Suppression of SRS induced crosstalk in RFvideo overlay TWDM-PON system using dicode coding Jun Li, Meihua Bi, Hao He,* and Weisheng Hu State Key Laboratory of Advanced Optical Communication Systems and Networks, Department of Electronic Engineering, Shanghai Jiao Tong University, Shanghai, 200240, China * [email protected]
Abstract: In this paper, we investigate the nonlinear Raman crosstalk in RF-video overlay time and wavelength division multiplexed passive optical network (TWDM-PON), and propose a novel spectrum-reshaping method based on dicode coding to mitigate this crosstalk. The dicode coding features ultra-low power spectral density in the low frequency region, which can reduce the nonlinear Raman crosstalk on the RF-video signal effectively. Experimental results show that, compared with traditional nonreturn-to-zero on-off keying (NRZ-OOK) signals, the crosstalk on RF-video signal can be reduced by 10 ~14 dB when the launch power per TWDMPON channel varies from 10-dBm to 15-dBm. The transmission of 10-Gb/s dicode signal over 20-km standard single mode fiber (SSMF) is also demonstrated with the receiver sensitivity of −31 dBm at bit error ratio (BER) of 3.8e-3. ©2014 Optical Society of America OCIS codes: (060.2330) Fiber optics communications; (060.4250) Networks; (190.5650) Raman effect.
References and links 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.
ITU-T recommendation G.989.1, “40-Gigabit-capable passive optical networks (NG-PON2): General requirements,” 2013. ITU-T recommendation G.989.2(draft), “40-Gigabit-capable passive optical networks: Physical media dependent layer specification,” 2013. Y. Ma, Y. Qian, G. Peng, X. Zhou, X. Wang, J. Yu, Y. Luo, X. Yan, and F. Effenberger, “Demonstration of a 40Gb/s time and wavelength division multiplexed passive optical network prototype system,” in Proc. OFC 2012, paper PDP5D.7. R. Gaudino, V. Curri, and S. Capriata, “Propagation impairments due to Raman effect on the coexistence of GPON, XG-PON, RF-video and TWDM-PON,” in Proc. of ECOC 2013, P.6.19. Y. R. Shen, Principles of Nonlinear Optics (Wiley-Interscience, 1984). V. Curri, S. Capriata, and R. Gaudino, “Outage probability due to Stimulated Raman Scattering in GPON and TWDM-PON coexistence,” in Proc. OFC 2014, paper M3I.2. F. Coppinger, L. Chen, and D. Piehler, “Nonlinear Raman Cross-Talk in a Video Overlay Passive Optical Network,” in Proc. OFC 2003, paper TuR5. F. Tian, R. Hui, B. Colella, and D. Bowler, “Raman crosstalk in ber-optic hybrid CATV systems with wide channel separations,” IEEE Photon. Technol. Lett. 16(1), 344–346 (2004). H. Kim, S. B. Jun, and Y. C. Chung, “Raman crosstalk suppression in CATV overlay passive optical network,” IEEE Photon. Technol. Lett. 19(9), 695–697 (2007). N. Cheng and M. Zhou, Litvin, Kerry Effenberger, Frank, “Delay Modulation for TWDM PONs,” in Proc. OFC 2014, paper W1D.3. A. Tanaka, N. Cvijetic, and T. Wang, “Beyond 5dB Nonlinear Raman Crosstalk Reduction via PSD Control of 10Gb/s OOK in RF-Video Coexistence Scenarios for Next-Generation PON,” in Proc. OFC 2014, paper M3I.3. A. Shahpari, J. D. Reis, S. Ziaie, R. Ferreira, M. Lima, A. N. Pinto, and A. Teixeira, “Multi system NextGeneration PONs impact on Video Overlay,” in Proc. ECOC 2013, paper Tu.3.F.3. D. Piehler, “Minimising nonlinear Raman crosstalk in future network overlays on legacy passive optical networks,” Electron. Lett. 50(9), 687–688 (2014). A. Li, C. J. Mahon, Z. Wang, G. Jacobsen, and E. Bodtker, “Experimental conrmation of crosstalk due to stimulated Raman scattering in WDM AM-VSB CATV transmission systems,” Electron. Lett. 31(18), 1538– 1539 (1995).
#214357 - $15.00 USD Received 19 Jun 2014; revised 5 Aug 2014; accepted 9 Aug 2014; published 25 Aug 2014 (C) 2014 OSA 8 September 2014 | Vol. 22, No. 18 | DOI:10.1364/OE.22.021192 | OPTICS EXPRESS 21192
15. Q. Guo and A. V. Tran, “Combined utilization of partial-response coding and equalization for high-speed WDMPON with centralized lightwaves,” Opt. Express 20(27), 27981–27991 (2012). 16. H. Kim, K. H. Han, and Y. C. Chung, “Performance limitation of hybrid WDM systems due to Stimulated Raman Scattering,” IEEE Photon. Technol. Lett. 13(10), 1118–1120 (2001). 17. B. Colella, F. J. Effenberger, C. Shimer, and F. Tian, “Raman Crosstalk Control in Passive Optical Networks,” in Proc. OFC 2006, paper NWD6. 18. P. Kabal and S. Pasupathy, “Partial-response signaling,” IEEE. Trans. Commun. 23(9), 921–934 (1975). 19. M. Bi, S. Xiao, H. He, L. Yi, Z. Li, J. Li, X. Yang, and W. Hu, “Simultaneous DPSK demodulation and chirp management using delay interferometer in symmetric 40-Gb/s capability TWDM-PON system,” Opt. Express 21(14), 16528–16535 (2013). 20. ITU -T Recommendation G.975.1, 2004, Appendix I.9.
1. Introduction The time and wavelength division multiplexed passive optical network (TWDM-PON) has been selected by Full Service Access Network (FSAN) as a primary broadband solution for the next generation PON stage2 (NG-PON2) [1,2]. It combines the benefits of time division multiplexed PON (TDM-PON) and wavelength division multiplexed PON (WDM-PON), and can provide at least 40-Gb/s downstream and 10-Gb/s upstream aggregate capacity by stacking 10-gigabit-capable PONs (XG-PONs) via multiple pairs of wavelengths [3]. Meanwhile, in order to guarantee a full back compatibility of TWDM-PON with gigabitcapable PON (GPON), XG-PON and RF-video, the upstream and downstream wavelengths of TWDM-PON are in C- (1524 nm~1544 nm) and L + (1596 nm~1603 nm) bands, respectively [2]. However, on the full coexistence scenario of downstream wavelengths, the fiber nonlinearity may occur in the feeder fiber. Among all the fiber nonlinearities, the stimulated Raman scattering (SRS) is the most serious one [4]. As we all know that, SRS causes power transfer from a shorter wavelength to a longer wavelength. And the power transfer efficiency increases as the wavelength spacing increases until about 110 nm, which corresponds to the maximum SRS efficiency [5]. Hence, on the coexistence of GPON operating at 1490 nm and TWDM-PON operating at 1600 nm, GPON would act as a Raman pump and transfer its modulated power to the TWDM-PON channels. In this situation, 4-dB power depletion would be experienced on GPON for the case of 8 TWDM-PON wavelengths [4,6]. In addition to the power depletion on GPON, the nonlinear Raman crosstalk also exists between TWDM-PON and RF-video channel, which causes significant impairments to the video signals in the full coexistence scenario. In the previous PON systems, the crosstalk only exists between GPON operating at 1490 nm and RF-video operating at 1550 nm, which has already been investigated in the last few years [7–9]. Recently, due to the occurrence of 40Gb/s TWDM-PON, the nonlinear Raman crosstalk on RF-video from TWDM-PON wavelength channels is attracting new interests. Owing to the existence of multiple wavelength channels in TWDM-PON, this crosstalk is more serious than that caused by GPON wavelength. Huawei technology and NEC laboratory have studied this nonlinear Raman crosstalk between RF-video and TWDM-PON. By using Miller code [10] and simple RF filtering [11] techniques, 10-dB and 5-dB Raman crosstalk reduction were achieved respectively. The basic idea of Raman crosstalk reduction is to reduce the low-frequency components of digital modulation signals, because the nonlinear Raman crosstalk is strongly related to the power spectral density (PSD) of digital modulation signals, and the lower the low-frequency power is, the lower the Raman crosstalk would be [12–14]. In this paper, taking advantage of dicode-coding’s characteristic of ultra-low power spectral density in low frequency region [15], we utilize it in TWDM-PON system to mitigate the nonlinear Raman crosstalk on RF-video signal. By the experiment, 10 to 14-dB Raman crosstalk suppression is achieved at the launch power of 10 dBm to 15 dBm per channel in case of four TWDM-PON wavelength channels. Furthermore, the 10-Gb/s dicode transmission is also demonstrated over 20 km SSMF with the receiver sensitivity of −31 dBm at BER of 3.8e-3.
#214357 - $15.00 USD Received 19 Jun 2014; revised 5 Aug 2014; accepted 9 Aug 2014; published 25 Aug 2014 (C) 2014 OSA 8 September 2014 | Vol. 22, No. 18 | DOI:10.1364/OE.22.021192 | OPTICS EXPRESS 21193
2. Nonlinear Raman crosstalk on RF-video from TWDM-PON channels (i) … RF-video
RF-video 1550nm
TX1
TXN TWDM-PON
Crosstalk (dB)
TX2
-30
TWDM-PON λ 1600nm
(ii)
1λ 4λ 8λ
-40 -50
ONU
ONU
-60 -70 -80 0
ONU 50 100 150 Frequency (MHz)
200
Fig. 1. The considered downstream transmission characterized by the coexistence of RF-video and TWDM-PON. Inset (i): Downstream wavelengths; (ii): the theoretical nonlinear Raman crosstalk on RF-video (1550 nm) from multiple TWDM-PON wavelengths (1596.8 nm-1602.4 nm). WDM: wavelength division multiplexing; ONU: optical network unit
The considered coexistence system for downstream transmission is depicted in Fig. 1. The RF-video channel and TWDM-PON wavelength channels are multiplexed in the optical line terminal (OLT) before launched into the feeder fiber. The nonlinear Raman crosstalk on RFvideo from N TWDM-PON wavelength channels can be expressed as follows [16]. Induced Power on λRF − video Crosstalk ( dB ) = 10log Digital Modulation Signals Power on λTWDM (1) 2 −2 α L ρ P − 2e −α L cos ( 2π fd i L ) SRS TWDM mCATV N 2 1 + e =10log gi × 2 Aeff mTWDM i =1 α 2 + ( 2π fd i )
Where gi and di is the Raman gain coefficient and group velocity mismatch between RF-video and TWDM-PON wavelength channel i. The optical modulation index of CATV and TWDM-PON signal is denoted as mCATV and mTWDM respectively. Besides, the effective polarization overlap factor ρ SRS , the effective area Aeff, the launch power per channel PTWDM, the fiber length L and the attenuation coefficient of the fiber a are also included in Eq. (1). We calculated the nonlinear Raman crosstalk as a function of the RF-video frequency f for different number of wavelengths, which is shown in the inset (ii) of Fig. 1. In this calculation, PTWDM is 0 dBm and the fiber length L is 20 km. It is clearly seen that, the low frequency is affected more seriously than the high frequency owing to the low-pass characteristic of Raman crosstalk [17]. Moreover, when the number of TWDM-PON wavelength channel increases, the Raman crosstalk also increases seriously. For instance, the Raman crosstalk at 50 MHz is −57 dB at one channel case (1 λ) while reaches −52 dB for the case of four channels (4 λ), and finally increases to −50 dB as for 8 TWDM-PON channels (8 λ). The serious crosstalk will significantly influence the performance of RF-video signal. Therefore, it’s necessary to reduce this nonlinear Raman crosstalk in the coexistence system. 3. Proposed Raman crosstalk mitigation method and experimental setup Dicode coding is one of the partial-response signaling formats [18], which can alter the distribution of signal spectrum through controlled inter-symbol interference between adjacent symbols. Theoretically, it can be realized by a simple 1-bit delay and subtract operation, #214357 - $15.00 USD Received 19 Jun 2014; revised 5 Aug 2014; accepted 9 Aug 2014; published 25 Aug 2014 (C) 2014 OSA 8 September 2014 | Vol. 22, No. 18 | DOI:10.1364/OE.22.021192 | OPTICS EXPRESS 21194
namely the subtraction between the current bit and the previous bit. However, in order to alleviate the error propagation effect, differential code should be done before 1-bit delay and subtract operation. The differential code signal bn is calculated as follows. b n = mod ( a n + b n−1 )
(2)
Where mod in Eq. (2) is the modulo-2 operation and an is the original binary signal. After 1bit delay and subtract operation, dicode signal cn is shown as follows. cn = bn − bn −1
(3)
For instance, supposing an original binary signal sequence an “1 1 0 1 0 0 1 0”, then the generated differential code signal bn and dicode signal cn would be “1 0 0 1 1 1 0 0” and “1 −1 0 1 0 0 −1 0” respectively. In the decode operation, both level “1” and level “-1” correspond to binary “1” while level “0” corresponds to binary “0”, which means that each bit can be decoded independently without error propagation.
Fig. 2. Measured spectra of 10-Gb/s NRZ, Miller code and dicode signal.
As analyzed in [14], the Raman crosstalk is dependent on the power spectrum, which means that decreasing the power of a frequency point in digital modulation signals would decrease the crosstalk level at the same frequency. Therefore, we measured the spectra of 10Gb/s NRZ, Miller code and dicode signals, and the corresponding results are shown in Fig. 2. The value in the vertical axis is the digital modulation signals power on λTWDM before amplification. Compared with NRZ format, both Miller code and dicode have a reduced power at low frequency region. In the frequency of 50 MHz, the power is reduced by 10-dB with Miller code while 30-dB power reduction can be achieved with dicode coding, which is shown in Fig. 2. Meanwhile, dicode coding has a narrower bandwidth compared with NRZ and Miller code, hence it relaxes the bandwidth requirement for high data rate transmission. In order to verify the capability of nonlinear Raman crosstalk mitigation with dicode coding, we set up an experiment as depicted in Fig. 3. Firstly, four distributed feedback (DFB) lasers each followed by a polarization controller (PC) are multiplexed using a 100GHz arrayed waveguide grating (AWG). Because of the lack of L-band devices such as AWG and Erbium doped fiber amplifier (EDFA), we used four DFBs (λ1-λ4) operating at 1564.32 nm, 1565.12 nm, 1565.92 nm, 1566.72 nm instead of L-band ones. Note that the nonlinear Raman crosstalk is strongly related to the wavelength spacing rather than the wavebands [10,16], so the experimental results will not be impacted by the operating wavelengths as long as the wavelength spacing is properly set. As 50-nm wavelength spacing is required to evaluate the influence between TWDM-PON and RF-video, we set the RF-video wavelength (λ5) at 1510.56 nm in our experiment. External modulation instead of direct modulation is
#214357 - $15.00 USD Received 19 Jun 2014; revised 5 Aug 2014; accepted 9 Aug 2014; published 25 Aug 2014 (C) 2014 OSA 8 September 2014 | Vol. 22, No. 18 | DOI:10.1364/OE.22.021192 | OPTICS EXPRESS 21195
used to avoid the influence of chirp in directly modulated lasers (DML) [19] and allow us to determine the crosstalk due only to the SRS. Then, the multiplexed TWDM wavelengths are
Fig. 3. Experimental setup of co-existed downstream transmission between RF-video and TWDM-PON. Inset (i) dicode waveform; (ii) the optical spectrum of downstream wavelengths. AWG: arrayed waveguide grating; PC: polarization controller; TOF: tunable optical filter; MZM: Mach-Zehnder Modulator.
injected to a standard single-electrode Mach-Zehnder Modulator (JDSU OC-92) biased at the quadrature point. This MZM is driven by ~4-Vpp amplified dicode signals, which is generated by an arbitrary waveform generator (Tektronix AWG70001A) with the sampling rate of 50 GS/s. A part of waveform is shown in the inset (i) of Fig. 3. In the experiment, 40000 bits are transmitted to test its performance. The TWDM-PON wavelength channels are multiplexed with the RF-video using a 1:2 coupler, and then transmitted over 20-km of standard single mode fiber (SSMF). The optical spectra of all channels are shown in the inset (ii) of Fig. 3. The launch power per TWDM-PON channel is varied while the launch power of RF-video is fixed at 4.33 dBm. At the receiver, we use two tunable optical filters (TOF) to select these wavelengths. The signal on RF-video channel (λ5) is detected by a 3.5-GHz PIN + TIA, and then measured by a 30-GHz electrical spectrum analyzer with the resolution bandwidth (RBW) and video bandwidth (VBW) of 30 kHz. The measured signal can be considered as the nonlinear Raman crosstalk because the RF-video channel is not modulated by video signals. The dicode waveform detected by a 10-Gb/s APD + TIA is captured by an 80-GS/s real-time digital storage oscilloscope (DSO) for demodulation offline. 4. Experiment results and analysis
Fig. 4. The nonlinear Raman crosstalk with NRZ, Miller code and Dicode signals for the case of four TWDM-PON wavelength channels
#214357 - $15.00 USD Received 19 Jun 2014; revised 5 Aug 2014; accepted 9 Aug 2014; published 25 Aug 2014 (C) 2014 OSA 8 September 2014 | Vol. 22, No. 18 | DOI:10.1364/OE.22.021192 | OPTICS EXPRESS 21196
Firstly, we measured the Raman crosstalk mitigation effect of our proposed scheme. The results of NRZ format and Miller code are also depicted for comparison, which is shown in Fig. 4. The value in the vertical axis corresponds to the induced power on λRF-video because of TWDM wavelengths. In this measurement, four TWDM-PON wavelength channels are utilized and each channel is operated at the launch power of 13 dBm. It can be observed that the nonlinear Raman crosstalk of dicode modulation is lower than NRZ and Miller code in the frequency range of 50 MHz to 200 MHz. Especially for the crosstalk at 50 MHz, ~8 dB and ~11 dB suppression can be obtained with Miller code and dicode coding, respectively. As for the other frequencies, the nonlinear Raman crosstalk can almost be ignored for dicode modulation. Then we further investigated the proposed Raman crosstalk mitigation method by measuring the Raman crosstalk under different launch power cases. NRZ, Miller code and dicode formats were all included for comparison. The results are shown in Fig. 5(a) taking 50 MHz as example, since the Raman crosstalk is most severe on 50 MHz. It can be clearly seen that the power of crosstalk increases as the launch power per channel increases for NRZ signals. By using Miller code, the crosstalk was mitigated by 5~10 dB. As for the dicode modulation, the nonlinear Raman crosstalk is similar to the noise floor when the launch power is limited to 11.5 dBm, which means that almost no crosstalk is existed in this case. The noise floor here refers to the noise on the spectrum analyzer with all TWDM-PON channels off. When we further increased the launch power, the crosstalk increased gradually. Only 4-dB crosstalk increase is observed when the launch power per channel is 15 dBm, which is ~13 dB better than the NRZ case and ~3dB better than the Miller code solution. This is because that the spectrum density of dicode signal is much lower than that of NRZ and Miller code signals especially for the low frequency range. When dicode modulation is used in the system, less digital modulation signals power in TWDM wavelengths would be transferred to RF-video wavelength, resulting in smaller Raman crosstalk. Besides, relative intensity noise (RIN) is also an important factor in our system, which can increase the power of Raman crosstalk on RF-video wavelength. As demonstrated in ref [7], the RIN on TWDM channels is related by the bit rate, the extinction ratio and the launch power per channel. The higher the bit rate is, the more serious the RIN is. When SRS happens, the RIN on each TWDM channel would be superimposed on the RF video wavelength, resulting in more Raman crosstalk.
Fig. 5. (a) The nonlinear Raman crosstalk at 50 MHz versus the launch power per channel for NRZ, Miller code and Dicode modulation. (b) BER and eye diagrams of NRZ and dicode coding after 20-km SSMF transmission.
Finally, we measured the sensitivity and eye diagram of dicode signal. The measured BER curves and eye diagram of dicode signal in back to back (BtB) and 20-km SSMF transmission cases are shown in Fig. 5(b). The results of NRZ format are also measured and depicted. For dicode modulation, there is almost no difference on the receiver sensitivity between BtB and
#214357 - $15.00 USD Received 19 Jun 2014; revised 5 Aug 2014; accepted 9 Aug 2014; published 25 Aug 2014 (C) 2014 OSA 8 September 2014 | Vol. 22, No. 18 | DOI:10.1364/OE.22.021192 | OPTICS EXPRESS 21197
20-km transmission cases. The receiver sensitivity at forward error correction (FEC) BER limit of 3.8e-3 [20] for NRZ and dicode modulation is −36 dBm and −31 dBm, respectively. The 5-dB power penalty compared with NRZ modulation came from the lower noise tolerance of the three-level dicode signal. However, it can be further compensated by using the partial response equalization method in [15], which is not included in our demodulation algorithm in this paper. 5. Conclusions In this paper, we have investigated the Raman crosstalk on RF-video from TWDM-PON wavelength channels, and proposed a spectrum-reshaping technique based on dicode modulation to mitigate this crosstalk. This code realized through 1-bit delay and subtract operation has 30-dB less power than traditional NRZ modulation at 50 MHz. Utilizing this ultra-low power spectral density of dicode coding, 10 to 14 dB Raman crosstalk suppression has been achieved when the launch power per channel varied from 10 dBm to 15 dBm in case of four TWDM-PON wavelength channels. Furthermore, we find that the Raman crosstalk can almost be ignored when the launch power is limited at 10 dBm due to the property of dicode coding. Finally, the 10-Gb/s dicode transmission is also demonstrated through 20 km SSMF with the receiver sensitivity of −31 dBm at the BER of 3.8e-3. Acknowledgments The work was jointly supported by the NSFC (61271216, 61090393, 61221001 and 60972032), the 863 Program and the 973 Program (2010CB328205, 2010CB328204 and 2012CB315602).
#214357 - $15.00 USD Received 19 Jun 2014; revised 5 Aug 2014; accepted 9 Aug 2014; published 25 Aug 2014 (C) 2014 OSA 8 September 2014 | Vol. 22, No. 18 | DOI:10.1364/OE.22.021192 | OPTICS EXPRESS 21198
Abstract: In this paper, we investigate the nonlinear Raman crosstalk in RF-video overlay time and wavelength division multiplexed passive optical network (TWDM-PON), and propose a novel spectrum-reshaping method based on dicode coding to mitigate this crosstalk. The dicode coding features ultra-low power spectral density in the low frequency region, which can reduce the nonlinear Raman crosstalk on the RF-video signal effectively. Experimental results show that, compared with traditional nonreturn-to-zero on-off keying (NRZ-OOK) signals, the crosstalk on RF-video signal can be reduced by 10 ~14 dB when the launch power per TWDMPON channel varies from 10-dBm to 15-dBm. The transmission of 10-Gb/s dicode signal over 20-km standard single mode fiber (SSMF) is also demonstrated with the receiver sensitivity of −31 dBm at bit error ratio (BER) of 3.8e-3. ©2014 Optical Society of America OCIS codes: (060.2330) Fiber optics communications; (060.4250) Networks; (190.5650) Raman effect.
References and links 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.
ITU-T recommendation G.989.1, “40-Gigabit-capable passive optical networks (NG-PON2): General requirements,” 2013. ITU-T recommendation G.989.2(draft), “40-Gigabit-capable passive optical networks: Physical media dependent layer specification,” 2013. Y. Ma, Y. Qian, G. Peng, X. Zhou, X. Wang, J. Yu, Y. Luo, X. Yan, and F. Effenberger, “Demonstration of a 40Gb/s time and wavelength division multiplexed passive optical network prototype system,” in Proc. OFC 2012, paper PDP5D.7. R. Gaudino, V. Curri, and S. Capriata, “Propagation impairments due to Raman effect on the coexistence of GPON, XG-PON, RF-video and TWDM-PON,” in Proc. of ECOC 2013, P.6.19. Y. R. Shen, Principles of Nonlinear Optics (Wiley-Interscience, 1984). V. Curri, S. Capriata, and R. Gaudino, “Outage probability due to Stimulated Raman Scattering in GPON and TWDM-PON coexistence,” in Proc. OFC 2014, paper M3I.2. F. Coppinger, L. Chen, and D. Piehler, “Nonlinear Raman Cross-Talk in a Video Overlay Passive Optical Network,” in Proc. OFC 2003, paper TuR5. F. Tian, R. Hui, B. Colella, and D. Bowler, “Raman crosstalk in ber-optic hybrid CATV systems with wide channel separations,” IEEE Photon. Technol. Lett. 16(1), 344–346 (2004). H. Kim, S. B. Jun, and Y. C. Chung, “Raman crosstalk suppression in CATV overlay passive optical network,” IEEE Photon. Technol. Lett. 19(9), 695–697 (2007). N. Cheng and M. Zhou, Litvin, Kerry Effenberger, Frank, “Delay Modulation for TWDM PONs,” in Proc. OFC 2014, paper W1D.3. A. Tanaka, N. Cvijetic, and T. Wang, “Beyond 5dB Nonlinear Raman Crosstalk Reduction via PSD Control of 10Gb/s OOK in RF-Video Coexistence Scenarios for Next-Generation PON,” in Proc. OFC 2014, paper M3I.3. A. Shahpari, J. D. Reis, S. Ziaie, R. Ferreira, M. Lima, A. N. Pinto, and A. Teixeira, “Multi system NextGeneration PONs impact on Video Overlay,” in Proc. ECOC 2013, paper Tu.3.F.3. D. Piehler, “Minimising nonlinear Raman crosstalk in future network overlays on legacy passive optical networks,” Electron. Lett. 50(9), 687–688 (2014). A. Li, C. J. Mahon, Z. Wang, G. Jacobsen, and E. Bodtker, “Experimental conrmation of crosstalk due to stimulated Raman scattering in WDM AM-VSB CATV transmission systems,” Electron. Lett. 31(18), 1538– 1539 (1995).
#214357 - $15.00 USD Received 19 Jun 2014; revised 5 Aug 2014; accepted 9 Aug 2014; published 25 Aug 2014 (C) 2014 OSA 8 September 2014 | Vol. 22, No. 18 | DOI:10.1364/OE.22.021192 | OPTICS EXPRESS 21192
15. Q. Guo and A. V. Tran, “Combined utilization of partial-response coding and equalization for high-speed WDMPON with centralized lightwaves,” Opt. Express 20(27), 27981–27991 (2012). 16. H. Kim, K. H. Han, and Y. C. Chung, “Performance limitation of hybrid WDM systems due to Stimulated Raman Scattering,” IEEE Photon. Technol. Lett. 13(10), 1118–1120 (2001). 17. B. Colella, F. J. Effenberger, C. Shimer, and F. Tian, “Raman Crosstalk Control in Passive Optical Networks,” in Proc. OFC 2006, paper NWD6. 18. P. Kabal and S. Pasupathy, “Partial-response signaling,” IEEE. Trans. Commun. 23(9), 921–934 (1975). 19. M. Bi, S. Xiao, H. He, L. Yi, Z. Li, J. Li, X. Yang, and W. Hu, “Simultaneous DPSK demodulation and chirp management using delay interferometer in symmetric 40-Gb/s capability TWDM-PON system,” Opt. Express 21(14), 16528–16535 (2013). 20. ITU -T Recommendation G.975.1, 2004, Appendix I.9.
1. Introduction The time and wavelength division multiplexed passive optical network (TWDM-PON) has been selected by Full Service Access Network (FSAN) as a primary broadband solution for the next generation PON stage2 (NG-PON2) [1,2]. It combines the benefits of time division multiplexed PON (TDM-PON) and wavelength division multiplexed PON (WDM-PON), and can provide at least 40-Gb/s downstream and 10-Gb/s upstream aggregate capacity by stacking 10-gigabit-capable PONs (XG-PONs) via multiple pairs of wavelengths [3]. Meanwhile, in order to guarantee a full back compatibility of TWDM-PON with gigabitcapable PON (GPON), XG-PON and RF-video, the upstream and downstream wavelengths of TWDM-PON are in C- (1524 nm~1544 nm) and L + (1596 nm~1603 nm) bands, respectively [2]. However, on the full coexistence scenario of downstream wavelengths, the fiber nonlinearity may occur in the feeder fiber. Among all the fiber nonlinearities, the stimulated Raman scattering (SRS) is the most serious one [4]. As we all know that, SRS causes power transfer from a shorter wavelength to a longer wavelength. And the power transfer efficiency increases as the wavelength spacing increases until about 110 nm, which corresponds to the maximum SRS efficiency [5]. Hence, on the coexistence of GPON operating at 1490 nm and TWDM-PON operating at 1600 nm, GPON would act as a Raman pump and transfer its modulated power to the TWDM-PON channels. In this situation, 4-dB power depletion would be experienced on GPON for the case of 8 TWDM-PON wavelengths [4,6]. In addition to the power depletion on GPON, the nonlinear Raman crosstalk also exists between TWDM-PON and RF-video channel, which causes significant impairments to the video signals in the full coexistence scenario. In the previous PON systems, the crosstalk only exists between GPON operating at 1490 nm and RF-video operating at 1550 nm, which has already been investigated in the last few years [7–9]. Recently, due to the occurrence of 40Gb/s TWDM-PON, the nonlinear Raman crosstalk on RF-video from TWDM-PON wavelength channels is attracting new interests. Owing to the existence of multiple wavelength channels in TWDM-PON, this crosstalk is more serious than that caused by GPON wavelength. Huawei technology and NEC laboratory have studied this nonlinear Raman crosstalk between RF-video and TWDM-PON. By using Miller code [10] and simple RF filtering [11] techniques, 10-dB and 5-dB Raman crosstalk reduction were achieved respectively. The basic idea of Raman crosstalk reduction is to reduce the low-frequency components of digital modulation signals, because the nonlinear Raman crosstalk is strongly related to the power spectral density (PSD) of digital modulation signals, and the lower the low-frequency power is, the lower the Raman crosstalk would be [12–14]. In this paper, taking advantage of dicode-coding’s characteristic of ultra-low power spectral density in low frequency region [15], we utilize it in TWDM-PON system to mitigate the nonlinear Raman crosstalk on RF-video signal. By the experiment, 10 to 14-dB Raman crosstalk suppression is achieved at the launch power of 10 dBm to 15 dBm per channel in case of four TWDM-PON wavelength channels. Furthermore, the 10-Gb/s dicode transmission is also demonstrated over 20 km SSMF with the receiver sensitivity of −31 dBm at BER of 3.8e-3.
#214357 - $15.00 USD Received 19 Jun 2014; revised 5 Aug 2014; accepted 9 Aug 2014; published 25 Aug 2014 (C) 2014 OSA 8 September 2014 | Vol. 22, No. 18 | DOI:10.1364/OE.22.021192 | OPTICS EXPRESS 21193
2. Nonlinear Raman crosstalk on RF-video from TWDM-PON channels (i) … RF-video
RF-video 1550nm
TX1
TXN TWDM-PON
Crosstalk (dB)
TX2
-30
TWDM-PON λ 1600nm
(ii)
1λ 4λ 8λ
-40 -50
ONU
ONU
-60 -70 -80 0
ONU 50 100 150 Frequency (MHz)
200
Fig. 1. The considered downstream transmission characterized by the coexistence of RF-video and TWDM-PON. Inset (i): Downstream wavelengths; (ii): the theoretical nonlinear Raman crosstalk on RF-video (1550 nm) from multiple TWDM-PON wavelengths (1596.8 nm-1602.4 nm). WDM: wavelength division multiplexing; ONU: optical network unit
The considered coexistence system for downstream transmission is depicted in Fig. 1. The RF-video channel and TWDM-PON wavelength channels are multiplexed in the optical line terminal (OLT) before launched into the feeder fiber. The nonlinear Raman crosstalk on RFvideo from N TWDM-PON wavelength channels can be expressed as follows [16]. Induced Power on λRF − video Crosstalk ( dB ) = 10log Digital Modulation Signals Power on λTWDM (1) 2 −2 α L ρ P − 2e −α L cos ( 2π fd i L ) SRS TWDM mCATV N 2 1 + e =10log gi × 2 Aeff mTWDM i =1 α 2 + ( 2π fd i )
Where gi and di is the Raman gain coefficient and group velocity mismatch between RF-video and TWDM-PON wavelength channel i. The optical modulation index of CATV and TWDM-PON signal is denoted as mCATV and mTWDM respectively. Besides, the effective polarization overlap factor ρ SRS , the effective area Aeff, the launch power per channel PTWDM, the fiber length L and the attenuation coefficient of the fiber a are also included in Eq. (1). We calculated the nonlinear Raman crosstalk as a function of the RF-video frequency f for different number of wavelengths, which is shown in the inset (ii) of Fig. 1. In this calculation, PTWDM is 0 dBm and the fiber length L is 20 km. It is clearly seen that, the low frequency is affected more seriously than the high frequency owing to the low-pass characteristic of Raman crosstalk [17]. Moreover, when the number of TWDM-PON wavelength channel increases, the Raman crosstalk also increases seriously. For instance, the Raman crosstalk at 50 MHz is −57 dB at one channel case (1 λ) while reaches −52 dB for the case of four channels (4 λ), and finally increases to −50 dB as for 8 TWDM-PON channels (8 λ). The serious crosstalk will significantly influence the performance of RF-video signal. Therefore, it’s necessary to reduce this nonlinear Raman crosstalk in the coexistence system. 3. Proposed Raman crosstalk mitigation method and experimental setup Dicode coding is one of the partial-response signaling formats [18], which can alter the distribution of signal spectrum through controlled inter-symbol interference between adjacent symbols. Theoretically, it can be realized by a simple 1-bit delay and subtract operation, #214357 - $15.00 USD Received 19 Jun 2014; revised 5 Aug 2014; accepted 9 Aug 2014; published 25 Aug 2014 (C) 2014 OSA 8 September 2014 | Vol. 22, No. 18 | DOI:10.1364/OE.22.021192 | OPTICS EXPRESS 21194
namely the subtraction between the current bit and the previous bit. However, in order to alleviate the error propagation effect, differential code should be done before 1-bit delay and subtract operation. The differential code signal bn is calculated as follows. b n = mod ( a n + b n−1 )
(2)
Where mod in Eq. (2) is the modulo-2 operation and an is the original binary signal. After 1bit delay and subtract operation, dicode signal cn is shown as follows. cn = bn − bn −1
(3)
For instance, supposing an original binary signal sequence an “1 1 0 1 0 0 1 0”, then the generated differential code signal bn and dicode signal cn would be “1 0 0 1 1 1 0 0” and “1 −1 0 1 0 0 −1 0” respectively. In the decode operation, both level “1” and level “-1” correspond to binary “1” while level “0” corresponds to binary “0”, which means that each bit can be decoded independently without error propagation.
Fig. 2. Measured spectra of 10-Gb/s NRZ, Miller code and dicode signal.
As analyzed in [14], the Raman crosstalk is dependent on the power spectrum, which means that decreasing the power of a frequency point in digital modulation signals would decrease the crosstalk level at the same frequency. Therefore, we measured the spectra of 10Gb/s NRZ, Miller code and dicode signals, and the corresponding results are shown in Fig. 2. The value in the vertical axis is the digital modulation signals power on λTWDM before amplification. Compared with NRZ format, both Miller code and dicode have a reduced power at low frequency region. In the frequency of 50 MHz, the power is reduced by 10-dB with Miller code while 30-dB power reduction can be achieved with dicode coding, which is shown in Fig. 2. Meanwhile, dicode coding has a narrower bandwidth compared with NRZ and Miller code, hence it relaxes the bandwidth requirement for high data rate transmission. In order to verify the capability of nonlinear Raman crosstalk mitigation with dicode coding, we set up an experiment as depicted in Fig. 3. Firstly, four distributed feedback (DFB) lasers each followed by a polarization controller (PC) are multiplexed using a 100GHz arrayed waveguide grating (AWG). Because of the lack of L-band devices such as AWG and Erbium doped fiber amplifier (EDFA), we used four DFBs (λ1-λ4) operating at 1564.32 nm, 1565.12 nm, 1565.92 nm, 1566.72 nm instead of L-band ones. Note that the nonlinear Raman crosstalk is strongly related to the wavelength spacing rather than the wavebands [10,16], so the experimental results will not be impacted by the operating wavelengths as long as the wavelength spacing is properly set. As 50-nm wavelength spacing is required to evaluate the influence between TWDM-PON and RF-video, we set the RF-video wavelength (λ5) at 1510.56 nm in our experiment. External modulation instead of direct modulation is
#214357 - $15.00 USD Received 19 Jun 2014; revised 5 Aug 2014; accepted 9 Aug 2014; published 25 Aug 2014 (C) 2014 OSA 8 September 2014 | Vol. 22, No. 18 | DOI:10.1364/OE.22.021192 | OPTICS EXPRESS 21195
used to avoid the influence of chirp in directly modulated lasers (DML) [19] and allow us to determine the crosstalk due only to the SRS. Then, the multiplexed TWDM wavelengths are
Fig. 3. Experimental setup of co-existed downstream transmission between RF-video and TWDM-PON. Inset (i) dicode waveform; (ii) the optical spectrum of downstream wavelengths. AWG: arrayed waveguide grating; PC: polarization controller; TOF: tunable optical filter; MZM: Mach-Zehnder Modulator.
injected to a standard single-electrode Mach-Zehnder Modulator (JDSU OC-92) biased at the quadrature point. This MZM is driven by ~4-Vpp amplified dicode signals, which is generated by an arbitrary waveform generator (Tektronix AWG70001A) with the sampling rate of 50 GS/s. A part of waveform is shown in the inset (i) of Fig. 3. In the experiment, 40000 bits are transmitted to test its performance. The TWDM-PON wavelength channels are multiplexed with the RF-video using a 1:2 coupler, and then transmitted over 20-km of standard single mode fiber (SSMF). The optical spectra of all channels are shown in the inset (ii) of Fig. 3. The launch power per TWDM-PON channel is varied while the launch power of RF-video is fixed at 4.33 dBm. At the receiver, we use two tunable optical filters (TOF) to select these wavelengths. The signal on RF-video channel (λ5) is detected by a 3.5-GHz PIN + TIA, and then measured by a 30-GHz electrical spectrum analyzer with the resolution bandwidth (RBW) and video bandwidth (VBW) of 30 kHz. The measured signal can be considered as the nonlinear Raman crosstalk because the RF-video channel is not modulated by video signals. The dicode waveform detected by a 10-Gb/s APD + TIA is captured by an 80-GS/s real-time digital storage oscilloscope (DSO) for demodulation offline. 4. Experiment results and analysis
Fig. 4. The nonlinear Raman crosstalk with NRZ, Miller code and Dicode signals for the case of four TWDM-PON wavelength channels
#214357 - $15.00 USD Received 19 Jun 2014; revised 5 Aug 2014; accepted 9 Aug 2014; published 25 Aug 2014 (C) 2014 OSA 8 September 2014 | Vol. 22, No. 18 | DOI:10.1364/OE.22.021192 | OPTICS EXPRESS 21196
Firstly, we measured the Raman crosstalk mitigation effect of our proposed scheme. The results of NRZ format and Miller code are also depicted for comparison, which is shown in Fig. 4. The value in the vertical axis corresponds to the induced power on λRF-video because of TWDM wavelengths. In this measurement, four TWDM-PON wavelength channels are utilized and each channel is operated at the launch power of 13 dBm. It can be observed that the nonlinear Raman crosstalk of dicode modulation is lower than NRZ and Miller code in the frequency range of 50 MHz to 200 MHz. Especially for the crosstalk at 50 MHz, ~8 dB and ~11 dB suppression can be obtained with Miller code and dicode coding, respectively. As for the other frequencies, the nonlinear Raman crosstalk can almost be ignored for dicode modulation. Then we further investigated the proposed Raman crosstalk mitigation method by measuring the Raman crosstalk under different launch power cases. NRZ, Miller code and dicode formats were all included for comparison. The results are shown in Fig. 5(a) taking 50 MHz as example, since the Raman crosstalk is most severe on 50 MHz. It can be clearly seen that the power of crosstalk increases as the launch power per channel increases for NRZ signals. By using Miller code, the crosstalk was mitigated by 5~10 dB. As for the dicode modulation, the nonlinear Raman crosstalk is similar to the noise floor when the launch power is limited to 11.5 dBm, which means that almost no crosstalk is existed in this case. The noise floor here refers to the noise on the spectrum analyzer with all TWDM-PON channels off. When we further increased the launch power, the crosstalk increased gradually. Only 4-dB crosstalk increase is observed when the launch power per channel is 15 dBm, which is ~13 dB better than the NRZ case and ~3dB better than the Miller code solution. This is because that the spectrum density of dicode signal is much lower than that of NRZ and Miller code signals especially for the low frequency range. When dicode modulation is used in the system, less digital modulation signals power in TWDM wavelengths would be transferred to RF-video wavelength, resulting in smaller Raman crosstalk. Besides, relative intensity noise (RIN) is also an important factor in our system, which can increase the power of Raman crosstalk on RF-video wavelength. As demonstrated in ref [7], the RIN on TWDM channels is related by the bit rate, the extinction ratio and the launch power per channel. The higher the bit rate is, the more serious the RIN is. When SRS happens, the RIN on each TWDM channel would be superimposed on the RF video wavelength, resulting in more Raman crosstalk.
Fig. 5. (a) The nonlinear Raman crosstalk at 50 MHz versus the launch power per channel for NRZ, Miller code and Dicode modulation. (b) BER and eye diagrams of NRZ and dicode coding after 20-km SSMF transmission.
Finally, we measured the sensitivity and eye diagram of dicode signal. The measured BER curves and eye diagram of dicode signal in back to back (BtB) and 20-km SSMF transmission cases are shown in Fig. 5(b). The results of NRZ format are also measured and depicted. For dicode modulation, there is almost no difference on the receiver sensitivity between BtB and
#214357 - $15.00 USD Received 19 Jun 2014; revised 5 Aug 2014; accepted 9 Aug 2014; published 25 Aug 2014 (C) 2014 OSA 8 September 2014 | Vol. 22, No. 18 | DOI:10.1364/OE.22.021192 | OPTICS EXPRESS 21197
20-km transmission cases. The receiver sensitivity at forward error correction (FEC) BER limit of 3.8e-3 [20] for NRZ and dicode modulation is −36 dBm and −31 dBm, respectively. The 5-dB power penalty compared with NRZ modulation came from the lower noise tolerance of the three-level dicode signal. However, it can be further compensated by using the partial response equalization method in [15], which is not included in our demodulation algorithm in this paper. 5. Conclusions In this paper, we have investigated the Raman crosstalk on RF-video from TWDM-PON wavelength channels, and proposed a spectrum-reshaping technique based on dicode modulation to mitigate this crosstalk. This code realized through 1-bit delay and subtract operation has 30-dB less power than traditional NRZ modulation at 50 MHz. Utilizing this ultra-low power spectral density of dicode coding, 10 to 14 dB Raman crosstalk suppression has been achieved when the launch power per channel varied from 10 dBm to 15 dBm in case of four TWDM-PON wavelength channels. Furthermore, we find that the Raman crosstalk can almost be ignored when the launch power is limited at 10 dBm due to the property of dicode coding. Finally, the 10-Gb/s dicode transmission is also demonstrated through 20 km SSMF with the receiver sensitivity of −31 dBm at the BER of 3.8e-3. Acknowledgments The work was jointly supported by the NSFC (61271216, 61090393, 61221001 and 60972032), the 863 Program and the 973 Program (2010CB328205, 2010CB328204 and 2012CB315602).
#214357 - $15.00 USD Received 19 Jun 2014; revised 5 Aug 2014; accepted 9 Aug 2014; published 25 Aug 2014 (C) 2014 OSA 8 September 2014 | Vol. 22, No. 18 | DOI:10.1364/OE.22.021192 | OPTICS EXPRESS 21198
