Open Access
Add to BibSonomyAbstract
While the gain-transient suppression of erbium-doped fiber amplifiers (EDFAs) has been widely studied, the large interval between upstream burst-mode signals from optical network units (ONUs) in time- and wavelength-division multiplexing passive optical networks (TWDM-PONs) presents new challenges. A non-gain-clamped EDFA acting as a preamplifier does not have the desired overshoot on the burst-mode signal when there are only a few ONUs in operation in the TWDM-PON. To solve this problem, we propose an all-optical gain-clamped EDFA (OGC-EDFA) that uses a distributed feedback laser diode to generate a saturation signal. An OGC-EDFA based on a ring laser configuration was also tested to compare the overshoot performance; the both OGC-EDFAs showed negligible overshoot performance. Given the negligible overshoot and wide input dynamic range of the OGC-EDFA, the proposed amplifier is thought to be a simple, low-cost solution for TWDM-PON applications.
© 2014 Optical Society of America
1. Introduction
Looking beyond currently deployed time-division multiple access (TDMA) passive optical networks (PONs) (20-km reach; 1:32 split), time- and wavelength-division multiplexing PONs (TWDM-PONs), which allow larger split ratios, are being developed [1,2]. TWDM-PONs have been selected as the next-generation PON2 (NG-PON2) for full service access networks (FSAN) [3]. Study group 15 of the International Telecommunication Union Telecommunication Standardization Sector (ITU-T) is currently standardizing NG-PON2, which includes both the TWDM-PON and point-to-point (PtP) wavelength-division multiplexing (WDM) PON for power saving and load balancing [4]. To support large split ratios and long distances, an optical amplifier will be used at the NG-PON2 optical line terminal (OLT) or as a research extender at remote sites [5]. The C-band erbium-doped fiber amplifier (EDFA) is a suitable preamplifier because the TWDM-PON relies on upstream signals in the C-band so that low-cost optical devices can be used on the ONU side. One of the main tasks of the preamplifier is to amplify burst-mode upstream signals at a negligible power penalty and sufficient input dynamic range, since bursts emanating from different ONUs can result in a significant change in the input power to the amplifier.
The timescale of burst-mode transmission in TWDM-PONs is very fast in comparison with that of channel add-dropping in WDM networks with optical add-drop multiplexers (OADMs).We will discuss details with later in this section. In the case of a 2.5-Gb/s upstream signal in a TWDM-PON, the burst length period is 125 μs. The preamble time of the burst overhead is set to a minimum of 64 ns and a maximum of 742 ns, which includes the 2R receiver settling time and clock and data recovery (CDR) locking time of the OLT burst-mode receiver as shown in Fig. 1.
Fig. 1 ITU-T XG-PON upstream physical (PHY) layer overhead [6, 7]. The TWDM-PON will have the same overhead configuration [8].
Figure 2 illustrates the upstream signal transmission in a TWDM-PON consisting of an OLT, preamplifier, optical splitters, and tunable ONUs. This TWDM-PON allows a 10-Gb/s upstream capacity utilizing 2.5 Gb/s × 4 wavelengths. In this work, we consider two possible scenarios: 1) all the upstream time slots are occupied by ONUs, resulting in the preamplifier experiencing quasi-continuous-wave (quasi-cw) WDM input signals with a 51.2-ns guard time between bursts as shown in Fig. 2(a), and 2(b) a power-saving mode in which almost all of the ONUs are in sleep mode and upstream burst slots are rarely occupied by several ONUs [9, 10]. From the reference [9], the ONU woke up at intervals of 10 ms with the 1G downlink, and it took 1.6 ms to check for the presence or absence of downstream traffic.
Fig. 2 TWDM-PON upstream transmissions (a) when all ONUs are in operation (b) when only a few ONUs are in operation.
As shown in Fig. 2(b), burst signals exist intermittently, and there is a large interval between upstream burst signals. Consequently, the TWDM-PON OLT preamplifier should be robust against burst-mode input signals with a large interval time. Furthermore, the differential path loss between the ONUs is 15 dB for gigabit-capable (G-), 10-G (XG-), and TWDM-PONs. The preamplifier should thus have a sufficient input dynamic range over 15 dB.
Several studies of burst-mode amplifiers have been conducted [11–13]. A 1.3-μm semiconductor optical amplifier (SOA) was studied to extend the gigabit Ethernet (GE-) PON system budget [12]. A hybrid electro-optical feedback gain-stabilized EDFA including a gain-compensation laser diode, loop controller, and pin-photodiode was reported for use as a long-reach WDM-PON burst-mode optical amplifier [13]; a 0.3-dB power penalty was reported, since the feedback scheme is an intrinsically slower process in comparison with the preamble time of the burst-mode signal.
Table 1 summarizes the key requirements of an EDFA for WDM network or TWDM-PON applications. It should be noted that the power settling time and input power changes are significantly different for each application.
Table 1. Comparison of EDFA requirements
A number of methods for gain-clamped EDFAs (GC-EDFAs) were reported to be applicable to dynamic add-drop WDM networks. An all-optical GC-EDFA (OGC-EDFA) configured with an external laser cavity and a feed-forward GC-EDFA employing a pump power control scheme are well-known techniques for clamping the population inversion state of the EDFA [14]. The pump power controlled EDFA is very suitable for metro or long haul WDM networks which require 50 ms of link protection time. The EDFA, however, completes a gain transient within a hundred microseconds, which is not sufficient for TWDM-PON applications.
In this paper, we report on an investigation of the amplification of burst-mode upstream signals in an EDFA used as a TWDM-PON preamplifier. Conventional EDFAs (C-EDFAs) show a significant gain transient for intermittent burst signals, and to solve this problem, we have added a saturation signal to clamp the gain of the EDFA, effectively eliminating the detrimental transient. We compare the overshoot performances of the C-EDFA and OGC-EDFA.
2. Experimental
Figure 3(a) presents a schematic diagram of the proposed OGC-EDFA, which consists of a 980-nm laser diode (LD) pumped C-band EDFA configured with a TO-Can-type 1535-nm LD at the input of the EDFA. The 1535-nm signal acts as a saturation signal to clamp the gain of the EDFA. Figure 3(b) shows the experiment setup for measuring the envelope and eye diagram of the amplified signal. We demonstrate a burst-mode signal by using a cw light source, pulse generator, and acousto-optic modulator (AOM) (400-ns rising/falling time) rather than a burst-mode transmitter since a TWDM-PON ONU transmitter is not commercially available. The optical pulse width of 125 μs corresponds to the time of the 2.5-Gb/s upstream signal packet in the TWDM-PON. The upstream signal wavelength is 1545 nm.
Fig. 3 (a) OGC-EDFA with saturation signal. (b) Experimental setup to measure overshoot and optical eye.
The gain of the C-EDFA is higher than that of the OGC-EDFA over the entire input range. The C-EDFA can be used for an input power of less than –30 dBm, but the change in the gain as a function of the input power limits the dynamic range of the EDFA. The 15-dB gain of the OGC-EDFA shown in Fig. 4 is maintained up to an input power of −10 dBm for a saturation signal of 0 dBm. Based on this nearly flat gain as a function of the input power, we expect good gain-clamped performance over a wide range of input powers. ITU-T G.989.2, which defines the TWDM-PON optical signal characteristics, gives the maximum upstream signal power to the OLT as −13 dBm [4]; the OGC-EDFA satisfies this requirement and could therefore be used as a preamplifier.
Stimulated lifetime corresponding to the stimulated emission rate is the reason of the gain dynamics of the EDFA [16]. The time dynamics of gain process give rise to transient effect that is described as an overshoot in this paper. The overshoot shows when a non-steady-state signal, for example burst-mode signal or low-speed modulated signal, is amplified. The front of the non-steady-state signal initially is amplified more due to a higher inversion of the EDFA. The end of the signal has less gain as it sees a low inversion. The proposed method to reduce overshoot effect includes the external steady-state signal to keep the inversion of the EDFA constant. Given that the inversion of the proposed EDFA is clamped by the saturation signal, a gain of an input signal does not vary.
To determine the dynamical characteristics of the EDFAs, we measured oscilloscope traces of the amplified optical pulse signals. Figures 5(a) and 5(b) show the oscilloscope traces of the amplified pulse signals. For an interval time of 0.03 μs, no gain transient from either the C-EDFA or the OGC-EDFA is observed because of the quasi-cw input signal. However, the shape of the output pulse of the C-EDFA changes as the interval time changes, whereas there is no significant variation in the output power of the OGC-EDFA when the interval time of the burst is varied.
The accumulated eye diagrams in Figs. 6(b) and 6(c) show thick one-levels created from the change in the signal power. It is quite probable that a significant error floor will be observed. Figures 6(d)–6(f) show the eye diagrams of the OGC-EDFA. Spots in the eye diagram are the artifact on the measurement when a pulsed 2.5 Gb/s signal inputs to digital communication analyzer (DCA).
Fig. 6 Accumulated eye diagrams of C-EDFA (a) 0.03 μs, (b) 125 μs, and (c) 1878 μs) and OGC-EDFA (d) 0.03 μs, (e) 125 μs, and (f) 1878 μs).
To quantify the change in the output power, we defined the overshoot as the percentage ratio of the maximum and minimum powers of the output signal. The overshoot of the C-EDFA increased rapidly to 160% for a change in the interval time from 0.03 μs to 125 μs, as shown in Fig. 7(a).The maximum overshoots of the C-EDFA and OGC-EDFA were 262% and 100.3%, respectively, for an interval time of 1,878 μs. In comparison with the C-EDFA, the strongly fixed population inversion status of the EDFA meant that the OGC-EDFA showed no overshoot peak.
The overshoot as a function of the saturation signal power to the OGC-EDFA is shown in Fig. 8(a).A saturation signal power of −3 dBm guarantees a negligible power penalty. We also measured the saturation signal power required to maintain an overshoot of 101% as a function of the OGC-EDFA input power. A saturation signal power that is 12 dB higher than the input power is enough to suppress the overshoot of the EDFA effectively.
Fig. 8 (a) Overshoot as a function of the saturation signal power. (b) Saturation signal power required to maintain an overshoot of less than 101% as a function of the EDFA input power.
TWDM-PON upstream wavelength plans are defined as a wide range (1524-1544 nm), a narrow reduced Range (1524-1540 nm), and a narrow Range (1532-1540 nm) in ITU-T G.989.2 recommendation draft. To avoid crosstalk between the upstream signal and the saturation signal the wavelength of saturation signal should be out of the TWDM-PON upstream wavelength plans. We used 1549.3 nm of DFB-LD as a saturation signal source. Figure 9 shows the measured eye diagram of the signal when the wavelength of the saturation tone is 1549.3 nm. No distortion of eye diagrams were observed.
Fig. 9 Accumulated eye diagrams the OGC-EDFA with 1549.3 nm saturation signal (a) 0.03 μs. (b) 125 μs. (c) 1878 μs.
The OLT burst-mode receiver determines a decision threshold level based on the bits of the initial part of the received upstream signal and clamps the decision threshold level until a burst-mode reset signal is received from the OLT media access control (MAC) module. A sudden received power change in the optical receiver will increase the power penalty of the signal owing to the non-optimized decision threshold level. (Fig. 10(a)).
Fig. 10 (a) Explanation of power penalty increase with non-optimized decision threshold level. (b) Power penalty as a function of the decision level offset.
We simulated a power penalty at bit error rate (BER) of 10−4 as function of the decision threshold level to investigate the variation in the power penalty from the decision threshold level to the signal, as shown in Fig. 10(b). In the TWDM-PON system, BER of upstream signal of 10−4 is recommended because FEC is widely used to PON system to extend link budget [4]. Simulations were performed with VPI and a direct-modulation 2.5-Gb/s distributed feedback (DFB-) LD with a 6-dB extinction ratio and a pin-photodiode receiver. The results indicate that a significant bit error will arise in the C-EDFA case. However, we expect that there will be no power penalty if the OGC-EDFA is employed, since negligible overshoot on the amplified signals was observed. Since we don’t have the resources to measure the BER of burst mode signal, it is difficult to directly compare the BER performance of the C-EDFA and the OGC-EDFA. In the future work, we will compare our proposed method and other method.
We also investigated the overshoot performance of a typical OGC-EDFA with a ring laser cavity with optical couplers, an optical filter, and an attenuator to adjust the 15-dB signal gain (Fig. 11(a)).The oscilloscope trace in Fig. 11(b) shows that there was a small overshoot, and the ripple on the trace because of the relaxation oscillation of the gain-clamping laser signal [17, 18]. Because the power change in the signal power was 4% the measured eye diagram indicated no significant power change of the amplified output signal as shown in Fig. 11(c).
Fig. 11 (a) OGC-EDFA employing a ring laser cavity. (b) Oscilloscope trace of a signal with an optical pulse width of 125 μs and an interval time of 1,878 μs. (c) Accumulated eye diagram of the OGC-EDFA (1,878 μs).
A gain-clamped SOA can also be used for burst-mode signal amplification [19], but there are drawbacks of a low gain and high noise figure when it is used as a preamplifier at the OLT. Recently, a compact EDFA that is compatible with conventional XFP optical transceivers with respect to the size was commercialized [20]. The proposed OGC-EDFA adopting a saturation signal can also be fabricated as a compact EDFA by using the same fabrication technique.
3. Conclusion
We have shown that the C-EDFA does not have the desired overshoot on the upstream signal when only a few ONUs are in operation in a TWDM-PON. Significant improvement in the overshoot performance is possible by employing an OGC-EDFA. By inserting a low-cost, TO-CAN-type DFB-LD in front of the EDFA, the overshoot reduced to almost zero on a 2.5-Gb/s burst-mode signal, which is essential for eliminating the power penalty under large-interval-time conditions. Since the OGC-EDFA has a wide range of input powers and interval times, the EDFA satisfies the requirements of a TWDM-PON preamplifier.
Acknowledgments
This research was funded by the MSIP (Ministry of Science, ICT & Future Planning), Korea in the ICT R&D Program 2014 [14-000-05-002]. The authors gratefully acknowledge the colleagues in optical access research section for the useful comments on an early version of the paper. The authors also wish to thank the reviewers for the constructive and helpful comments on the revision of this article.
References and links
1. Y. Ma, Y. Qian, G. Peng, X. Zhou, X. Wang, J. Yu, Y. Luo, X. Yan, and F. Effenberger, “Demonstration of a 40 Gb/s time and wavelength division multiplexed passive optical network prototype system,” OFC/NFOEC 2012, paper PDP5D.7 (2012).
2. E. Wong, M. Mueller, and M. C. Amann, “Characterization of energy-efficient and colorless ONUs for future TWDM-PONs,” Opt. Express 21(18), 20747–20761 (2013). [CrossRef] [PubMed]
3. FSAN next generation PON task group, http://www.fsan.org/task-groups/ngpon/.
4. 40-Gigabit-capable passive optical networks (NG-PON2): Physical media dependent (PMD) layer specification, ITU-T G.989.2 (2013).
5. K. Kim, J. Lee, S. Lee, J. Lee, and Y. Jang, “Low-cost, low-power, high-capacity 3R OEO-Type reach extender for a long-reach TDMA-PON,” ETRI. J. 34(3), 352–360 (2012). [CrossRef]
6. X. Qiu, “Burst mode receiver technology for short synchronization,”, OFC/NFOEC 2013, Tutorial OW3G.4 (2013).
7. J. Kim, J. J. Lee, S. Lee, and Y. Kim, “Physical media dependent prototype for 10-Gigabit-capable PON OLT,” ETRIJ 35(2), 245–252 (2013). [CrossRef]
8. 40-Gigabit-capable passive optical networks (NG-PON2): Transmission convergence (TC) layer specification, ITU-T G.989.3, under study in ITU-T SG15.
9. R. Kubo, J.-i. Kani, H. Ujikawa, T. Sakamoto, Y. Fujimoto, N. Yoshimoto, and H. Hadama, “Study and demonstration of sleep and adaptive link rate control mechanisms for energy efficient 10G-EPON,” J. Opt. Commun. Netw. 2(9), 716–729 (2010). [CrossRef]
10. H. H. Lee, K. Kim, J. Lee, and S. Lee, “Efficient power-saving 10-Gb/s ONU using uplink usage-dependent sleep mode control algorithm in WDM-PON,” ETRI J. 35(2), 253–258 (2013). [CrossRef]
11. Y. Awaji, H. Furukawa, B. J. Puttnam, and N. Wada, “Burst-mode optical amplifier,” OFC/NFOEC 2010, paper OTh14 (2010).
12. N. Suzuki and J. Nakagawa, “First demonstration of full burst optical amplified GE-PON uplink with extended system budget of up to 128 ONU splits and 58 km reach,” ECOC 2005, paper Tu1.3.3 (2005).
13. P. Ossieur, C. Antony, A. M. Clarke, A. Naughton, H.-G. Krimmel, Y. Chang, C. Ford, A. Borghesani, D. G. Moodie, A. Poustie, R. Wyatt, B. Harmon, I. Lealman, G. Maxwell, D. Rogers, D. W. Smith, D. Nesset, R. P. Davey, and P. D. Townsend, “A 135 km, 8192-split, carrier distributed DWDM-TDMA PON with 2 x 32 x 10 Gb/s capacity,” J. Lightwave Technol. 29(4), 463–474 (2011). [CrossRef]
14. H. Nakaji and M. Shigematsu, “Wavelength dependence of dynamic gain fluctuation in a high-speed automatic gain controlled erbium-doped fiber amplifier,” IEEE Photon. Technol. Lett. 15(2), 203–205 (2003). [CrossRef]
15. H. H. Lee, D. Lee, H. S. Chung, and H. J. Lee, “Effective suppression of signal-wavelength dependent transients in a pump-controlled L-band EDFA,” IEEE Photon. Technol. Lett. 16(6), 1462–1464 (2004). [CrossRef]
16. P. C. Becker, N. A. Olsson, and J. R. Simpson, Erbium-Doped Fiber Amplifier, (Academic Press, 1999).
17. J. T. Ahn and K. H. Kim, “All-optical gain-clamped Erbium-doped fiber amplifier with improved noise figure and freedom from relaxation oscillation,” IEEE Photon. Technol. Lett. 16(1), 84–86 (2004). [CrossRef]
18. G. Luo, J. L. Zyskind, J. A. Nagel, and M. A. Ali, “Experimental and theoretical analysis of relaxation-oscillations and spectral hole burning effects in all-optical gain-clamped EDFA’s for WDM networks,” J. Lightwave Technol. 16(4), 527–533 (1998). [CrossRef]
19. J. Sugawa, H. Ikeda, S. Matsuda, and M. Suzuki, “Optical amplifier technologies for high power budget PON Systems,” COIN2012, paper WF.3 (2012).
20. Licomm Corp, “XOA,”http://licomm.com/2011/product/product_detail.asp?Depth1=1&Depth2=4&Seq=96.
