Open Access
Add to BibSonomy
Abstract
A reconfigurable passive optical network (PON) downlink has been experimentally demonstrated. A multi-carrier signal is generated in the digital domain and converted to an optical signal by a Mach-Zehnder intensity modulator. A wavelength selective switch (WSS) demultiplexes the sub-carriers for a distribution network. Each sub-carrier signal is a duo-binary filtered binary phase shift keying and directly detected at the receiver side. By changing the sub-carrier allocation and WSS port setting, the PON network can be easily reconfigured without changing the hardware configuration. The maximum capacity of 75-Gb/s with 10-km transmission and 128-way split has been achieved.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
The line-rate of passive optical network (PON) reaches 50-Gb/s in the latest ITU-T recommendation G.9804, to meet with the growing demand from individual users and versatile applications [1]. To fully appreciate the potential of the hardware, digital signal processing (DSP)-based flexible PON is intensively investigated [2,3]. In the DSP-based flexible PON system, the modulation format and forward error correction (FEC) code are optimized for individual users according to the splitting loss and transmission distance, and the total performance is maximized. On the other hand, hardware-level flexibility is also an important feature for future PON systems where versatile application needs to be accommodated by reconfiguring the network depending on the demand, and reconfigurable photonic integrated circuit has been developed [4]. A typical situation that requires network reconfiguration is a temporal big event where many participants require huge traffic, such as Olympic games. In such case, the data traffic to the site will temporally become huge. To accommodate this huge traffic, new branch and large bandwidth need to be assigned. On the other hand, when the event is over, the branch needs to be closed and the network is reconfigured to be normal state. In general, a reconfiguration of PON needs hardware change and therefore requires time and cost. So, a reconfigurable PON system that requires no hardware change is expected. In such a flexible PON system, a multi-carrier transmission is suitable for dynamic bandwidth and port allocation [5–8], because the port and bandwidth assignment can be easily changed by controlling the sub-carrier setting. On the other hand, flexible multi-carrier PON system usually requires multiple transmitters with different wavelengths to accommodate large capacity traffic [5,6], which is not cost effective. Digital subcarrier multiplexing (DSM) is an attractive approach to generate multi-carrier signal by a single transmitter, although it requires expensive IQ modulator to generate fully independent data streams on sub-carriers, even for on-off keying (OOK) [8–10]. In addition, although a coherent optical transceiver can achieve highly flexible point to multipoint (P2MP) connections without optical wavelength filtering, an intensity modulation and direct detection (IM-DD) is still attractive for a PON system where the cost is always primary concern [11].
In this paper, we experimentally demonstrate reconfigurable PON downlink based on a single Mach-Zehnder intensity modulator (MZ-IM) and a wavelength selective switch (WSS). A multi-carrier signal is generated in digital domain and converted to analog optical signal by a high-speed digital-to-analog converter (DAC) and an MZ-IM which is more cost effective than IQ modulator, and the carriers are separated into several branches by a WSS. As typically illustrated in the inset spectra i) and ii) of Fig. 3, the network can be easily reconfigured by changing the carrier frequency allocation and WSS port assignment. In this demonstration, owing to high bandwidth of intensity modulator and DAC, as well as high resolution of WSS, 4 × 12.5-Gb/s signal with 5-GHz guard-band has been successfully distributed after transmitted over 10-km single mode fiber (SMF), and the system is reconfigured to 50-Gb/s + 2 × 12.5-Gb/s. To justify the insertion loss of WSS, we chose the low-loss C-band wavelength. In addition, multi-carrier transmission with low baud-rate is adopted, rather than high baud-rate single-carrier, to avoid the chromatic dispersion (CD) penalty. The reconfigurability has been successfully demonstrated, and the downlink transmission of 10-km, 128-way split has been achieved with maximum total capacity of 75-Gb/s.
2. Operation principle and experimental setup
Figure 1 illustrates the principle of multi-carrier signal generation by a single intensity modulator. The base-band data signal of n th (n = 1, 2, …, N) stream is multiplied by cos(2πfnt), where fn is the sub-carrier frequency, and 1-N data signals are combined in digital domain. The digitally multiplexed signal is converted to analog waveform by a DAC to drive an optical intensity modulator. A typical optical spectrum of sub-carrier multiplexed signal is shown in Fig. 1. The sub-carrier multiplexed signal is transmitted over trunk-span fiber and a WSS demultiplexes sub-carriers for optical distribution network (ODN). The demultiplexed sub-carrier signals are directly detected at the end-user optical network units (ONUs). It should be noted that the sub-carrier frequencies fn and −fn (n = 1, 2, …, N) are carrying the same data and the corresponding branches are shared via time domain multiplexing (TDM). The modulation format of each carrier is duo-binary filtered binary phase shift keying (BPSK) with Nyquist pulse shaping for bandwidth saving, and therefore the data sequence is differentially encoded at the transmitter side. Figure 2 illustrates the process of signal generation. Since the impulse response of duo-binary filter is [1,1], BPSK signal sequences of [0, π] and [π, 0] make destructive interference, whereas [0, 0] and [π, π] make constructive interference, and the 3-dB bandwidth become half of the original Nyquist signal which leads to low CD penalty. By receiving the signal with single photodetector (PD), the signal is converted to simple on-off keying (OOK) signal.
Fig. 3. Experimental setup for reconfigurable PON downlink. Insets: wavelength assignment of i) before and ii) after reconfiguration.
Figure 3 shows the experimental setup. A multi-carrier signal waveform is generated by MATLAB offline processing, and an arbitrary waveform generator (AWG) with 256-GSa/s converts the digital data to analog waveform to drive the single intensity modulator (dual-drive MZM). The bandwidths of the DAC and MZM are respectively 70- and 50-GHz, and the total frequency response of the transmitter is compensated in digital domain. The DAC has inverted and non-inverted output, and they are directly connected to two input ports of dual-drive MZM, that is biased at null-output point, to generate a real-value optical waveform. The Vπ (π-shift voltage for single input port) and the extinction ratio of dual-drive MZM are respectively 1.9-V and 30-dB, and the amplitude of the applied waveform is 0.85-Vpp. It means that the dual-drive MZM is operating in linear-response region, and the waveform from DAC is accurately converted to the real-value optical waveform. The frequency of laser light source is set at 193.5-THz. The generated multi-carrier optical signal is transmitted over 10-km SMF that has 2.2-dB insertion loss including the connectors and separated into several branches by a WSS. By reconfiguring the sub-carrier frequency allocation and WSS port assignment, the PON downlink network can be reconfigured. The receiver side is the single photodetector integrated with transimpedance amplifier (TIA), and the 3-dB bandwidth is about 30-GHz. The received signal is digitized by a real-time oscilloscope with 160-GSa/s and re-sampled to 2-Sa/s, and offline digital equalization filter is applied. For high baud-rate signal, Volterra nonlinear equalizer (VNLE) is used to compensate for CD [12], whereas linear feed forward equalizer is enough for low baud-rate signal. After digital equalization, bit-error-rate (BER) has been measured to evaluate the system performance.
3. Experimental results
Figure 4 shows the spectra and measured BERs of 4-sub-carriers x 12.5-Gb/s signal, i.e., the total capacity of the system is 50-Gb/s, which is assumed as the normal configuration of this system (in the case of no extra bandwidth request). The frequency in Fig. 4 indicates the difference from that of laser light source. Four sub-carrier frequencies are allocated with 5-GHz guard-band, and therefore the sub-carrier frequencies are 17.5-, 35.0-, and 52.5-GHz. Due to intensity modulation, negative carrier frequency copy is generated, and it is also used for branching. The powers of sub-carriers are equalized by compensating for the frequency response of MZM in the process of offline waveform generation. The center carrier (baseband) is intendedly given 3-dB higher power than the other sub-carriers, because the data stream for center carrier has only one frequency channel whereas the other streams have two (positive and negative) frequencies for branching, as illustrated in Fig. 1. The signal from the transmitter has total power of 8.7-dBm, and its spectrum is shown in Fig. 4(a). After 10-km transmission, the signal is demultiplexed by a WSS. The port frequency setting of WSS is summarized in Table 1. Due to 1-GHz frequency setting resolution of WSS, the bandwidths of ports are 17- or 18-GHz, although the frequency space of sub-carriers is 17.5-GHz. In addition, only positive frequency sub-carriers are tested for system evaluation, because the negative frequency sub-carriers are the complete copies of those on positive frequencies owing to the real-value waveform generated by the dual-drive MZM. Although the typical insertion loss of WSS is 4.5-dB, extra 0.5-1.0-dB loss was found in each branch due to small bandwidth setting. As a result, the powers at the output of WSS are about −7.5-dBm for sub-carriers, whereas that of center carrier is −5.3-dBm. The demultiplexed spectrum of 17.5-GHz sub-carrier is shown in Fig. 4(b), and Fig. 4(c)and 4(d) show the BERs of 4-sub-carriers for back-to-back (BtoB) and after transmission over 10-km SMF, respectively. 4-sub-carriers are carrying different data sequences, which are pseudo-random binary sequences (PRBSs) with the length of 215, 211, 29, and 27. At the receiver side, 5-tap linear feed forward equalizer (FFE) is applied and the typical eye-diagrams after equalization are shown in the insets of Fig. 4(c) and 4(d). The penalties of 10-km transmission are almost negligible. The center carrier shows 0.5-dB penalty from the other carriers, due to the residual base-band components of sub-carrier modulation which stems from the imperfect MZM and modulation waveforms. The received powers at the BER of 1 × 10−3 are −23.2-dBm for sub-carriers and −22.7-dBm for center carrier, and the corresponding loss budgets are respectively 15.7-dB and 17.4-dB, which means 128-way split (=16-branch(12 dB) x 3-sub-carriers x 2(pos/neg)-frequencies + 32-branch(15 dB)) can be achieved tolerating more than 2-dB extra loss.
Fig. 4. (a) optical spectrum of sub-carrier multiplexed 4 × 12.5-Gb/s signal, (b) demultiplexed spectrum at 17.5-GHz sub-carrier, and the measured BERs for (c) back-to-back and (d) after 10-km transmission.
Table 1. Port frequency setting of WSS for 4 × 12.5-Gb/s
Next, we have tried network reconfiguration. We assume that a large bandwidth for new branch is temporally requested, and 50-Gb/s is assigned on the center carrier for this bandwidth request. The traffic of existing users is re-assigned on 2-sub-carriers, instead of four in normal configuration, and each sub-carrier is output from two ports of WSS, as illustrated in the inset ii) of Fig. 3. In this case, the temporal new branch is easily configured without changing the hardware configuration, although the maximum data-rate for existing users is reduced. The spectrum of sub-carrier multiplexed signal is shown in Fig. 5(a), and the port frequency setting of WSS is summarized in Table 2. Two 12.5-Gb/s streams are allocated on 36.25- and 53.75-GHz, and therefore the signals keep 5-GHz guard-band. The total output power from the transmitter side is 8.9-dBm, and the powers at the output of WSS are −2.0-dBm for 50-Gb/s and −10.0-dBm for 12.5-Gb/s signals. Figure 5(b) and 5(c) are the demultiplexed spectra for 50-Gb/s and 12.5-Gb/s signals, and Fig. 5(d) and 5(e) are the measured BERs for BtoB and 10-km transmission. The received powers of 12.5-Gb/s branches at the BER of 1 × 10−3 are −23.5-dBm, and the corresponding loss budgets are 13.5-dB, where still 128-way split (=16-branch x 2-sub-carriers x 2(pos/neg)-frequencies x 2-port) can be achieved tolerating 1.5-dB extra loss. On the other hand, 50-Gb/s signal shows serious penalty after 10-km transmission, although the BER is still under 10−3 after digital equalization. Due to high baud-rate, the impact of CD is critical, and the bandwidth limitation of the receiver (30-GHz) also degrades the signal quality. The digital equalization is the combination of VNLE and 3-tap decision feedback equalizer (DFE). The tap length of first, second, and third order terms of VNLE are respectively 11, 11, and 3. Although the 50-Gb/s has large transmission penalty, the reconfiguration of PON network has been successfully demonstrated.
Fig. 5. (a) optical spectrum of sub-carrier multiplexed 50-Gb/s + 2 × 12.5-Gb/s signal, demultiplexed spectra of (b) baseband 50-Gb/s signal and (c) 12.5-Gb/s signal at 36.25-GHz sub-carrier, and measured BERs for (d) back-to-back and (e) after 10-km transmission.
Table 2. Port frequency setting of WSS for 50-Gb/s + 2 × 12.5-Gb/s
To improve the quality of 50-Gb/s transmission, another carrier frequency allocation is investigated. For 50-Gb/s signal, CD pre-compensation is applied in digital domain at the transmitter side. The baseband CD pre-compensated signal SCDC, which has complex value, is generated and converted to carrier frequency fc. By summing with its complex conjugate, as the following equation,
In the experiment, fc is set to 45-GHz and 25-Gb/s signal is assigned on the center carrier for existing users, where the guard-band is 7.5-GHz, as shown in Fig. 6(a), and Fig. 6(c) and 6(d) are the demultiplexed spectra. The amount of CD pre-compensation is −160-ps/nm, which is not a measured value but a typical value for 10-km SMF. The total power from the transmitter side is 8.8-dBm. Seeing from the measured BERs in Fig. 6(b), the quality of 50-Gb/s signal is drastically improved by CD pre-compensation. It should be noted that the digital equalizer at the receiver side is only second order VNLE with 7-taps. In addition, the power of 25-Gb/s signal is +1.0-dBm at the output of WSS and the loss budget at the BER of 1 × 10−3 is 22.3-dB, keeping 128-way split tolerating 1.3-dB extra loss.
Fig. 6. (a) optical spectrum of sub-carrier multiplexed 50-Gb/s + 25-Gb/s signal, (b) measured BERs after 10-km transmission, and demultiplexed spectra of (c) 25-Gb/s and (d) 50-Gb/s signals.
In every case, the achievable splitting number for existing users is 128-way tolerating >1.3-dB extra loss. In a practical system, however, more loss margin might be required for safety, and the splitting number will be reduced depending on the design policy. In addition, 12.5- and 25-Gb/s branches respectively achieved below −23- and −21-dBm received optical power at the BER of 1 × 10−3, owing to the sufficient guard band. It implies that splitting number (i.e., loss budget) for existing users can be kept in many cases by controlling the power and bandwidth assignment for sub-carriers.
4. Summary
We have experimentally demonstrated reconfigurable PON downlink by using a single intensity modulator and a WSS. The normal configuration is 4-sub-carriers x 12.5-Gb/s system with 10-km, 128-way split at maximum. To justify the insertion loss of WSS, low-loss C-band wavelength is selected, and the low baud-rate multi-carrier system shows negligible CD penalty. The network is reconfigured to make a new temporal branch with 50-Gb/s capacity. The traffic of existing users is re-assigned on 2-sub-carriers with reduced capacity. Although 50-Gb/s signal shows serious CD penalty, the BER achieved <10−3, and the network is successfully reconfigured without changing the hardware configuration. In addition, digital domain CD pre-compensation at the transmitter side drastically improve the transmission quality of 50-Gb/s signal, achieving the total capacity of 75-Gb/s.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. D. Nesset, “The progress of higher speed passive optical network standardisation in ITU-T,” in European Conference on Optical Communication2021 (ECOC2021).
2. B. Li, D. Nesset, D. Liu, Z. Ye, and L. Li, “DSP enabled next generation flexible PON for 50 G and beyond,” in Optical Fiber Communications Conference and Exhibition2022 (OFC2022), M3G.1.
3. R. Bonk and T. Pfeiffer, “New use cases for PONs beyond residential services,” in Optical Fiber Communications Conference and Exhibition2022 (OFC2022), Tu2G.1.
4. A. Kaszubowska-Anandarajah, K. Sivapalan, E. Martin, D. Gutierrez-Pascual, F. Smyth, J. Braddell, P. Lakshmijayasimha, and P. M. Anandarajah, “Reconfigurable PIC transmitter for short reach applications,” in Optical Fiber Communications Conference and Exhibition2022 (OFC2022), M4G.1.
5. L. Zhou, G. Peng, and N. Chand, “Demonstration of a novel software-defined flex PON,” Photon Netw Commun 29(3), 282–290 (2015). [CrossRef]
6. C. Yeh, C. Chow, M. Yang, and D. Hsu, “A flexible and reliable 40-Gb/s OFDM downstream TWDM-PON architecture,” IEEE Photonics J. 7(6), 1–9 (2015). [CrossRef]
7. J. Bäck, P. Wright, J. Ambrose, A. Chase, M. Jary, F. Masoud, N. Sugden, G. Wardrop, A. Napoli, J. Pedro, M. A. Iqbal, A. Lord, and D. Welch, “CAPEX savings enabled by point-to-multipoint coherent pluggable optics using digital subcarrier multiplexing in metro aggregation networks,” in European Conference on Optical Communication2020 (ECOC2020).
8. Y. Zhang, M. O’Sullivan, and R. Hui, “Digital subcarrier multiplexing for flexible spectral allocation in optical transport network,” Opt. Express 19(22), 21880–21889 (2011). [CrossRef]
9. T. Rahman, D. Rafique, B. Spinnler, A. Napoli, M. Bohn, A. M. J. Koonen, C. M. Okonkwo, and H. de Waardt, “Digital subcarrier multiplexed hybrid QAM for data-rate flexibility and ROADM filtering tolerance,” in Optical Fiber Communication Conference2016 (OFC2016), Tu3K.5.
10. M. Qiu, Q. Zhuge, M. Chagnon, Y. Gao, X. Xu, M. Morsy-Osman, and D. V. Plant, “Digital subcarrier multiplexing for fiber nonlinearity mitigation in coherent optical communication systems,” Opt. Express 22(15), 18770–18777 (2014). [CrossRef]
11. D. Welch, A. Napoli, J. Back, et al., “Digital subcarrier multiplexing: enabling software-configurable optical networks,” in Journal of Lightwave Technology, 2022, https://doi.org/10.1109/JLT.2022.3211466.
12. N. Stojanovic, F. Karinou, Z. Qiang, and C. Prodaniuc, “Volterra and Wiener equalizers for short-reach 100 G PAM-4 applications,” J. Lightwave Technol. 35(21), 4583–4594 (2017). [CrossRef]
