Single wavelength 50 Gb/s passive optical network (PON) is an excellent candidate for meeting high capacity requirements. In this paper, we experimentally investigate a symmetrical 50 Gb/s time division multiplexed passive optical network (TDM-PON) system in the O-band based on 25G optics. Semiconductor optical amplifier (SOA) is used in optical line terminal (OLT) side to improve link power budget. We initially investigate the performances of SOA as a booster amplifier with different gain in the downstream and make a trade-off between receiver sensitivity and power budget. The performances of 50 Gb/s non-return-to-zero (NRZ) in the downstream with avalanche photodiode (APD) receiver and upstream with SOA-PIN receiver with different equalization schemes are evaluated. Experimental results show that up to 34.97 dB link power budget is achieved in the downstream direction with 7-tap feed forward equalization (FFE), and 33.76 dB link power budget is achieved in the upstream direction with only 3-tap FFE filtering.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
In view of the developing of the fifth generation (5G) mobile fronthaul, 4k/8k video streaming services and cloud computing, the capacity of optical networks has been boosting. In the early years, the institute of electrical and electronics engineers (IEEE) 802.3 Working Group  aimed to achieve 100G-EPON by means of multiplexing four wavelengths with 25 Gb/s per wavelength data rates to meet high capacity requirements. But from the perspective of large-scale deployment, going from 10 Gb/s per wavelength data rates to 25 Gb/s is a too small step as an upgrade . Lately International Telecommunications Union Telecommunication Standardization Sector (ITU-T)  Study Group devoted themselves to the development of single wavelength 50 Gb/s technology in the O-band . This proposal not only reduces the number of optical devices but also saves half of the wavelength resources compared with 25 Gb/s per wavelength scheme, consequently reducing the difficulty of wavelength management . To satisfy the data rate requirements, more advanced modulation formats emerge as the times require, including carrier-less amplitude and phase (CAP), discrete multi-tone (DMT) and 4-level pulse-amplitude modulations (PAM-4) with higher spectra efficiency [6–8]. However, these modulation schemes always require a high linearity of optical devices, have inferior receiver sensitivity and maybe demand complex transmitter and receiver [9,10]. NRZ modulation format is still a good candidate in consideration of its single threshold detecting-receiver , technical maturity and simple equalization technology. But beyond all that, power budget will be a serious task for single wavelength 50 Gb/s transmission system. SOA has attracted wide attention for its advantages of small size, low cost and easy integration. Particularly it has the ability to amplify signals of various wavelengths. A symmetrical experiment of 50 Gb/s PAM-4 TDM-PON based on 25G optics in the O-band is achieved in . It can support the PR-30 link loss budget with the use of SOA and digital signal processing (DSP) of 47-tap feed-forward equalization (FFE) and 11-tap decision feedback equalization (DFE). A SOA-PIN/TIA receiver module is demonstrated in , with a sensitivity at 25 Gb/s and 40 Gb/s NRZ of −23 dBm and −21 dBm respectively based on 10G optics. Minghui Tao et al. accomplished the experiment of 50-Gb/s NRZ signal based on 40 GHz-bandwidth EML and APD with a 3-dB bandwidth of about 18 GHz at 1550 nm. A receiver sensitivity of about −23.8 dBm is obtained with the use of receiver-side equalizer based on FFE, followed by a maximum-likelihood sequence estimation (MLSE) decoder . And 50 Gb/s/λ PAM-4 transmission is demonstrated over 20 km SSMF using a 10G-class transmitter with 21dB loss budget in C-band with the help of SVM-modified-FFE in . However, compared with multilevel formats, binary format (NRZ) can achieve higher output power provided by SOA . And based on our investigation to date, per-wavelength data rates at 50 Gb/s TDM-PON with SOA is not fully investigated. In order to meet the requirements of low cost and high power budget, we investigate the performance of 50 Gb/s NRZ with the assistance of SOA and DSP technology.
A symmetrical 50 Gb/s NRZ experimental scheme in the O-band is demonstrated in this paper, which can support over 33 dB power budget with the help of SOA and simple equalization in both transmission directions. We initially investigate the performances of SOA as a booster amplifier with different gain in the downstream. There is a trade-off between receiver sensitivity and launch power. As a result, 11.77 dBm launch power is selected. In the experimental scheme, all SOAs are employed at the OLT side. This approach enables all optical network unit (ONU) users to share the cost of SOA compared to SOA at the ONU side. Consequently, the cost of the system is reduced. Afterwards, the performances of 50 Gb/s NRZ in the downstream transmission with 25G APD receiver and upstream transmission with SOA-PIN receiver is discussed. Experimental results show that with up to 34.97 dB link power budget is achieved in the downstream direction with 7-tap feed forward equalization (FFE) filter and 33.76 dB link power budget is achieved in the upstream direction with 3-tap FFE filter only. Accordingly, we believe that the proposed experimental scheme will make an important contribution to the next-generation 50 Gb/s PON system.
2. Experimental setup
Figure 1 depicts the experimental setup for symmetrical 50 Gb/s TDM-PON transmission based on 25G class optics. The downstream and upstream signal is separated by an optical circulator (OC) in the experimental scheme. In the downstream link, the NRZ signal with 50 Gb/s data rate is generated by multiplexing two 25 Gb/s pseudo-random binary sequence (PRBS) with a word length of obtained from a pulse pattern generator (PPG, Anritsu MP 1800A). And the signal is directly modulated by a 25G electro-absorption modulated laser (EML) with the center wavelength of 1310 nm. The output power of the EML is around 3.26 dBm. As depicted in Fig. 1, SOA-1 in OLT acts as a power booster to increase the laser output power, which has a saturation output power of 11.8 dBm, small-signal gain of 12.7 dB and noise figure of 7.1 dB. In order to obtain the optimal input optical power of SOA, we add an attenuator to control power injected into SOA. Then an isolator is placed behind the SOA to prevent light reflection. At receiver side, a variable optical attenuator (VOA) and a LAN-WDM optical filter with a 3 dB pass-band of 3.87 nm are placed after 25-km standard single mode fiber (SSMF) to adjust the received optical power and to mitigate the out-of-band noise, respectively. Then the signal is detected by a 25G avalanche photodiode (APD) integrated with linear TIA. And finally the output electrical signal is sampled by digital signal analyzer (DSA, Key sight DSAZ592A) with 80 GS/s sampling rate and 33 GHz bandwidth. The frequency response of the transceivers in the downstream is shown as inset (b) in Fig. 1. The 3 dB bandwidth of the whole system is 17 GHz. For the upstream link, the signal generation and the EML transmitter is the same as in the downstream direction. SOA-2 is placed at the receiver side for pre-amplification. It has a saturation output power of 14.2 dBm, small-signal gain of 30 dB and noise figure of 7.4 dB under 300 mA driving current. The receiver is a 35G PIN. The frequency response of the transceivers in the upstream is shown as inset (c). The 3 dB bandwidth of upstream link is 19 GHz. Offline DSP including timing recovery, re-sampling, equalization and finally bit-error-ratio (BER) calculation is accomplished in both directions for demodulation as shown in Fig. 1 (a). Note that the SOA for downstream power boosting and upstream signal pre-amplification both locates in OLT side. This approach enables all optical network unit (ONU) users to share the cost of SOA compared to SOA at the ONU side. As a result, the cost of the system is reduced.
3. 50 Gb/s NRZ downstream performance with linear APD receiver
Before conducting a downlink experiment, we first studied the optimal input power of SOA-1 and made a trade-off between the receiver sensitivity and launch power to maximize the power budget. Firstly, we fixed the input power of SOA at 2 dBm to get 11.77 dBm output power. Then we varied the power launched into 25-km fiber from 11.77 dBm to 5 dBm using VOA2 and measured the BERs under −23 dBm received power. As shown in Fig. 2 (a), similar sensitivities are obtained within 5~11 dBm launch power range, which means that no fiber nonlinearity introduced penalty is observed under 11 dBm launch power. Then we adjusted the power launched into SOA and measured the sensitivity variation. The best receiver sensitivity of −25.78 dBm is obtained with −2 dBm SOA’s input power, when the output power is only 9.44 dBm. Therefore, for maximizing the power budget, a tradeoff is needed. Table 1 lists the sensitivities under different launch power, and the maximal power budget of 36.74 dB is achieved under 11.77 dBm launch power when the power sent into SOA is 2 dBm. Therefore, in the following experiments, we set the launch power at 11.77 dBm in the downstream direction.
In the development of PON system, the wavelength coexistence is essential and optical filters are indispensable to avoid interference between close wavelengths. In order to verify the impact of the LAN-WDM on the system, we compared the performance of the system with or without LAN-WDM as shown in Fig. 3. The performance with optical filter is improved by 0.6 dB compared to the situation without optical filter.
Then we analyzed the equalization performance of FFE and DFE filtering with the different number of taps, as shown in Fig. 4. For 25-km SSMF transmission, we range the FFE tap number from 3 to 13 to evaluate the equalization performance. As a consequence, −23.2 dBm sensitivity is achieved at the LDPC limit by using 7-tap FFE . Further increasing the tap number of FFE makes negligible improvement. The performance of DFE filter is also evaluated. As depicted in Fig. 4 (b), 3-tap FFE and 1-tap DFE provides −23.75 dBm sensitivity, which is ~0.55 dB better than 7-tap FFE. No significant improvement is achieved when more taps are used.
In addition, the performance of MLSE is evaluated and compared with the FFE and DFE filtering. As Fig. 4 (c) describes, −25 dBm sensitivity is obtained using MLSE, which is 1 dB better than DFE and 2 dB higher than FFE case. As our experiment implements in the O band, the chromatic dispersion (CD) induced penalty is avoided, and the sensitivity measured after 25-km fiber transmission is similar with back-to-back (B2B) case. Figure 4 (d) shows the impact brought by the SOA booster, the sensitivity is ~1 dB lower by using SOA for all equalization methods. Considering the power boosting effect, the overall power budget is increased by 6.41 dB using SOA in downlink.
According to the standards established by ITU-T, the downstream standard wavelength of 50G-PON is 1342 nm. Our experiment is based on the 1310 nm wavelength in the downstream due to the lack of EML with the center wavelength of 1342 nm. To evaluate the downstream dispersion penalty, EML is replaced by a tunable light source and a Mach-Zehnder (MZ) modulator operated at the O-band. In the experiment, we set the tunable light source to 1342 nm wavelength as the input source of the MZ modulator. The sensitivity curves and the eye diagrams of the received signal in B2B and 25 km SSMF transmission cases are depicted in Fig. 5. It can be concluded that the dispersion penalty of the 1342 nm wavelength after 25-km SSMF transmission is 0.2 dB. Without considering the chirp and linewidth of the laser, the dispersion penalty between 1310 nm and 1342 nm wavelength is very small in 50G-PON. Further study is needed to quantify the dispersion penalty when the laser chirp and the worst-case fiber dispersion at 1342 nm are taken into consideration.
4. 50 Gb/s NRZ upstream performance with SOA + PIN receiver
Similarly, we verified the performance of 50 Gb/s NRZ with SOA pre-amplification and PIN receiver in the upstream link, and the results as shown in Fig. 6. −30.5 dBm receiver sensitivity can be achieved by using 3-tap FFE. Adding one feedback tap improves the sensitivity to −30.8 dBm, as illustrated in Fig. 6 (b). MLSE provides the highest sensitivity of −31.8 dBm. And electrical eye diagram comparison in three cases after 25-km SSMF transmission is displayed in Fig. 6 (d). Inset (i) and (ii) in the Fig. 6 (d) show the electrical eye diagram received by 25G APD in the downstream without and with SOA at the −18 dBm received optical power. And inset (iii) shows the electrical eye diagram received by 35G PIN in the upstream with SOA at the −25 dBm received optical power. The signal is converted into duo-binary format due to the bandwidth limitation of the transceiver, which can be seen from the illustration. From inset (i) and (ii), we can see that after introducing SOA, eye diagram has deteriorated. In addition, the corresponding extinction ratio was calculated. With the introduction of SOA, the extinction ratio is reduced by 0.22 dB. The specific values are shown in Table 2. From the above, we can conclude that SOA introduced nonlinearity and degraded the extinction ratio. Therefore, the off-line DSP module is indispensable.
In the PON system, upstream is operated in burst mode. In our experiments as above, only one ONU was considered. A VOA is placed after 25-km SSMF to adjust the input power of SOA and the sensitivity curves can be obtained. Then another VOA is used to guarantee that the PIN works below saturation power. When applied to the burst mode in the actual situation, the signal power transmitted by different ONUs is different and the received power of the PIN after SOA is not fixed. Therefore, the gain control of the SOA is required to ensure that the signal with power in a certain dynamic range can be normally received . Apart from this, the gain control of the SOA is also essential in order to ensure that the PIN works properly.
We have experimentally demonstrated a symmetrical 50 Gb/s TDM-PON system in the O-band with up to 34.97 dB link power budget with the help of SOA and FFE filter. We initially investigated the performances of different gain of SOA as a booster amplifier and made a tradeoff between launch power and receiving sensitivity. Hereafter, we discussed the performances of 50 Gb/s NRZ in the downstream with APD receiver and upstream with SOA-PIN receiver. The link loss budget is increased by 6.41 dB and 5.2 dB in the downlink and uplink respectively by using SOA. 7-tap FFE can achieve 34.97 dB link power budget in the downlink and 3-tap FFE can achieve 33.76 dB link power budget in the upstream as summarized in Table 3, which will make an important contribution to the next-generation 50 Gb/s PON system.
National Natural Science Foundation of China (NSFC) (Project No. 61420106011, 61601277, 61601279); Shanghai Science and Technology Development Funds (Project No. 17010500400, 18511103400, 16YF1403900).
1. IEEE P802.3ca 100G-EPON Task Force, http://www.ieee802.org/3/ca/
2. V. Houtsma, D. van Veen, and R. Bonk, “Options for Single wavelength 50G TDM-PON,” in IEEE 802.3ca, Sep. Meeting (IEEE, 2017), paper houtsma_3ca_1_0917.
3. G.989 series of ITU specifications, https://www.itu.int/rec/T-REC-G/en.
4. G.hsp.50Gpmd ITU-T work programme, https://www.itu.int/ITUT/workprog/wp_item.aspx?isn=14550.
5. E. Harstead and J. Johnson, “25G and 50G EPON downstream wavelength plan,” in IEEE 802.3ca, March Meeting (IEEE, 2018), paper harstead_3ca_1b_0318.
6. C.-C. Wei, K.-Z. Chen, L.-W. Chen, C.-Y. Lin, W.-J. Huang, and J. Chen, “High-capacity carrierless amplitude and phase modulation for WDM long-reach PON featuring high loss budget,” J. Lightwave Technol. 35(4), 1075–1082 (2017). [CrossRef]
7. X. Xu, E. Zhou, G. N. Liu, T. Zuo, Q. Zhong, L. Zhang, Y. Bao, X. Zhang, J. Li, and Z. Li, “Advanced modulation formats for 400-Gbps short-reach optical inter-connection,” Opt. Express 23(1), 492–500 (2015). [CrossRef] [PubMed]
8. T. Tanaka, M. Nishihara, T. Takahara, W. Yan, L. Li, Z. Tao, M. Matsuda, K. Takabayashi, and J. Rasmussen, “Experimental demonstration of 448-Gbps+ DMT transmission over 30-km SMF,” in Optical Fiber Communication Conference (OFC) (2017), paper M2I. 5.
9. D. van Veen and V. Houtsma, “High Speed TDM-PON Beyond 10G,” in Optical Fiber Communication Conference (OFC) (2016), paper Tu3C. 3. [CrossRef]
10. D. van Veen and V. Houtsma, “Symmetrical 25-Gb/s TDM-PON with 31.5-dB Optical Power Budget Using Only Off-the-Shelf 10-Gb/s Optical Components,” J. Lightwave Technol. 34(7), 1636–1642 (2016). [CrossRef]
11. J. Zhang, J. S. Wey, J. Yu, Z. Tu, B. Yang, W. Yang, Y. Guo, X. Huang, and Z. Ma, “Symmetrical 50-Gb/s/λ PAM-4 TDM-PON in O-band with DSP and Semiconductor Optical Amplifier Supporting PR-30 Link Loss Budget,” in Optical Fiber Communication Conference (OFC) (2018), paper M1B.4. [CrossRef]
12. C. Caillaud, P. Chanclou, F. Blache, P. Angelini, B. Duval, P. Charbonnier, D. Lanteri, G. Glastre, and M. Achouche, “Integrated SOA-PIN Detector for High-Speed Short Reach Applications,” J. Lightwave Technol. 33(8), 1596–1600 (2015). [CrossRef]
13. M. Tao, J. Zhang, X. Dong, K. Zhang, L. Zhou, H. Zeng, Y. Luo, S. Li, and X. Liu, “Improved Dispersion Tolerance for 50G-PON Downstream Transmission via Receiver-Side Equalization,” in Optical Fiber Communication Conference (OFC) (2019), paper M2B.3. [CrossRef]
14. X. Miao, M. Bi, J. Yu, L. Li, and W. Hu, “SVM-Modified-FFE Enabled Chirp Management for 10G DML-based 50Gb/s/λ PAM4 IM-DD PON,” in Optical Fiber Communication Conference (OFC) (2019), paper M2B.5. [CrossRef]
15. R. Bonk, “SOA for future PONs,” in Optical Fiber Communication Conference (OFC) (2018), paper Tu2B.4. [CrossRef]
16. M. Laubach, S. Yang, R. Hirth, and G. Kramer, “FEC Proposal (LDPC) for NGEPON,” in IEEE 802.3ca, May Meeting (IEEE, 2017), paper laubach_3ca_1_0517.
17. D. Liu and T. Yang, “Analysis of Dynamic Range in 100G EPON with SOA as Preamplifier,” in IEEE 802.3ca, Jan. Meeting (IEEE, 2018), paper liu_3ca_3_0118.