] is utilized to suppress the noise from the photo-diode (PD) during channel estimation with training sequence (TS). DFT-spread is also adopted to reduce the PAPR induced nonlinear impairments . For 11.85-Gbit/s 256QAM-OFDM signal, with the aid of ISFA-based channel estimation and PAPR reduction are enabled by DFT-spread. The bit-error ratio (BER) after 20-km SSMF transmission can be improved from 6.4 × 10−3 to 6.8 × 10−4 when the received optical power is −6dBm.
2.1 ISFA based channel estimation
Channel estimation is a critical procedure in OFDM transmission systems . With accurate channel estimation, the physical impairments of the ðber transmission link such as CD and PMD can be obtained. Hence, the subsequent channel equalization can be performed to restore the signal quality. Time-domain averaging method can be applied during the channel estimation to suppress noise term. However, it will lead to SE degradation [5, 12]. ISFA is an alternative scheme without SE degradation. ISFA is based on the moving average theory to suppress the noise during the channel estimation. To this end we note it as a low pass filter (LPF) with odd taps only [12, 13].
In the short-reach DDO-OFDM system, the noise from the PD is dominant and can be approximated with Gaussian-like distributions. Numerical simulation is carried out to verify the noise suppression with ISFA in the back-to-back (BTB) DDO-OFDM system. In our simulation, the additive white Gaussian noise (AWGN) channel model is used to represent the optical link when only the dominant noise from the PD is considered. The power, length and baud rate of the original AWGN sample are set to 0dBm, 215 and 10Gbaud, respectively. The noise power in frequency domain with different ISFA taps which represent the size of average windowing is given in the Fig. 1(a).It can be seen that the noise power is reduced along with the increase of ISFA taps. In order to reduce the impact of noise during the channel estimation, the ISFA taps should not be too small. The obtained channel response with TS is composed by the accurate channel estimation in which the neighbor subcarrier channels are highly correlated and noise with Gaussian-like distributions. In order to suppress this Gaussian-liked noise, the average samples during ISFA should be very large. While the channel estimation will become inaccurate as the correlations among these averaged samples reduce. Therefore, there is a tradeoff between the noise suppression and the mean squared error of estimated channel in selecting ISFA taps. Then OFDM signal has transmitted in AWGN channel, the modulation format of the OFDM signal is 16QAM and the Fast Fourier Transform (FFT) size is 256. The BER versus the signal to noise ratio (SNR) for the OFDM signals with and without ISFA is shown in Fig. 1(b). The ISFA tap number is 11. The ISFA can introduce the sensitivity improvement of 3-dB receiver at the BER of 3.8 × 10−3, which demonstrates that the ISFA is effective to improve the estimation accuracy of the channel via noise suppression.
2.2 PAPR reduction with DFT-spread
The generation and recovery principle of DFT-spread OFDM signal are shown in Fig. 2(a). They introduce one more DFT operation in the signal generation and one more IDFT operation in the signal recovery than conventional OFDM scheme, respectively. k signal-carrying subcarriers in the positive frequency bins are firstly processed by k-point DFT and thus the signal length becomes N/2 after zero padding. Afterward the complex conjugation operation is implemented to satisfy the Hermitian symmetry and the result signal with the length of N is mapped from frequency domain to time domain by N-point IDFT. The second N-point DFT operation in the receiver is similar to the conventional OFDM scheme. We evaluate the PAPR performance by complementary cumulative distribution function (CCDF). CCDF denotes a probability distribution of the PAPR of current OFDM symbol is over a certain threshold. Figure 2(b) gives the calculated CCDF curves for traditional OFDM signal and DFT-spread OFDM signal. The PAPR of DFT-spread OFDM signal outperforms that of traditional OFDM signal. Accordingly, 2.5-dB PAPR improvement is attained at the probability of 1 × 10−4.
3. Experimental setup
The experimental setup for the 256QAM-OFDM signal transmission utilizing DML in intensity-modulation and direct-detection (IM-DD) system are given in Fig. 3. At the transmitter, the optical carrier at 1537.54 nm generated from a commercial DML has been directly modulated by an electrical 256QAM-OFDM signal. The OFDM signal is generated off-line in Matlab® and then uploaded into an arbitrary waveform generator (AWG, Tektronix 7122B) with 4-GSa/s sample rate. Here the FFT size for electrical OFDM generation is 256. 100 subcarriers in the positive frequency bins are used to convey data and other 100 subcarriers in the negative frequency are filled with Hermitian symmetric data to generate real value OFDM signal, the first subcarrier is set to zero for DC-bias and the rest 55 null subcarriers at the edge are reserved for oversampling. The subcarrier symbol rate is 15.625MSa/s without considering CP. All the 100 information-bearing subcarriers in the positive frequency bins are mapped via 256QAM. A 14-sample CP is added to the 256 samples and thus each OFDM symbol contains 270 samples. One TS is inserted into every 160 OFDM symbols in order to realize time synchronization and channel estimation. The total bit rate is 11.85Gb/s (4 × 100/270 × 8Gb/s≈11.85Gb/s) and the signal bandwidth is 1.56GHz (100/256 × 4GHz≈1.56GHz). A LPF is used after digital to analog convertor (DAC) to suppress residual alias spectrum and then the OFDM signal is amplified to 2.4V (peak-to-peak) by an electrical amplifier (EA) before injected into a distributed feedback (DFB) based DML. For optical OFDM modulation, the DML with 10-GHz bandwidth and 20-MHz linewidth is biased at 52mA to produce 6-dBm average output power. Since the bandwidth of OFDM signal is narrow and the optical fiber transmission distance is short, intensity noise converted form phase noise induced by the CD in fiber link  after optical-to-electrical (O/E) conversion is negligible even the linewidth of DML is 20-MHz in the experimental setup . The inset (a) of Fig. 3 shows the optical spectra (0.01-nm resolution) of the optical carrier before and after modulated by the electrical OFDM signal, respectively. From the figure we can see the optical spectrum slight broadening after OFDM signal is modulated onto the optical carrier. As the power of the optical carrier located in the middle of optical OFDM spectrum is much higher than that of the signal, the model of the optical-to-electrical conversion of optical OFDM signal at the receiver can be found in our previous work  and the signal-to-signal beating noise after PD can be neglected due to the high power level carrier. The generated signal is injected into the first erbium-doped fiber amplifier (EDFA), which is mainly used to adjust the launch power into fiber. After fiber transmission, an optical attenuator (ATT) is applied to adjust the received optical power for sensitivity measurement. Another EDFA is cascaded to pre-amplify the signal and then a tunable optical filter (TOF) with 0.33-nm bandwidth is used to suppress the out-of-band amplified-spontaneous-emission (ASE) noise from the EDFA. It is worth noting that here we use EDFA and TOF before PD to demonstrated long reach (>40km) PON. For practical use of short reach (<20km) PON, EDFA and TOF can be removed to relax the cost of PON. O/E conversion is implemented via a PD with 3-dB bandwidth of 10GHz. The power injected into the PD is maintained at 3dBm by the second EDFA. Note that the required received optical power for the PD is −16dBm for 10-Gb/s on-off keying (OOK) signal. The converted electrical signal is sent into a real-time oscilloscope (Agilent DSO81304B) with 20-GSa/s sample rate. Subsequently, it can be processed by off-line digital signal processing (DSP). The off-line DSP includes CP removal, FFT, channel estimation with ISFA, one-tap equalization, 256QAM de-mapping and BER calculation. In this experiment, each OFDM symbol contains 800 bits and the BER is obtained by direct error counting with 640 OFDM symbols (640 × 800 = 512000 bits). The resolution of DAC in the AWG and analog to digital convertor (ADC) in real-time oscilloscope is 10 and 8 bits, respectively.
The inset (b) of Fig. 3 gives the received 256QAM constellation in the electrical back to back (eBTB) case. Digital pre-equalization is used to compensate for static distortions of the AWG and DML. Time-domain averaging method is adopted to acquire static channel response with 161 OFDM symbols. After obtaining the channel response, pre-equalization can be implemented to compensate high frequency power attenuation. The insets (c) and (d) of Fig. 3 show the electrical spectra of the received OFDM signal before and after pre-equalization, respectively. The pre-equalization can flatten the electrical spectrum and is vital for the successful implementation of our proposed system.
4. Experimental results and discussions
In our proposed 256QAM DDO-OFDM system, the ISFA based on the moving average theory is applied during the channel estimation to suppress the noise. As mentioned in section 2.1, the ISFA can be regarded as a LPF with only odd taps and taps should be optimized. Figures 4(a) and 4(b) show the amplitude and phase of the estimated channel response H as a function of the index of modulated subcarriers without and with ISFA, respectively. The received optical power is maintained at −9dBm and the adopted ISFA tap number is 7. Evidently, the ISFA can significantly make both amplitude and phase ñuctuations to be smoother. In order to find the optimized ISFA taps, Fig. 4(c) shows the measured Q-factor versus the number of ISFA taps. The Q-factor can be derived from the BER and DFT-spread is not adopted. The measurements are given with the received optical power of −6, −9 and −11dBm in both optical back-to-back (OBTB) and after 20-km SMF. The optical power launched into fiber is maintained at 6dBm and the received optical power is adjusted by the attenuator. It can be seen that the optimal ISFA taps should be 5 and in the following discussion the ISFA taps are fixed at 5. Figure 4(d) gives the measured Q-factor versus the launch power into 20-km SSMF. The received optical power is maintained at −6dBm. Both ISFA and DFT-spread are adopted. It can be seen that the optimal launch power into fiber is 6dBm. When the launch power is larger than 6dBm, the system performance is degraded due to the enhanced nonlinear fiber transmission effects. The received 256QAM constellations at the launch power of −4, 6 and 10dBm are inserted in Fig. 4(d), respectively. The constellation at the launch power of 6dBm exhibits the best performance.
Figure 5(a) shows the BER versus the received optical power before and after 20-km SSMF transmission with 6-dBm launch power. The dash line in Fig. 5(a) represents the channel estimation without ISFA, while the solid lines indicate the channel estimation with ISFA. The w and w/o represent the implementation with and without DFT-spread, respectively. With the aid of ISFA based channel estimation and PAPR reduction with DFT-spread, the BER of 11.85-Gbit/s 256QAM-OFDM signal after 20-km SSMF transmission can be improved from 6.4 × 10−3 to 6.8 × 10−4 with −6dBm received optical power. Nonlinearity during the electrical-to-optical conversion by the DML driven by the signal with high voltage after EA cannot be avoided and it leads to the error floor. Moreover, the significant receiver sensitivity improvement due to the DFT-spread in the OBTB case is quite similar to that after 20-km fiber transmission. When the launch power into fiber is only 6-dBm, the nonlinear noise for fiber transmission compared the transmitter nonlinearity (mainly including DML modulation nonlinearity and EA nonlinearity) can be neglected. Which means the DFT-spread is mainly used to suppress the nonlinear noise from transmitter. If the launch power is increased further, the fiber nonlinearity will become more and more obvious. At that situation the DFT-spread can suppress not only the nonlinearity in transmitter but also the nonlinearity in optical fiber. Figure 5(b) shows the BER of the 11.85-Gbit/s 256QAM-OFDM signal versus fiber span at 6-dBm launch power into fiber. The BER performance degrades with the increase of fiber span, but the BER is still no more than 7% HD-FEC threshold of 3.8 × 10−3 after more than 35-km fiber transmission. Figure 5(c) shows the BER versus the raw bit rate of the DAC after 20-km SSMF transmission and the raw bit rate is obtained by adjusting the sample rate of AWG. The received optical power is maintained at −6dBm. The BER is below the 7% HD-FEC threshold of 3.8 × 10−3 when the raw bit rate is no more than 15Gbit/s, and below the 20% soft-decision FEC (SD-FEC) threshold of 2.7 × 10−2  when the raw bit rate is no more than 28Gbit/s. The received 256QAM constellations at the bit rates of 6, 15 and 24Gbit/s are inserted in Fig. 5(c), respectively.
For the first time, we demonstrate a simple DDO-OFDM system with 256QAM modulation format utilizing a cost-effective DML. The BER of 11.85-Gbit/s 256QAM-OFDM signal after 20-km SSMF transmission is 6.8 × 10−4 with −6 dBm received optical power. After 20-km SSMF transmission, the BER of 15Git/s 256QAM-OFDM signal is less than 3.8 × 10−3. We believe this scheme is cost-effective and easy to be implemented for the next generation short-reach fiber-optics access system with high capacity and SE.
References and links
1. D. Qian, J. Yu, J. Hu, P. N. Ji, and T. Wang, “11.5Gb/s OFDM transmission over 640km SSMF using directly modulated laser,” in Proc. ECOC 2008, paper Mo3E4 (2008).
2. R. P. Giddings, X. Q. Jin, and J. M. Tang, “First experimental demonstration of 6Gb/s real-time optical OFDM transceivers incorporating channel estimation and variable power loading,” Opt. Express 17(22), 19727–19738 (2009). [CrossRef] [PubMed]
3. X. Q. Jin, R. P. Giddings, E. Hugues-Salas, and J. M. Tang, “Real-time demonstration of 128-QAM-encoded optical OFDM transmission with a 5.25bit/s/Hz spectral efficiency in simple IMDD systems utilizing directly modulated DFB lasers,” Opt. Express 17(22), 20484–20493 (2009). [CrossRef] [PubMed]
4. 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]
5. M. F. Huang, J. Yu, D. Qian, N. Cvijetic, and G. K. Chang, “Lightwave centralized WDM-OFDM-PON network employing cost-effective directly modulated laser,” in Proc. OFC 2009, paper OMV5 (2009).
6. 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]
7. 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]
8. J. Yu, M. Huang, D. Qian, L. Chen, and G. K. Chang, “Centralized lightwave WDM-PON employing 16-QAM intensity modulated OFDM downstream and OOK modulated upstream signals,” IEEE Photon. Technol. Lett. 20(18), 1545–1547 (2008). [CrossRef]
9. J. Silva, A. Cartaxo, and M. Segatto, “A PAPR reduction technique based on a constant envelope OFDM approach for ﬁber nonlinearity mitigation in optical direct-detection systems,” IEEE/OSA J. Opt. Commun. Netw. 4(4), 296–303 (2012). [CrossRef]
10. W.-R. Peng, “Analysis of laser phase noise effect in direct-detection optical OFDM transmission,” J. Lightwave Technol. 28(17), 2526–2536 (2010). [CrossRef]
11. Z. Cao, J. Yu, W. Wang, L. Chen, and Z. Dong, “Direct-detection optical OFDM transmission system without frequency guard band,” IEEE Photon. Technol. Lett. 22(11), 736–738 (2010). [CrossRef]
12. Q. Yang, N. Kaneda, X. Liu, and W. Shieh, “Demonstration of frequency-domain averaging based channel estimation for 40 Gb/s CO-OFDM with high PMD,” IEEE Photon. Technol. Lett. 21(20), 1544–1546 (2009). [CrossRef]
14. Y. Tang, W. Shieh, and B. S. Krongold, “DFT-spread OFDM for fiber nonlinearity mitigation,” IEEE Photon. Technol. Lett. 22(16), 1250–1252 (2010). [CrossRef]
15. F. J. Effenberger, “The XG-PON System: Cost Effective 10 Gb/s Access,” J. Lightwave Technol. 29(4), 403–409 (2011). [CrossRef]
16. S. Yamamoto, N. Edagawa, H. Taga, Y. Yoshida, and H. Wakabayashi, “Analysis of laser phase noise to intensity noise conversion by chromatic dispersion in intensity modulation and direct detection optical-fiber transmission,” J. Lightwave Technol. 8(11), 1716–1722 (1990). [CrossRef]
17. D. Chang, F. Yu, Z. Xiao, N. Stojanovic, F. N. Hauske, Y. Cai, C. Xie, L. Li, X. Xu, and Q. Xiong, “LDPC convolutional codes using layered decoding algorithm for high speed coherent optical transmission,” OFC 2012, OW1H.4 (2012).