We propose arrayed optical amplifiers that share a single pump laser with the aim of realizing full-add/drop colorless, directionless, contentionless ROADM nodes and demonstrate its feasibility in experiments. The experimental results show that the fabricated arrayed optical amplifiers can be made to correspond properly to wavelength path reconfigurations by adjusting a splitting ratio of the variable splitter between the pump laser and eight EDFAs, and cause no significant penalty for 128-Gbit/s PDM-QPSK signal transmission.
© 2012 OSA
To deal with the rapid growth in internet traffic driven by the spreading use of various new services such as cloud computing and wireless backhaul, optical transport networks are evolving to achieve more flexibility as regards wavelength routing and wavelength assignment in addition to supporting bit rates of 100G and beyond. For example, the flexibility needed to realize mesh-based network topologies and dynamic capacity allocation is a key element in the design of next-generation networks. In such networks, multi-degree reconfigurable optical add/drop multiplexing (ROADM) with colorless, directionless and contentionless (C/D/C-less) functions is expected to play an important role in restoring wavelength paths for link or node failures and reconfiguring networks to optimize the utilization of network resources, because any add/drop port can support all wavelengths and connect in any fiber direction . Several node architectures have been proposed for implementing a C/D/C-less function in multi-degree ROADM [2–6]. In , we demonstrated experimentally a C/D/C-less ROADM node consisting of wavelength selective switches (WSS) with a high port count and transponder aggregators (TPA) using silica-based planar lightwave circuit (PLC) technology. However, if we are to realize a C/D/C-less node with a full-add/drop ratio, a study of the feasibility of implementing additional amplifiers in add/drop paths remains a crucial issue as pointed out in [5–7].
At the European Conference on Optical Communication held in 2012, we proposed an arrayed optical amplifier (AOA) that shares a single pump laser and reported experimental results showing its feasibility for use in the add/drop paths of a C/D/C-less node for 128-Gbit/s polarization-division-multiplexing quadrature phase-shift keying (PDM-QPSK) signal transmission systems . Although the concept of pump sharing between several amplifiers has been already reported in [9, 10], this is the first time it is applied to the amplifiers in the add/drop paths of a ROADM node. In this paper, we describe the design concept of our proposed AOA in more detail and report experimental results related to the performance of the fabricated AOA module. Moreover, we describe the performance requirements for optical amplifiers in the C/D/C-less node, for which we could not give a satisfactory explanation in  due to space limitations. The rest of this paper is organized as follows. In Section 2, we describe the C/D/C-less node configuration and the need for optical amplifiers in the add/drop paths. Then, we describe the design concept of our proposed AOA in Section 3. Section 4 presents experimental results showing that the fabricated AOA corresponds properly to the wavelength path reconfigurations and causes no significant penalty for a 128-Gbit/s PDM-QPSK signal transmission. Section 5 concludes this paper.
2. Performance requirements for optical amplifiers in C/D/C-less ROADM node
First, we describe the C/D/C-less ROADM node configuration. Figure 1 shows the configuration of an M-degree C/D/C-less node, where M denotes the number of fiber directions, which we assume to be eight in this work. The C/D/C-less node consists of colorless switches for connecting input and output express paths (hereafter referred to as wavelength cross-connects; WXC), TPAs and transmitters/receivers (Tx/Rx). On the drop side (blue line in Fig. 1), the incoming wavelength-division-multiplexing (WDM) signals from the degree-1 input fiber are delivered to another WXC or dropped to the TPA by way of the 1 × p WSS in WXC-1. The dropped signals are connected to the TPA through the 1 × q WSS. The M × r TPA collects the dropped signals from all the WXCs and allocates them to each desired Rx. On the add side (red line in Fig. 1), the output signal from the Tx is routed to the desired WXC by the r × M TPA. The p × 1 WSS in WXC-1 collects add signals that pass through the q × 1 optical coupler and signals from other WXCs, and then launches them into the degree-1 output fiber.
Here, let us discuss the add/drop ratio of the C/D/C-less node as shown in Fig. 1. A WXC composed of the 1 × p (or p × 1) WSS can be connected to (p-M + 1)·q·r Rx (or Tx) by way of the 1 × q WSS (or q × 1 optical coupler) and the M × r (or r × M) TPA. Thus, the available maximum add/drop ratio, Radd/drop, is expressed as11], since such multicast switch-based TPA has an intrinsic splitting loss due to the 1 × r splitter, the adoption of a WSS with a higher port count, p or q, is an appropriate way to minimize the loss increases for the add/drop paths. However, a higher q value is undesirable for the add side, because the q × 1 optical coupler also has intrinsic loss according to the q value. Thus, considering that the previously reported maximum port count of the WSS is forty-three , a practical way of realizing a full-add/drop C/D/C-less node is as follows: We use a high port count (p ≥ 23) WSS in the WXC, a low port count (q ~4) WSS between the WXC and TPA, and an 8 × 12 multicast switch-based TPA. In this work, we demonstrated experimentally a full-add/drop C/D/C-less ROADM node by using a 1 × 43 WSS, a 1 × 4 WSS and an 8 × 12 TPA as an example. The results are described in Section 4.
Next, we explain the need for optical amplifiers in the add/drop paths of the full-add/drop C/D/C-less node and their performance requirements. Figure 2(a) and 2(b) show a level diagram of the signal power for the drop and add-side paths, respectively. In these Figs., A ~D indicate the power monitor points in the add/drop paths. We calculated the add/drop-path losses based on the fact that the loss of a high port count WSS such as a 1 × 23 or 1 × 43 WSS is 10 dB, a low port count WSS has a loss of 5 dB, and the TPA loss is 14 dB including the splitting loss of 10.8 dB. For the drop side, as shown in Fig. 2(a), if the output power of the pre-amplifier is 0 dBm/ch, the input power of the receiver is reduced to −29 dBm/ch. Since the minimum signal power at the input of a digital coherent receiver is specified at −18 dBm by an implementation agreement published by the Optical Internetworking Forum (OIF) , we must at least compensate for drop-path losses of more than 10 dB. Similarly, for the add side, as shown in Fig. 2(b), given that the power of the output signal from the Tx is + 1 dBm/ch , the input power of the post-amplifier is reduced to about −30 dBm/ch. Since the maximum gain of the commercially available optical amplifier for a WDM system with 96 wavelength channels is around 20 dB, an add-path loss of more than 10 dB must be compensated for if we are to launch the WDM signals with a power of around 0 dBm/ch into the output fiber. Therefore, it is obvious that additional optical amplifiers with a gain of more than 10 dB must be introduced into the add/drop paths to realize a full-add/drop C/D/C-less ROADM node.
Figure 3 shows the configuration of the C/D/C-less node with additional amplifiers in the add/drop paths. In this Fig., triangles at “D1” ~“D3” and “A1” ~“A3” represent possible positions at which to insert the additional amplifiers for the drop and add-side paths, respectively. Table 1 shows the estimated number of wavelength channels input into the amplifier and the number of the amplifier modules for each insertion point. In this estimation, we limit the number of amplifier insertion points to one for each scenario, “D1” ~“D3” and “A1” ~“A3”, for simplicity. Note that while the amplifier at “A3” handles a single wavelength channel sent from the Tx, twelve wavelength channels are launched into the amplifier at “D3” in front of the Rx, because we select a desired wavelength channel by tuning the frequency of the local oscillator (LO) in the coherent receiver instead of using a tunable filter to reduce the number of required optical components. This estimation indicates that more amplifiers with a lower pump power are needed as the amplifier insertion point approaches the Tx/Rx.
Then, we explain how to decide the most suitable position if we are to implement the concept of pump sharing. In terms of reducing the number of required optical components, it is desirable to insert the amplifiers at “A1” and “D1”. However, considering the effect of failures of the AOA, “A1/D1” has a serious drawback compared with “A2/D2” or “A3/D3”. For example, when a single pump laser is shared between N amplifiers at “A1/D1”, 48·N Tx/Rx lose the connection to the WXC by the failure of the pump laser. On the other hand, if the pump laser shared between M ( = 8) or r ( = 12) amplifiers at “A2/D2” or “A3/D3”, respectively, the number of the unavailable Tx/Rx is limited to r ( = 12). When comparing “A2/D2” and “A3/D3” on the basis that the identical AOA is introduced to the add/drop paths, there is no difference in the required pump power per one amplifier. Thus, in this work, we determined the amplifier insertion position at “A2/D2” rather than “A3/D3” to reduce the optical components. So, our proposed AOA is installed between a 1 × 4 WSS (or 4 × 1 optical coupler) and TPA as shown in Fig. 4 .
3. Design concept of arrayed optical amplifiers
In this section, we explain the design concept and configuration of our proposed AOA. As described in Section 2, if we assume that a commercially available erbium-doped fiber amplifier (EDFA) module is installed between a 1 × 4 WSS (or 4 × 1 optical coupler) and a TPA in the C/D/C-less node, a few hundred discrete EDFA modules are required in a single node. This approach would be undesirable in terms of equipment size and cost. Thus, we proposed the AOA as a way of resolving this problem. Figure 5 shows the configuration of our proposed AOA. The AOA is composed of input/output power monitors, a single pump laser, signal/pump couplers, EDFAs and isolators. All the EDFAs consist of the same kinds of commercially-available erbium doped fibers with the length of about 3 m. The pump power is distributed to eight EDFAs through a splitter with a variable splitting ratio, and then precisely adjusted with variable optical attenuators (VOA) . This variable splitter with eight VOAs was fabricated using silica-based PLC technology. The 1 × 8 splitter consists of seven 1 × 2 splitters cascaded in three stages. Each 1 × 2 splitter is composed of the thermo-optic PLC switch based on the Mach-Zehnder interferometer. The excess losses of the 1 × 8 splitter and the VOA are 1.0 and 0.9 dB, respectively. The loss variation and polarization dependent loss (PDL) of these components are less than 0.1 dB. Moreover, the tap couplers for the power monitor and signal/pump couplers are also integrated in one PLC chip [16, 17]. The power tap ratio of the tap coupler is designed to be about 5% and its excess loss and PDL is 0.9 and 0.1 dB, respectively. As regards the signal/pump coupler, the average value of the excess loss and PDL are 0.7 and 0.1 dB, respectively, for both wavelengths. The loss variation is negligible (< 0.1 dB) for the wavelength of 1550 nm, but that for the wavelength of 980 nm is 0.2 dB. We believe that this integration approach is more promising for reducing the component size and manufacturing cost as the number of required optical components increases.
In the design of the AOA that shares a single pump laser between several EDFAs, it is very important that the output power of the shared pump laser is the same as that of a discrete EDFA module rather than a high power laser if we are to achieve effective reductions in power consumption and cost. We explain why the single pump laser can be shared in our AOA in detail below. Since the number of wavelength channels dropped from WXC-m (m = 1 ~M) to one M × r TPA is not greater than r, the relation given below exists as regards the pump power, Ppump, m, needed for a single EDFA-m.Equation (2) means that if we use a discrete EDFA module instead of an AOA, a pump laser with a maximum output power of r·P is needed for every EDFA module. On the other hand, it is obvious from Eq. (3) that the output power of r·P is sufficient for the shared pump laser in the AOA. Of course, this logic is applicable to the add-side AOA.
Next, we describe the pump power control of the AOA. Since wavelength channels are added or dropped in response to network reconfiguration or unexpected faults, the pump power control must limit the power excursion of the surviving channels. To respond to the designed reconfiguration, a response speed of a few milliseconds is sufficient for the pump power control. However, we need fast pump power control systems that operate on a microsecond timescale to respond to the abrupt changes in the number of wavelength channels resulting from unexpected faults. Their introduction complicates the control system and needs optical components with a fast response speed, thus leading to increased cost. We think that the performance degradation caused by transitional gain variations could be disregarded if we used a coherent receiver with a wide dynamic range . Therefore, the fast pump power control is avoided in this work.
For our experimental demonstration, we designed an AOA to compensate for the insertion loss of the 8 × 12 TPA. So, the designed gain value equaled the TPA loss of about 14 dB, and the maximum number of input wavelength channels was twelve for each EDFA. We used a single 976-nm laser with an output power of 500 mW as the shared pump laser.
4. Experimental results
We confirmed the feasibility of our proposed AOA in the add/drop paths of the C/D/C-less ROADM node experimentally. First, we ensured that the pump power was properly distributed to each EDFA and the optical signals were amplified without any penalties. Figure 6 shows the measured gain value and noise figure (NF) of each EDFA in the fabricated AOA. The power level diagram of this experiment corresponds to that of the add/drop paths shown in Fig. 2. Multi-channel (up to 12) continuous-wave (CW) lights with powers of −15 and −13 dBm/ch in the C-band were launched into the AOA for the drop and add side, respectively. Then, we measured the output power and NF for one CW light at a frequency of 193.1 THz. As shown in Fig. 6, the measured gain was equal to the expected value of 14 dB regardless of the input wavelength channels for all the EDFAs (Amp. 1 ~8). Fig. 7 shows the attenuation value of each VOA. The variation in these values resulted from the errors in the splitting ratio of the variable splitters or the variation in the excess losses of optical components between eight pump light paths. In addition, we launched multi-channel 128-Gbit/s PDM-QPSK signals instead of the CW lights and measured the bit error ratio (BER). Figure 8(a) and 8(b) show the experimental setup and results for the drop and add-side AOA, respectively. The received power level and optical signal-to-noise ratio with a 0.1-nm resolution was adjusted to −14 dBm/ch and 16 dB, respectively. Note that the multi-channel signals were launched into the coherent receiver without using an optical filter to eliminate unwanted channels. We suppressed the penalty resulting from this multi-channel detection by setting the local oscillator (LO) power at 13 dBm so that the power ratio between the LO and received signals was sufficiently large. The measured BER indicate that there were no penalties compared with the BER of 4 × 10−3 obtained in an experiment with a back-to-back configuration without an AOA or TPA. These results reported in Figs. 6 and 8 show that the pump power was properly distributed and the signals were successfully amplified without any penalties.
Next, we investigated the transient response of the fabricated AOA. We studied a worst-case scenario, corresponding to eleven wavelength channels being added or dropped. Figure 9 shows the transient behavior of the power of the surviving channel. Eleven channels were added or dropped using optical switches with switching times of 200 ms and 250 ns to emulate the designed wavelength path reconfigurations and unexpected network faults, respectively. For a switching time of 200 ms (Fig. 9(a)), the fabricated AOA successfully suppressed the power excursion with a gain offset of less than 0.5 dB. On the other hand, with fast switching (Fig. 9(b)), the surviving channel power excursion reached 10.8 dB. The power level returned to a steady state in about 20 ms. We think that the difference in transient responses between eight amplifiers may be caused by manufacturing errors in VOAs or their control system. Although the suppression of this power excursion will be a subjects for future investigation, we believe that the power excursion could be within the input power margin of a coherent receiver with a wide dynamic range of 20 dB , because the abrupt changes in the number of wavelength channels is confined to the decreases caused by unexpected faults such as a fiber cut in actual networks.
We report the results of a transmission experiment designed to demonstrate wavelength path reconfigurations. Figure 10(a) and 10(b) show the concept and configurations of twelve wavelength paths. The add and drop nodes were linked via three different routes, Paths I ~III. Paths I and II had two and one nodes that signals passed through, respectively, and these nodes were connected to each other by using 40-km dispersion-shifted fibers. 128-Gbit/s PDM-QPSK signals at frequencies of 192.1 (f1), 193.1 (f2) and 195.9 (f3) THz were transmitted through the Paths I, II and III, respectively. The routes of the remaining nine wavelength channels including a frequency of 193.0 (f0) THz were switched by using the add and drop-side TPA in the following order, A → B → C → B → A. Figure 10(c) shows the measured BER. The f1, f2 and f3 results were unaffected by the absence or presence of other wavelength channels. This result indicates that our fabricated AOA corresponded properly to the wavelength path reconfigurations.
We proposed an AOA with a shared pump laser to reduce the module size, manufacturing cost and power consumption compared with the introduction of discrete EDFA modules to realize a full-add/drop C/D/C-less ROADM node. We confirmed the feasibility of our proposed AOA experimentally. The experimental results show that the pump power was properly distributed to each EDFA by adjusting the splitting ratio of the variable splitter and 128-Gbit/s PDM-QPSK signals were successfully amplified without any penalties. Moreover, our experiments indicated that the fabricated AOA has sufficient responsiveness to wavelength path reconfigurations.
References and links
1. S. Gringeri, B. Basch, V. Shukla, R. Egorov, and T. J. Xia, “Flexible architectures for optical transport nodes and networks,” IEEE Commun. Mag. 48(7), 40–50 (2010). [CrossRef]
2. R. Jensen, A. Lord, and N. Parsons, “Colourless, directionless, contentionless ROADM architecture using low-loss optical matrix switches,” ECOC 2010, Mo.2.D.2 (2010).
3. S. Nakamura, S. Takahashi, M. Sakauchi, T. Hino, M. B. Yu, and G. Q. Lo, “Wavelength selective switching with one-chip silicon photonic circuit including 8×8 matrix switch,” OFC/NFOEC 2011, OTuM2 (2011).
4. R. Jensen, A. Lord, and N. Parsons, “Highly scalable OXC-based contentionless ROADM architecture with reduced network implementation costs,” OFC/NFOEC 2012, NW3F.7 (2012).
5. Y. Sakamaki, T. Kawai, T. Komukai, M. Fukutoku, T. Kataoka, T. Watanabe, and Y. Ishii, “Experimental demonstration of multi-degree colorless, directionless, contentionless ROADM for 127-Gbit/s PDM-QPSK transmission system,” Opt. Express 19(26), B1–B11 (2011). [CrossRef] [PubMed]
6. W. I. Way, “Optimum architecture for M×N multicast switch-based colorless, directionless, contentionless, and flexible-grid ROADM,” OFC/NFOEC 2012, NW3F.5 (2012).
7. G. J. Cowle, and M. Bolshtyansky, Optical amplifier trends for CDC node network architectures,” OECC 2012, 6D2–4 (2012).
8. Y. Sakamaki, T. Kawai, M. Fukutoku, T. Kataoka, and K. Suzuki, “Full-add/drop C/D/C-less ROADM achieved by developing arrayed optical amplifiers with a shared pump laser,” ECOC 2012, P3.03 (2012).
9. D. R. Zimmerman and L. H. Spiekman, “Amplifiers for the masses: EDFA, EDWA, and SOA amplest for metro and access applications,” J. Lightwave Technol. 22(1), 63–70 (2004). [CrossRef]
10. M. Bolshtyansky, H. Cheng, P. Colbourne, Z. W. Dong, D. Dougherty, K. Y. Huang, G. Wills, and G. Cowle, “Planar waveguide integrated EDFA,” OFC 2008, PDP17 (2008).
11. T. Watanabe, K. Suzuki, and T. Takahashi, “Silica-based PLC transponder aggregators for colorless, directionless, and contentionless ROADM,” OFC/NFOEC 2012, OTh3D.1 (2012).
12. Y. Ishii, K. Hadama, J. Yamaguchi, Y. Kawajiri, E. Hashimoto, T. Matsuura, and F. Shimokawa, “MEMS-based 1×43 wavelength-selective switch with flat passband,” ECOC 2009, PD 1.9 (2009).
13. Optical Internetworking Forum, “Implementation agreement for integrated dual polarization intradyne coherent receivers,” (2010). http://www.oiforum.com/public/documents/OIF-DPC-RX-01.0.pdf.
14. Optical Internetworking Forum, “Implementation agreement for integrated polarization multiplexed quadrature modulated transmitters,” (2010). http://www.oiforum.com/public/documents/OIF-PMQ-TX-01.0.pdf.
15. Y. Hashizume, Y. Inoue, T. Kominato, T. Shibata, and M. Okuno, “Low-PDL 16-channel variable optical attenuator array using silica-based PLC,” OFC 2004, WC4 (2004).
16. K. Jinguji, N. Takato, Y. Hida, T. Kitoh, and M. Kawachi, “Two-port optical wavelength circuits composed of cascaded Mach-Zehnder interferometers with point-symmetrical configurations,” J. Lightwave Technol. 14(10), 2301–2310 (1996). [CrossRef]
17. H. Ono, T. Watanabe, K. Suzuki, A. Mori, T. Takahashi, and T. Sakamoto, “An erbium-doped fibre amplifier with widely variable gain employing integrated components on a planar lightwave circuit,” ECOC 2011, Th. 11. LeCervin. 6. (2011).
18. K. Murata, T. Saida, K. Sano, I. Ogawa, H. Fukuyama, R. Kasahara, Y. Muramoto, H. Nosaka, S. Tsunashima, T. Mizuno, H. Tanobe, K. Hattori, T. Yoshimatsu, H. Kawakami, and E. Yoshida, “100-Gbit/s PDM-QPSK coherent receiver with wide dynamic range and excellent common-mode rejection ratio,” Opt. Express 19(26), B125–B130 (2011). [CrossRef] [PubMed]