Open Access
Add to BibSonomy
Abstract
This paper introduces the concept of a symmetric 10 Gbit/s high power-budget TDM-PON based on digital coherent technology and confirms its feasibility through a bidirectional transmission experiment with a transmission distance of 40 km and power budget of more than 50 dB. Burst-mode upstream 10 Gbit/s binary-phase-shift-keying (BPSK) signals synchronized by the clock recovered from downstream 10 Gbit/s NRZ signals are detected by using an optical pre-amplifier and coherent detection based on real-time burst-mode digital signal processing (DSP) in the optical line terminal (OLT). The real-time DSP implements coefficient handover in the adaptive equalizer to allow the reception of burst-mode upstream BPSK signals with short preamble length. An experimental bit error performance evaluation of the real-time burst-mode DSP yields the receiver sensitivity of -45.1 dBm for upstream burst-mode BPSK with a preamble length of 1.3 μs. For downstream signals, the receiver sensitivity of -38.9 dBm is achieved by using a chirp-controlled transmitter with optical post-amplifier so as to avoid the signal distortion created by the chromatic dispersion of single mode fiber (SMF) when the launched power is increased.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Fiber to the home (FTTH) has been widely deployed to provide broadband services to customers. In Japan, the number of FTTH subscribers exceeded 33 million in 2020 [1]. One of the key technologies supporting this massive FTTH deployment is time division multiplexed passive optical networks (TDM-PONs), which offer cost effective optical access by sharing the fiber to the central office (CO) and optical line terminal (OLT) in a CO through optical power splitters among a number of optical network units (ONUs) at subscriber premises.
To provide PON-based optical broadband services more cost-effectively, a drastic improvement in the link budget is necessary to support PON systems with extended transmission distance and/or increased splitting ratios; the maximum loss budget between ONU and OLT of current TDM-PONs is typically 29 to 31 dB. Higher loss budget PON systems enable us to realize CO consolidation, and hence achieve dramatic reductions in capital expenditure (CAPEX) and operation expenditure (OPEX). This is attractive for not only the existing network operators but also new network operators who are considering installing a new network infrastructure because higher loss budget PON systems can reduce the initial cost by reducing the number of COs and the amount of equipment required. Optical amplifiers used as repeaters have been investigated to extend the transmission distance of PON systems [2–6]. However, installing active devices outside the CO creates problems for the network operator, such as power supply, failure monitoring, and so on. This demands unrepeated systems with much higher link budgets [7–10]. From the viewpoint of cost, all equipment used to increase the link budget should be placed in the CO. Therefore, utilizing optical amplifiers as pre-/post-amplifiers in the OLT is a promising approach to improving the link budget in a cost-effective manner. The downstream signals use the optical amplifier as a post-amplifier to increase the launched power and improve the link budget. Unfortunately, undesirable nonlinear signal distortion is created in single mode fiber (SMF) at very high input powers. Thus, the launched power should be determined so as to avoid optical nonlinear effects in SMF. For example, the launched power of 21 dBm and the power budget of 51 dB were achieved in downstream transmission by using a post-amplifier, such as an erbium doped fiber amplifier (EDFA), for 10 Gbit/s intensity modulated signals in the C-band [11].
Upstream transmission, on the other hand, cannot use optical fiber amplifiers as a post-amplifiers due to the high cost issues created by installing the optical amplifier in the ONU. Thus, the optical amplifier should be set in the OLT and used as a pre-amplifier before reception of upstream signals. Unfortunately, improving upstream receiver sensitivity by optical pre-amplification suffers from the amplified spontaneous emission (ASE) noise generated by the optical amplifier as the upstream signals have much lower signal power. To improve upstream reception sensitivity, one effective solution is digital signal processing (DSP)-based optical coherent detection with an optical pre-amplifier. Digital coherent reception inherently achieves much larger link budgets than direct detection [12].
The major technical challenge in applying digital coherent detection for optical access is realizing the coherent reception of the PON’s burst-mode upstream signals. The upstream PON signals are bursty with significant power differences burst by burst. That limits the dynamic range of receiver sensitivity. Thus, a level-stabilization technique is required that can equalize the received power of upstream burst signals. In addition, since optical signal characteristics, such as frequency bandwidth of transmitter in ONU and the channel distortion created by chromatic dispersion, are different with each burst signal, DSP-based adaptive equalization must be optimized for each burst signal. However, the conventional DSP-based adaptive equalizer requires relatively long periods to calculate the equalization coefficients. Therefore, DSP technologies with very quick response time are required to realize the adaptive equalizer for the short burst signals expected. In our previous works, we proposed a DSP-based burst-mode coherent receiver with a coefficient handover method that can reduce the convergence time of tap coefficients for adaptive equalization for each burst frame [13,14]. The high receiver sensitivity of -44.7 dBm for burst-mode 20 Gbit/s quadrature phase shift keying (QPSK) was demonstrated by a burst-mode coherent receiver with offline DSP [13]. Furthermore, receiver dynamic range of 22 dB was achieved by using an automatic level control EDFA to equalize the received power of each burst-mode signal [14]. However, a burst-mode coherent receiver with the coefficient handover method had not been demonstrated using real-time burst-mode DSP technologies. In addition, while each technology has been investigated individually, it remained necessary to evaluate the performance of bidirectional transmission.
In this paper, we propose the design of a symmetric 10 Gbit/s DSP-based TDM-PON system; it employs digital coherent technology for upstream and intensity-modulation and direct detection (IM-DD) scheme with optical post-amplifier for downstream, to realize very large link budgets. The feasibility of our proposal together with our implementation of the tap coefficient handover method for adaptive equalization in a real-time burst-mode DSP are experimentally demonstrated. The burst-mode upstream binary-phase-shift-keying (BPSK) signals synchronized to the clock recovered from downstream 10 Gbit/s IM signals yielded the loss budget of over 50 dB by real-time burst-mode DSP-based coherent detection. To the best of our knowledge, this is the first real-time and bidirectional demonstration of a DSP-based TDM-PON with the power budget over 50 dB and transmission distance of 40 km. Moreover, the coefficient handover method is tested for the first time in a real-time experiment to confirm very high receiver sensitivity can be achieved for burst-mode upstream signals with reasonable burst preamble length.
The remainder of this paper is organized as follows. Section II details the concept of our proposal to realize a longer reach and higher splitting ratio PON system based on digital coherent technology. Section III describes the real-time burst-mode DSP for the reception of burst-mode BPSK signals, and the coefficient handover method that achieves an adaptive filter with very short response time. Section IV shows our experimental setup and results of a feasibility study on service restoration using a prototype testbed for metro-access networks accommodating for multiple services. Finally, Section V concludes this paper.
2. Proposed concept
Figure 1 shows schematics of the current PON system and the proposed future PON system. The CO-sited OLT is connected to multiple ONUs (subscribers) by optical splitters and feeder fiber. In current PON systems, shown in Fig. 1(a), transmission distance and splitting ratio are usually limited to 20 km and 32, respectively, because both upstream and downstream transmission use IM-DD. The transmission distance and splitting ratio are determined by the link budget defined by transmitter (Tx) output power and receiver (Rx) sensitivity. Longer reach and higher splitting ratio PON systems can be realized by using higher-power-budget optical transceivers (TRxs) to improve the link budget. Such enhancement enables us to consolidate the OLTs now set in small-scale COs into a few large-scale COs, as shown in Fig. 1(b). This enhanced PON system has the potential to reduce the CAPEX and OPEX by reducing the number of COs. In this work, we target a symmetric 10-Gbit/s-class PON system with power budget of 50 dB. Here, the 1.5 μm wavelength band is used for both upstream and downstream signals because of the low optical loss in SMF. By assuming that the optical fiber loss and optical splitter loss were 0.3 dB/km and 3.5 dB/N for splitting ratio of 1:2N, respectively, the target system can support a 40 km transmission distance and accommodate over 2048 ONUs. While the power consumption of the proposed system is for further study, significant reductions in the total power consumption of the optical access network can be expected because our proposed 2048-split system as it corresponds to 32 sets of 64-split OLT and a layer-2 switch to further aggregate the traffic from the OLTs.
To realize the target system, optical post-/pre-amplifiers are used in the optical TRx in the OLT. Since these optical amplifiers are shared by the many users accommodated by the OLT, this configuration is more cost-effective than setting an optical amplifier in each ONU for improving the link budget. Unfortunately, the ASE noise generated by optical pre-amplification of the upstream signals degrades efficiency. To offset this problem with upstream signals, DSP-based optical coherent detection with an optical pre-amplifier is an attractive solution [12].
Figure 2 shows a schematic of the proposed TDM-PON system based on digital coherent technology. For downstream transmission, the Rx configuration of the ONUs is simplified by the use of the IM-DD scheme with post-amplification. This configuration can realize PON systems with high link budgets for downstream transmission [9]. On the other hand, for upstream transmission, the ONU Tx uses an advanced modulation format, such as phase modulation (PM). In principle, a PM signal with coherent detection enables us to achieve higher receiver sensitivity than IM signals [15]. The upstream signals are detected by digital coherent detection technology in the OLT. This greatly improves Rx sensitivity and thus the link budget.
One key issue is ensuring that the DSP-based Rx can detect the upstream burst-mode signals. The signal distortion depends on the characteristics of the ONU Tx and chromatic dispersion (which depends on optical fiber length). In addition, the received power of upstream signals at the OLT Rx differs with each burst signal. Therefore, the equalization performed by the DSP must be optimized burst by burst. Furthermore, to successfully detect the burst signals, the OLT DSP must have very short response times to compensate these differences in the burst-mode upstream signal. We have already proposed a burst-mode coherent receiver (BM-CR) and demonstrated its feasibility by simulations [13]. In the following section, we describe the BM-CR for the digital coherent detection of burst-mode upstream signals.
3. Real-time burst-mode digital coherent receiver
Figure 3(a) shows schematic of the real-time burst-mode coherent Rx; it consists of an intradyne coherent receiver (ICR), local oscillator (LO), analog-to-digital convertor (ADC) and real-time burst-mode DSP. When receiving upstream burst-mode signals with digital coherent reception, the wide variation in the optical powers of upstream burst signals can create problems. The variation limits the receiver’s dynamic range when using digital coherent technology because the amplitude of the signal input to the ADC after the ICR varies. When the receiving power variation is excessive, the ADC’s dynamic range cannot be fully utilized, and large quantization error is created. To ensure that the coherent Rx achieves wide dynamic range, an auto-level-controlling (ALC)-EDFA can be used as a pre-amplifier prior to ICR [14]. The ALC-EDFA consists of a burst-mode automatic-gain-controlling EDFA (AGC-EDFA) as a pre-amplifier, an optical bandpass filter (OBPF) with a bandwidth of 0.48 nm to eliminate the ASE noise, and an ALC circuit. The ALC circuit is used to keep the output power from the ALC-EDFA constant regardless of the input power of the burst frames as shown in Fig. 3(b). This function is achieved by changing the attenuation values of a variable optical attenuator (VOA) through fast feedforward (FF) control. Here, we used an acousto-optic modulator as the VOA. The response time of this FF control circuit with the VOA is around 50 ns. Figure 3(c) shows measured ALC-EDFA output powers. The powers are almost constant between input powers of -45 dBm and -21 dBm. Thus, a wide dynamic range in received power at the coherent receiver can be realized.
Fig. 3. (a) Configuration of the real-time burst-mode coherent receiver with ALC-EDFA, (b) optical signals input into ICR w/ and w/o ALC-EDFA, (c) output power from ALC-EDFA.
Another issue in applying the DSP technology to burst transmission is how to optimize the finite impulse response (FIR) filter tap coefficients for each burst. As conventional DSP-based adaptive equalizers need relatively long times for the convergence of FIR filter tap coefficients, a very long preamble is needed in each burst, thus reducing the bandwidth efficiency. Our proposed solution, the tap coefficient handover method, can reduce the convergence time drastically [13]. Figure 4 shows the scheme of our proposal method. The coefficient handover method is conducted as part of the ONU activation process as follows.
First, the OLT broadcasts a serial number (SN) grant to poll a newly connected ONU. The ONU sends a SN response that contains a preamble long enough to calculate the coefficients for equalization. The OLT receives this response and calculates the equalization coefficients. Then, the converged coefficients for that ONU are stored in memory together with the ONU-ID. Subsequently, the stored coefficients are read when OLT receives the next upstream burst signal from the ONU. The coefficient is then refined upon reception of the burst, and is taken over to the next burst from the ONU. Therefore, short preambles can be used for receiving all bursts after the activation process. The assignment and management of the ONU-ID is performed by the MAC function, and the calculation of the coefficients of the FIR filter is performed by the DSP. Thus, the proposed method requires cooperation between the MAC function and the DSP function as described in Ref. [13].
4. Experiment and results
4.1 Experimental setup
To evaluate the feasibility of our proposal, we implemented the coefficient handover method in our real-time burst-mode DSP described in Ref. [16] and conducted bit-error-rate (BER) measurements on a symmetric 10 Gbit/s PON system.
Figure 5 shows the experimental setup. At the OLT side, the downstream Tx consists of a distributed feedback laser diode (DFB-LD), LiNbO3 based Mach-Zehnder modulator (MZM) for On-Off-keying (OOK), and an EDFA as booster. The optical signal generated from the DFB-LD was modulated the Z-cut MZM yielding a chirp-controlled 10 Gbit/s OOK signal. Next, the downstream signal was amplified by an EDFA and transmitted through SMF. At the ONU side, the Rx consisted of an SOA as a pre-amplifier, an OBPF to eliminate the ASE noise generated from the SOA and an avalanche photo diode (APD). The data signal and 10 GHz clock were recovered from the received 10 Gbit/s signal by using the clock and data recovery (CDR) technique and sent to the error detector #1(ED #1). The clock signal output from CDR was sent to the pulse pattern generator #1 (PPG #1) to synchronize upstream and downstream signals.
Fig. 5. (a) Expermental setup; PPG: pulse pattern generator, LD: laser diode, IQ-mod.: IQ modulator, (b) burst frame configuration.
The upstream signals were generated by a 10 Gbit/s burst-mode BPSK Tx consisting of an external cavity laser (ECL), LiNbO3 based in-phase and quadrature-phase (IQ)-modulator for BPSK modulation, and an SOA as the burst-mode shutter. This configuration is effective in avoiding the wavelength drift induced by self-heating in the laser diode (LD) [17,18]. This is because the injection current into the LD can be kept constant. The wavelength and linewidth of the ECL were 1532 nm and 10 kHz, respectively. It should be noted that an IQ-modulator was used in this study only to prove the concept. LiNbO3 based BPSK Txs are not suitable for ONU installation because of their large cost and size. A more cost-effective BPSK Tx can be developed on a push-pull type MZM. In particular, the Indium Phosphide (InP) based MZM is an attractive candidate because it can be monolithically integrated with laser sources [19,20].
Figure 5(b) shows the configuration of the transmitted burst frames. Each burst frame consists of a preamble of 1.3 μs, payload of 65 μs with a pseudo-random bit sequence (PRBS) pattern of 27-1 and an end of burst (EOB) of 1.3 μs. Note that EOB is a fixed bit pattern set at the end of each burst frame as a margin for transmitter disable. A guard time of 68 μs was used to separate burst frames. Although it is possible to insert another burst frame into this guard time, doing so is for a future study. The OLT Rx used an ALC-EDFA as a pre-amplifier and ICR with the real-time burst-mode DSP to receive the upstream burst-mode signals. The output power and the linewidth of LO were 9.2 dBm and 10 kHz, respectively. The carrier frequency offset between the upstream signal and the LO was set to within ±20 MHz. Note that the state of polarization of the upstream signal and the LO were manually controlled in this experiment because our DSP supported single polarization. In practice, polarization diversity receiving techniques could be realized by using a conventional method [12].
4.2 Results and discussion
Since the optical power of downstream signals launched into the SMF can be very high, optical nonlinear effects, such as Brillouin effect, must be countered. Figure 6 shows the optical loss due to Brillouin effect and loss budget of downstream transmission as a function of downstream signal power input to SMF. Optical loss by Brillouin effect increased with downstream signal power input; the loss budget degraded significantly when downstream signal power exceeded 12 dBm. Thus, downstream signal power input to SMF was set to 12 dBm. Figure 7 shows the receiver sensitivity and loss budget for upstream launched power as output by the burst-mode SOA in the back-to-back arrangement. Here, upstream launched power was controlled by the SOA’s injection current; power into the SOA was constant. Rx sensitivity was significantly decreased due to the chirp effect induced in the SOA when upstream launched power exceeded 6 dBm. Thus, the upstream launched power was set to 6 dBm. As a result, the loss budgets in the back-to-back configuration for downstream and upstream signals were more than 50 dB.
Figure 8 shows the measured burst-signal waveforms for received powers of -45 dBm and -25.5 dBm with and without ALC-EDFA, respectively. As a result, the input power difference of 19.5 dB was well suppressed by the ALC-EDFA, which offers wide dynamic range for burst-mode upstream transmission. The measured BER characteristics of downstream and upstream signals are depicted in Fig. 9. For downstream signals, receiver sensitivity improvement of 2.2 dB was observed after transmission through the 40 km SMF by setting the chirp parameter of the Z-cut LiNbO3 based MZM in Tx in the OLT to a negative value. Consequently, receiver sensitivity of -38.9 dBm and loss budget of 50.9 dB were achieved as shown in Fig. 9(a). Figure 9(b) shows the results for upstream burst-mode BPSK signals. The receiver sensitivity of -45.1 dBm was achieved by the real-time burst-mode DSP and the penalty after 40 km SMF transmission was negligible. As a result, the loss budget of 50.9 dB was achieved for upstream burst-mode BPSK signals by using the OLT-based BM-CR. In addition, wide receiver dynamic range of 19.5 dB was obtained by using an ALC-EDFA. Experiments showed that our proposed PON system can accommodate over 2048 users with 40 km transmission distance. The feasibility of the symmetric 10 Gbit/s 40 km reach DSP-based high power budget TDM-PON system has thus been confirmed.
The proposed system configuration can improve the power budget of future optical access systems as well. For example, the proposed technologies can be applied to a PON that employs both TDM and wavelength division multiplexing (WDM) technologies, such as a TWDM-PON, for upgrading PON systems. However, in order to realize a budget of over 50 dB for PONs beyond 10G, it is necessary to compensate the signal noise ratio (SNR) degradation imposed by the higher bit rate. This is particularly true for downstream transmission since it is difficult to increase the output power from the OLT, and the application of digital coherent technologies to the ONU side should be considered. In that case, the issue is how to realize a cost-effective ONU.
5. Summary
We proposed and experimentally evaluated a high power budget symmetric 10 Gbit/s DSP-based TDM-PON system.
The post-amplifier assisting IM-DD scheme with chirp-controlled transmitter was deployed for downstream signals resulting in a power budget of 50.9 dB for 40 km transmission. For the burst-mode upstream BPSK signals, by using a real-time BM-CR in an OLT that implemented the coefficient handover method, very high receiver sensitivity of -45.1 dBm was achieved for PON burst frames with regular preamble length. The upstream signals, synchronized to the clock recovered from downstream 10 Gbit/s IM signals, had negligible power penalty after 40 km transmission. These results confirm that our DSP based TDM-PON can achieve high power budget TDM-PON systems that support transmission distances of 40 km and 2048 users. Significant CO consolidation can be expected by employing the proposed system.
Disclosures
The authors declare no conflicts of interest.
References
1. Ministry of Internal Affairs and Communications Japan, “Trends in the Number of Subscribers for Broadband Services,” (2020), https://www.soumu.go.jp/johotsusintokei/field/tsuushin02.html
2. K.-I. Suzuki, Y. Fukada, D. Nesset, and R. Davey, “Amplified gigabit PON systems [Invited],” J. Opt. Netw. 6(5), 422–433 (2007). [CrossRef]
3. I. Voorde, C. Martin, J. Vandewege, and X. Qiu, “The SuperPON demonstrator: an exploration of possible evolution paths for optical access networks,” IEEE Commun. Mag. 38(2), 74–82 (2000). [CrossRef]
4. Z. Belfqih, F. Saliou, P. Chanclou, T. Soret, and N. Genay, “Bidirectional amplifier for standard PON architecture in burst mode configuration with class B+ attenuation range,” in Optical Fiber Communication (OFC) (2007), paper OWS1.
5. R. P. Davey, D. B. Grossman, M. Rasztovits-Wiech, D. B. Payne, D. Nesset, A. E. Kelly, A. Rafel, S. Appathurai, and S. H. Yang, “Long-Reach Passive Optical Networks,” J. Lightwave Technol. 27(3), 273–291 (2009). [CrossRef]
6. S. Shimizu, S. Kinoshita, K. Kitayama, and N. Wada, “0.73-W extremely low-power-consumption optical amplifier repeater for 10G-EPON systems,” in 40th European Conference on Optical Communication (ECOC) (2014), pp. 1–3.
7. K. Taguchi, H. Nakamura, K. Asaka, S. Kimura, and N. Yoshimoto, “Long-reach-tunable WDM/TDM-PON using synchronized gain-clamping SOA technology [invited],” J. Opt. Commun. Netw. 5(10), A144–151 (2013). [CrossRef]
8. K. Taguchi, K. Asaka, S. Kimura, and N. Yoshimoto, “High output power and burst extinction ratio (-tunable ONU transmitter using burst-mode booster SOA for WDM/TDM-PON,” J. Opt. Commun. Netw. 7(1), 1–7 (2015). [CrossRef]
9. K. Taguchi, “Long-Reach and High-Splitting Technologies for 40-Gbit/s-Class λ-tunable TWDM-PON,” in Optical Fiber Communication (OFC) (2016), paper W4C.4.
10. M. Fujiwara and R. Koma, “Increasing Splitting Ratio of Extended-Reach WDM/TDM-PON by Using Central Office Sited Automatic Gain Controlled SOAs,” IEICE TRANS. Commun. E100.B(8), 1388–1396 (2017). [CrossRef]
11. Z. Li, L. Yi, and W. Hu, “Comparison of downstream transmitters for high loss budget of long-reach 10G-PON,” in Optical Fiber Communication (OFC) (2014), paper Tu2C.4.
12. K. Kikuchi and S. Tsukamoto, “Evaluation of sensitivity of the digital coherent receiver,” J. Lightwave Technol. 26(13), 1817–1822 (2008). [CrossRef]
13. R. Koma, M. Fujiwara, J. I. Kani, K. I. Suzuki, and A. Otaka, “Burst-mode digital signal processing that pre-calculates FIR filter coefficients for digital coherent pon upstream,” J. Opt. Commun. Netw. 10(5), 461–470 (2018). [CrossRef]
14. R. Koma, M. Fujiwara, J. I. Kani, K. I. Suzuki, and A. Otaka, “Wide dynamic range burst-mode digital coherent detection using fast ALC-EDFA and pre-calculation of FIR filter coefficients,” in Optical Fiber Communication (OFC) (2015), pp. 6–8.
15. A. W. Davis, M. J. Pettitt, J. P. King, and S. Wright, “Phase Diversity Techniques for Coherent Optical Receivers,” J. Lightwave Technol. 5(4), 561–572 (1987). [CrossRef]
16. R. Koma, M. Fujiwara, J. Kani, S. Y. Kim, T. Suzuki, K. Suzuki, and A. Otaka, “Demonstration of Real-Time Burst-Mode Digital Coherent Reception with Wide Dynamic Range in DSP-Based PON Upstream,” J. Lightwave Technol. 35(8), 1392–1398 (2017). [CrossRef]
17. W. Poehlmann, D. V. Veen, R. Farah, T. Pfeiffer, and P. Vetter, “Wavelength Drift of Burst-Mode DML for TWDM-PON,” J. Opt. Commun. Netw. 7(1), A44–51 (2015). [CrossRef]
18. R. Bonk, W. Poehlmann, D. van Veen, J. Galaro, R. Farah, H. Schmuck, and T. Pfeiffer, “The underestimated challenges of burst-mode WDM transmission in TWDM-PON,” Opt. Fiber Technol. 26, 59–70 (2015). [CrossRef]
19. S. Matsuo, H. Ishii, T. Sato, R. Sato, K. Kasaya, A. Ohki, Y. Shibata, and H. Oohashi, “Monolithically integrated tunable DFB laser array and Mach-Zehnder modulator,” in 36th European Conference and Exhibition on Optical Communication (ECOC) (2010), pp. 1–3.
20. C. Kazmierski, N. Chimot, F. Jorge, A. Konczykowska, F. Blache, J. Decobert, F. Alexandre, A. Garreau, and R. D. Silva, “80Gb/s multi-level BPSK experiment with an InP-monolithic source based on prefixed optical phase switching,” in 26th International Conference on Indium Phosphide and Related Materials (IPRM) (2014), pp. 1–2.
