Pilot-based blind phase estimation for coherent optical OFDM system Xuebing Zhang, Jianping Li, Chao Li, Ming Luo, Haibo Li, Zhixue He, Qi Yang, Chao Lu 3 and Zhaohui Li,* Institute of Photonics Technology, Jinan University, Guangzhou, Guangdong, 563, China State Key Lab. of Optical Comm. Technologies and Networks, Wuhan, Hubei, 4374, China 3 Photonics Research Centre, The Hong Kong Polytechnic University, Hong Kong *li_zhaohui@hotmail.com Abstract: A pilot-based blind (PBB) phase estimation method for the coherent optical orthogonal frequency-division multiplexing (CO-OFDM) system is demonstrated in this paper. Instead of inserting the pilotsubcarriers that are loaded with the already known information in the OFDM signal, the unknown and specially designed signal is used to replace the signal on the pilot subcarriers to decrease the waste of spectrum and has demonstrated good performance in the phase noise compensation. Therefore, the spectral efficiency (SE) is further improved compared with the conventional pilot-aided (PA) phase noise estimation method. Both the proposed PBB and conventional PA estimation methods are compared in a CO-OFDM transmission experiment, which is modulated by 4 quadrature amplitude modulation (4-QAM) formats and transmitted over 76-km standard single-mode fiber (SSMF) without optical dispersion compensation. The experimental results show that the proposed PBB method can achieve the similar performance as the conventional PA method. 4 Optical Society of America OCIS codes: (6.33) Fiber optics communications; (6.66) Coherent communications; (6.48) Modulation. References and links. S. J. Savory, G. Gavioli, R. I. Killey, and P. Bayvel, Electronic compensation of chromatic dispersion using a digital coherent receiver, Opt. Express 5(5), 6 (7).. C. R. S. Fludger, T. Duthel, D. van den Borne, C. Schulien, E.-D. Schmidt, T. Wuth, J. Geyer, E. de Man, G.-D. Khoe, and H. de Waardt, Coherent equalization and POLMUX-RZ-DQPSK for robust -GE transmission, J. Lightwave Technol. 6(), 64 7 (8). 3. X. Liu, S. Chandrasekhar, B. Zhu, P. J. Winzer, A. H. Gnauck, and D. W. Peckham, 448-Gb/s reducedguardinterval CO-OFDM transmission over km of ultra-large-area fiber and five 8-GHz-grid ROADMs, J. Lightwave Technol. 9(4), 483 49 (). 4. D. Qian, M. F. Huang, E. Ip, Y. K. Huang, Y. Shao, J. Hu, and T. Wang, Highcapacity/spectral efficiency.7-tb/s WDMtransmission using PDM-8QAM-OFDM over65-km SSMF within C- and L-bands, J. Lightwave Technol. 3(), 54 548 (). 5. W. Shieh, Maximum-likelihood phase and channel estimation for coherent optical OFDM, Photonics Technol. Lett. (8), 65 67 (8). 6. X. Yi, W. Shieh, and Y. Ma, Phase noise effects on high spectral efficiency coherent optical OFDM transmission, J. Lightwave Technol. 6(), 39 36 (8). 7. V. Syrjala, M. Valkama, N. N. Tchamov, and J. Rinne, Phase noise modelling and mitigation techniques in OFDM communications systems. in Proc. WTS(9), PP 7. 8. S. T. Le, T. Kanesan, M. E. McCarthy, E. Giacoumidis, I. D. Phillips, M. F. C. Stephens, M. Tan, N. J. Doran, A. D. Ellis, and S. K. Turitsyn, Experimental demonstration of data-dependent pilot-aided phasenoiseestimation for CO-OFDM. in Proc. OFC (4), Tu3G.4. 9. S. Hussin, K. Puntsri, and R. Noé, Performance analysis of RF-pilot phase noise compensation techniques in coherent optical OFDM systems. in Proc. NOC (), pp 5.. C. Zhao, C. Yang, F. Yang, F. Zhang, and Z. Chen, A CO-OFDM system with almost blind phase noise suppression, Photonics Technol.Lett. 5(7), 73 76 (3). (C) 4 OSA September 4 Vol., No. 9 DOI:.364/OE..888 OPTICS EXPRESS 888
. M. K. Lee, S. C. Lim, and K. Yang, Blind compensation for phase noise in OFDM systems over constant modulus modulation, Transactions on Communications 6(3), 6 65 ().. Y. Ha and W. Chung, Non-Data-Aided phase noise suppression scheme for CO-OFDM systems, Photonics Technol. Lett. 5(7), 73 76 (3). 3. S. Cao, P. Y. Kam, and C. Yu, Decision-aided, pilot-aided, decision-feedback phase estimation for coherent optical OFDM systems, Photonics Technol. Lett. 4(), 67 69 (). 4. X. Yi, W. Shieh, and Y. Tang, Phase estimation for coherent optical OFDM, Photonics Technol. Lett. 9(), 99 9 (7). 5. D. Ai, H. Bao, and Y. Yang, A new method of calculating generation linear distribution factors by least square fitting. in Proc. SUPERGEN(9), pp. 5.. Introduction To meet the ever-increasing optical bandwidth demand, the optical communication system supporting -Tb/sand beyond per channel will soon be required. Coherent optical orthogonal frequency-division multiplexing (CO-OFDM) is a promising candidate technology for the required large capacity, long haul transmission and has been developed rapidly in recent years [ 4].However, the phase noise is still one of the main limitations for the CO-OFDM transmission performance [5 7].It is important to compensate the phase noise with a cost as low as possible for the CO-OFDM system. So far, there are two main research directions: i) non-blind-based method; ii) blind-based method. For the former scheme, many methods have been proposed and studied widely, such as pilot-aided, RF pilot-aided and pseudo pilot-aided [8 ]. But all of them need one or more pilots to transmit the already known information for the phase noise estimation, which will decrease the spectral efficiency (SE) of the system. On the contrary, the SE will be improved for the blind-based method in the phase noise estimation since none of already known data is required. By using the specially designed digital signal processing (DSP) algorithm, the phase noise can be calculated from the unknown information. So the blind methods have higher SE, which is attractive for the large capacity transmission system. Recently, a blind compensation for phase noise in OFDM systems over constant modulus modulation has been proposed []. Non-data-aided phase noise suppression scheme for CO-OFDM systems has also been proposed []. In addition, the interesting joint phase estimation method for coherent optical OFDM systems has been proposed and demonstrated [3]. Joint compensation algorithm combines two different techniques to achieve better performance than only using one technique. In this paper, a pilot-based blind (PBB) phase estimation method for CO-OFDM system is proposed. Instead of inserting the pilot-subcarriers which are loaded with the already known information in the OFDM signal, the unknown and specially designed signals are used to replace the signals on the pilot subcarriers to improve the SE and the good performance in the phase noise compensation is demonstrated. Therefore, the SE is further improved compared with the conventional pilot-aided (PA) phase noise estimation method. We experimentally demonstrated a CO-OFDM transmission with 4 quadrature amplitude modulation (4-QAM) format over 76-km standard single-mode fiber (SSMF), in which both the proposed PBB and conventional pilot-aided (PA) estimation methods are applied and compared. The experimental results show that the proposed PBB method can achieve the similar performance as the conventional PA method.. Principle of the PBB phase noise estimation method The principle of the proposed PBB phase noise estimation method is shown in Fig.. Figure (a) shows the two-dimensional time/frequency structure of one OFDM frame by using the PBB estimation method. K subcarriers are selected for the phase estimation. Unlike transmitting the already known signal used in the conventional PA estimation method, the subcarriers are used to transmit the unknown signal with amplitude modulation (AM) format as shown in Fig. (b). In half of the K selected subcarriers, the amplitudes of the AM signal are, 3, 5, 7 and N, respectively. Then, at the receiver, all the received random values on half of the K subcarriers and the zero value are used for the linear fitting. The angle between (C) 4 OSA September 4 Vol., No. 9 DOI:.364/OE..888 OPTICS EXPRESS 889
the fitting line and Re axis is the coarse rotated phase noise φ as shown in Fig. (c). In order to increase the accuracy of linear fitting, the amplitudes of, 3, 5, 7 and -N in the left half of the K selected subcarriers and zero value repeat the processing. Combining the two results, we get the final coarse phase noise. Then zero value and all of the positive random values and negative random values on the K subcarriers are used for the more accurate linear fitting to implement the fine phase noise estimation. Additionally, the order of the AM signal on the selected subcarriers can be flexible adjusted along with the modulation order of the data signal on the other subcarriers to improve the transmission capacity. Pilot subcarriers Time (a)...... K... Subcarriers (b) Im Im 3 5 7 K... N -N... Re -7-5 -3 - Re Im Linear fitting (c) Fine estimation Coarse estimation Re Coarse estimation Fig.. The principle of the proposed PBB phase noise estimation method. (a) Twodimensional time/frequency structure of one OFDM frame by using the conventional PA estimation method; (b) The proposed amplitude modulation formats; (c) The linear fitting for phase noise estimation Figure shows the main DSP procedures of the transmission system. Pseudo-random binary sequence (PRBS) is split into two paths. The upper path is mapped to the 4-QAM signal and the lower path is mapped to the 4-AMsignal for the phase noise estimation. Then the 4-AM signal is pre-modulated to decrease peak-to-average power ratio by increasing the random by multiplying the random ± or ± j, which is saved as a table for demodulation at the receiver. Then all the signals are used to generate the optical OFDM signal. After being transmitted through the optical channel, the inverse processes are required to recover the 4- QAM and 4-AM signals, respectively. The 4-AM signal is separated from the signal after the (C) 4 OSA September 4 Vol., No. 9 DOI:.364/OE..888 OPTICS EXPRESS 89
demodulation of OFDM signal (De-OFDM) for the phase estimation. As the feedback to the De-OFDM process, the output signal of the phase estimation is used to compensate the phase noise of the OFDM signal. Then the 4-QAM and 4-AM signals are de-mapped, respectively. Although both of the joint blind phase estimation method [3] and the PBB method can achieve good performance, the proposed PBB method doesn t need the processing of demodulation and decision-feedback compared with joint blind phase estimation method [3]. Therefore, the algorithm structure of the PBB method is relatively simple. PRBS 4-QAM Mapping 4-AM Mapping OFDM Generation Pre- Modulation Optical Channel De-OFDM De-Pre- Modulation Phase Estimation 4-QAM Demapping 4-AM Demapping Fig.. The main DSP procedures of the transmission system 3. Experimental setup The experimental setup of the CO-OFDM system using the PBB/PA phase noise estimation methods is illustrated in Fig. 3. In the experiment, an external-cavity laser (ECL) with linewidth less than -KHz is used as the optical carrier modulated by the OFDM signal that drives I/Q modulator. An arbitrary waveform generator (AWG) running at -GS/s is employed to produce OFDM/4QAMbasebandsignal.The FFT size is 56, in which 64 subcarriers are loaded with OFDM/4QAM signals. The center one subcarrier is unloaded to avoid the DC influence. 4 subcarriers are selected for the phase noise estimation. For the PBB method, the 4 subcarriers are loaded with 4-AM and the proposed phase noise estimation algorithm is used. However, for the PA method, the 4 subcarriers are the pilots loaded with the already known 4-QAM signal. In this experiment, TSs are periodically inserted in the front of each OFDM/4QAM frame, which is then followed by 5 payload symbols. The cyclic prefix length is the /8 FFT size. The data rate of OFDM signal is 3.-Gb/s for 4- QAM loading case. The transmission link is constructed by spans of 8 km SSMF with Raman amplification. In the receiver, the typical coherent receiver is used to detect the optical OFDM signal into electrical signal, and then is sampled by a digital storage oscilloscope (DSO Tektronix DSA74B) operating at 5 GS/s. Off-line DSP is done by the MATLAB program. ECL I AWG GS/s Q Optical IQ Mod. Transmitter OFDM Generator EDFA VOA 8km RAMAN ECL EDFA Fig. 3. Experimental setup of the OFDM system Optical Hybrid Receiver Oscillator scope I Q Offline Processing 4. Experimental results and Discussions Similar to [4], the number of the selected subcarriers is one of the important issues in the proposed PBB phase noise estimation method. The Q factor versus the number of the selected subcarriers at optical back-to-back is shown in Fig. 4. Here, the Q factor is defined as the average signal-to-noise ratio of the QAM and AM signals. The Q factor of subcarriers loading is.-db that is lower than the others when more than subcarriers are used. Thus (C) 4 OSA September 4 Vol., No. 9 DOI:.364/OE..888 OPTICS EXPRESS 89
subcarriers are not enough for the first-order fitting in the PBB estimation method. When 4 and more subcarriers are used, the Q factor states almost.5-db.thus,4 subcarriers are adequate for the phase estimation based on PBB method. As a result, only 4 subcarriers are used for phase noise estimation in the demonstration. Because the computational complexity depends on the number of linear fitting points, 5 points will not add much computational complexity. The linear fitting processing is carried out in three stages. First and second are used for coarse estimation. Each of them contains 3 points. According to [5], only 34 multiplications and 7 additions are needed. While in the third stage, 5 points are used for linear fitting. The processing in this stage consumes 76multiplications and 67additions. 4 8 4 6 8 4 Subcarrier Number Fig. 4. Q factor versus the number of selected subcarriers at optical BB. From the previous analysis, 4-AM signal with the absolute amplitudes of, 3, 5 and 7 is used in the PBB estimation method. But the higher amplitude signal will occupy more optical power, which can decrease the performance of the signal on the other subcarriers. The absolute amplitudes of the AM signal can be adjusted to accomplish better performance. Therefore, we investigate the Q factor performance versus amplitude factor(af) of the AM signal at optical back-to-back and over 76-km SSMF as shown in Fig. 5.Here, the AF is defined as the absolute amplitudes of 4-AM signal divided by the uniform amplitudes of, 3, 5 and 7. For example, when the AF is., the absolute amplitudes of 4-AM signal are.,.3,.5 and.7 respectively. It can be seen that the factor of.4is adequate for the phase estimation because the Q factor performance is barely improved beyond.4. 5 5 BB-PBB 76km-PBB..4.6.8. Amplitude Factor Fig. 5. Q factor performance versus amplitude factor. From the analysis above, we then set the number and amplitude factor of the selected subcarriers are 4 and.4, respectively. Meanwhile, for the purpose of comparison, 4 pilot (C) 4 OSA September 4 Vol., No. 9 DOI:.364/OE..888 OPTICS EXPRESS 89
subcarriers are also used in the PA phase estimation method. The Q factor versus launched power over 76-km SSMF with PBB and PA phase noise estimation methods are shown in Fig. 6. The optimized launched powers of the PBB and PA methods are.3-dbm and.6dbm, respectively. The.3-dBmincrease is mainly caused by the optical power radio decreasing of the subcarriers with 4-QAM format signal in the whole signal. The selected 4 subcarriers for the phase noise estimation occupy more optical power of the whole signal compared with the PA method. Thus a little higher launched power is obtained in the PBB method. 8.5.5.5.5 -.5 -.5 - - -.5 -.5 6 - - -.5 - -.5.5.5 - - -.5 - -.5.5.5 4 PBB PA -9-6 -3 Launch power (dbm) 3 6 Fig. 6. Q factor versus launch power over 76-km SSMF. We further evaluate the transmission performance of the PBB and PA phase noise estimation systems over 76-km fiber links. Figure 7 shows the Q factor versus optical signal-to-noise (OSNR) curves of the systems. The OSNR values are almost the same (~6.dB) at the E-3 limit (Q factor = 9.8-dB) for the two phase noise estimation methods at optical BB. After 76-km SSMF, the required OSNR values are 7.-dB and 6.6-dB respectively at the E-3 limit. There is only.4-db OSNR decrease. Thus the proposed PBB method can accomplish the comparable performance with the conventional PA method. 6.6dB.4dB e-3 Limit BB-PBB 8 BB-PA 76km-PBB 76km-PA 4 4 6 8 OSNR (db) 4 Fig. 7. Q factor versus OSNR curves using two schemes. 5. Conclusion In this paper, a PBB phase estimation method for CO-OFDM has been proposed. A 3.-Gb/s CO-OFDM transmission experiment modulated with 4-QAM format signal over 76-km SSMF without optical dispersion compensation is demonstrated to compare the proposed PBB method with the conventional PA estimation method. Compared to the conventional method, the similar performance of the proposed blind method is obtained. #64 - $5. USD Received 4 Jul 4; revised Sep 4; accepted 6 Sep 4; published Sep 4 (C) 4 OSA September 4 Vol., No. 9 DOI:.364/OE..888 OPTICS EXPRESS 893
Acknowledgments The authors would like to acknowledge the support of National High Technology 863 Research and Development Program of China (No. 3AA33), (No. 3AA343), National Natural Science Foundation of China (NSFC) under Grant (No. 6379, 64356)and New Century Excellent Talents in University (NCET--679). (C) 4 OSA September 4 Vol., No. 9 DOI:.364/OE..888 OPTICS EXPRESS 894