Open Access
Add to BibSonomyAbstract
A novel mode-selective optical packet switching, based on mode-multiplexers/demultiplexers and multi-port optical micro-electro-mechanical systems (MEMS) switches, has been proposed and experimentally demonstrated. The experimental demonstration was performed using the LP01, LP11a and LP11b modes of a 30-km long mode-division multiplexed few-mode fiber link, utilizing 40 Gb/s, 16-QAM signals.
© 2015 Optical Society of America
1. Introduction
Packet switching offers better network resource utilization than circuit switching due to the statistical multiplexing. Optical packet switching (OPS) is the technology that challenges to perform high-speed header processing, switching, and/or buffering in the optical domain. OPS can potentially offer higher throughput and better energy efficiency than its electrical counterpart.
Focusing on the header processing, several optical and opto-electronic (OE) techniques, such as correlation of optical codes [1], in-band multi-wavelength [2], subcarrier multiplexing [3], and in-band RF tone [4], have been proposed. However, these techniques incur latency due to inevitable OE conversion and/or electrical processing. On the other hand, mode-division multiplexing (MDM) has been gaining much attention for its potential to increase the transmission capacity using a special class of multimode fibers (MMF), called few-mode fibers (FMF). To date, the highest demonstrated mode count is 15 modes [5].
There are two ways to treat the modes. One approach is the so-called “superchannel” [6], in which the modes are switched in bundles, while the other is the mode-unbundling, in which each mode can be switched individually, e.g. by adding or dropping one of the modes via an MDM reconfigurable add/drop multiplexer (ROADM) [7].
In this paper, we concentrate on the mode-unbundling case of MDM. We propose and experimentally demonstrate a novel mode-selective switching, in which a mode acts as a label that carries the information of the output port of the switch. Hence, the switching is performed on-the-fly using mode-multiplexers/demultiplexers (mux/demux) [8] without any extra header processing. Suppose that an individual guided mode in a FMF or MMF represents a specific output port of the switch, an optical packet on a specific mode at the input port is forwarded to the designated output port by passing through a mode-demux and the optical switch, if the state of the switch is properly set. This allows the use of a slow but high port-count micro-electro-mechanical systems (MEMS) switch [9], which can potentially preserve the mode field pattern.
Particularly, after the brief introduction in chapter 1, the operation principle of the proposed mode-selective switching is explained in details in chapter 2. In chapter 3, the operation is demonstrated using the LP01, LP11a and LP11b modes of a 30-km long differential mode delay (DMD)-compensated MDM link [10], utilizing 40 Gb/s, 16-QAM signals. A conclusion is finally provided in chapter 4.
2. Operation principle of mode-selective switching
The operation principle of mode-selective switching is that a propagation mode can act as a label of header information, defined by routing decision in the transmitter as presented in Fig. 1.The conventional wavelength routing is a similar scheme since a wavelength acts as a label, e.g. generalized multiprotocol label switching (GMPLS) uses a wavelength as a shim header.
The schematic of mode-selective switching is illustrated in Fig. 2. For simplicity, we consider 2 × 2 OPS with the two-mode case, consisting of a 4 × 2 optical switch fabric and mode-demuxes, which demultiplex the LP01 and LP11 modes in an MDM system. It is assumed that the input ports are connected with two-mode waveguides. Depending on the scenario, the output ports can also be multimode waveguides using mode-mux in the output of the switch. The optical switch is set so that the packets on the LP01 mode are destined to the output port 1, while the packets on the LP11 mode are destined to the output port 2 as shown in Fig. 2(a).
Fig. 2 Mode-selective switching in the case of LP01/LP11 modes. (a) 2 × 2 configuration without contention using 4 × 2 optical switch. (b) With contention; for simplicity the buffer is neglected.
The routing table for the mode-selective switching case depicted in Fig. 2(a) is presented in Table 1.
Table 1. Routing table of 2 × 2 switch using two modes
Based on this table, each input port corresponds to a fiber input, while the optical switch ports connect the LP01/LP11 outputs from the mode-demuxes to the corresponding outputs of the switch. For example, an LP01-labeled signal coming from the second FMF (incoming port #2) will enter the optical switch from port #3 and will be switched to the output port #1. In order to preserve the mode field patterns of LP01 and LP11 modes, optical MEMS switches can be preferably applied, because the MEMS mirror does not alter the mode field pattern. Therefore future implementations can be fabricated to incorporate both demultiplexing and switching functionalities within the same module.
Contention will occur when two packets are destined to the same output port at the same time as shown in Fig. 2(b), and the contention resolution requires buffers such as fiber delay lines and/or electronic random access memory. However, the contention resolution is out of scope and not shown in the figure. It is noted that inevitable imperfections of the mode- demultiplexer induces mode crosstalk between the two ports.
In Fig. 3, the re-configurability of the switch is illustrated. The output ports of LP01 mode and LP11 are swapped from those in Fig. 2(a), therefore acting as a mode converter/swapper. More importantly, considering the existence of a time synchronization mechanism that ensures no time-overlapping of packets on different modes during propagation or prior to the switch, the crosstalk can be negligibly small, eliminating the need for multiple-input multiple-output (MIMO) processing at the receiver. As the port-count increases, more modes can be used. This will encounter challenges and might require a sophisticated design for low-loss mode-demux and mode excitation with high extinction ratio.
Fig. 3 Reconfigurability of mode-selective switching. The ports are swapped from those in Fig. 2(a).
3. Experimental demonstration
3.1 Experimental setup
We demonstrated the proposed mode-selective switching operation, using the LP01, LP11a, and LP11b modes of a two-mode MDM link. Our experimental setup is presented in Fig. 4. In the transmitter side, the optical signals were generated by an I-Q modulator, driven by a 10 GBaud arbitrary waveform generator (AWG) and an optical carrier at 1550 nm, generated by a 100 kHz linewidth laser. The routing decision was performed using two 1 × 2 LiNbO3 electro-optic switches (LN SW) in tandem, which switched the signals in either LP01, LP11a, or LP11b port of a multicore-fiber-based mode-mux [8]. Polarization controllers were used in the input of the mode-mux for crosstalk minimization. The signals were then launched into a 30-km DMD-compensated FMF link [10], able to carry 6 spatial and polarizations modes in total, composed of a 26-km negative-DMD span and a 4-km equally positive-DMD span.
The mode-selective switch was composed of a similar mode-demux and six input/output ports of a 128 × 128 MEMS switch [9]. The MDM part had overall path loss 4~6 dB for LP01 and 11~13 dB for LP11 paths and crosstalk of −9~-13 dB. The entire DMD of the link was measured to be <0.6 ns. Prior to the switch, the losses of the MDM part were compensated using single-mode erbium-doped fiber amplifiers (EDFA), followed by narrow band optical filters. The input power to the MEMS switch was set around 0 dBm and the path losses of the MEMS switch were 2~3 dB.
In the receiver side, the outputs of the MEMS switch were detected by a dual-polarization optical coherent receiver, using a 1 kHz linewidth local oscillator (LO). Prior to the coherent receiver, the optical signal-to-noise ratio (OSNR) was adjusted by an amplified spontaneous emission (ASE) noise-loading element and measured using an optical spectrum analyzer (OSA). The signals were then captured by a 40GSps real-time oscilloscope. Please note that each signal was processed individually as two separate receivers using single-input single-output (SISO) digital signal processing (DSP). For that, a 421-symbol long training sequence was used for timing, carrier frequency, and channel estimation; followed by 3072 16-QAM symbols. The symbol count was chosen in order to be compatible with a 1500-byte (or 12000-bit) Ethernet frame.
3.2 Experimental results
The measurements were performed separately for the two configurations of the MEMS switch shown in Fig. 5. For each configuration, we measured the bit-error-rate (BER) performance of the 16-QAM signals for all modes by changing the state of the LN SW in the transmitter. The BER measurement was measured for over 106 transmitted bits for each case, by error counting. We performed our measurements using the optical noise-loading element, considering a 7% forward error correction (FEC) overhead with threshold of 3.8 × 10−3 [11]. Please note that, for all cases, the received OSNR was higher than 30 dB without any noise-loading.
Fig. 5 Experimental results: (a) OSNR vs BER for configuration A, (b) constellations at OSNR = 16 dB for configuration A, (c) OSNR vs BER for configuration B, (d) constellations at OSNR = 16 dB for configuration B.
Our resulting BER versus OSNR performances, along with the corresponding constellation at OSNR = 16 dB, are presented in Fig. 5. The BER for all cases was well above the FEC threshold for OSNR ≥ 16 dB and the error-vector-magnitude (EVM) performance for the OSNR = 16 dB case was 19~20%. Since the packets are decorrelated both in spatial and temporal domains, the transmission is almost single-mode with some multipath nature due to modal crosstalk. However, all cases achieved similar performances, without any significant degradation due to the modal crosstalk. Particularly, the OSNR degradation compared to the BtB case for the FEC limit of 3.8 × 10−3 was around 2 dB.
3.3 Crosstalk concern
In addition, for demonstrating the operation of the proposed mode-selective switching, we captured the received waveforms at the 40 GSps real-time oscilloscope by changing the LN SW configuration in the transmitter side when using only LP01 and one of the LP11 modes. For the simultaneous representation, the LP01 and the LP11 outputs where combined prior to detection using a polarization beam combiner. The results for the MEMS configuration of Figs. 5(a) and 5(b), are presented in Fig. 6. Please note that the two shots were taken a few seconds apart. The Figs. 5(b) and 5(c) clearly show that the packet was properly routed based on the mode decided in the transmitter. However, it is important to note that the crosstalk (XT) accumulated in the 30-km FMF link was present for both cases. Considering the 4 × 2 switch configurations described in Fig. 2 and Fig. 3, this can pose performance degradation in the case that two packets are coming from different modes and different FMFs at the same time as shown in Fig. 5(a). A solution to this problem would be to implement a gate switch or a shutter in each input that operates based on the input power of the signal, or to assign each transmitter at different FMFs with different time-slots. In addition, weakly-coupled FMFs [12] or DSP-aided techniques can be adopted to minimize the impact of XT in either optical or digital domains.
Fig. 6 Crosstalk problem: (a) explanation, and waveforms captured by 40 GSps real-time oscilloscope for packets switched based on (b) LP01 and (c) LP11.
4. Conclusion
A novel mode-selective packet switching technique, based on mode-multiplexers/demultiplexers and multi-port MEMS switches, has been proposed. Advantages of the mode-selective switching are that the processing is performed on-the-fly requiring neither header processing, nor high-speed switching. The proposed mode-selective switching has been experimentally demonstrated using the LP01, LP11a, and LP11b modes of a 30-km DMD-compensated FMF link. FEC error-free 40 Gb/s, 16-QAM signals at an OSNR ≥ 16 dB have been successfully delivered, based on the mode-selective routing decision in the transmitter, for two different configurations of the multi-port MEMS switch.
Acknowledgments
The authors would like to thank T. Yamamoto from NTT Device Technology Labs for kindly providing us with the multi-port MEMS switch. This work has been partially supported by the “Basic Technologies for High-Performance Opto-Electronic Hybrid Packet Router” project funded by the National Institute of Information and Communications Technology (NICT), Japan, and by the FP7 EU-Japan STRAUSS project funded by the Ministry of Internal Affairs and Communications, Japan.
References and links
1. K. Kitayama and N. Wada, “Photonic IP routing,” IEEE Photonics Technol. Lett. 11(12), 1689–1691 (1999). [CrossRef]
2. N. Calabretta, W. Wang, T. Ditewig, O. Raz, F. Gomez Agis, S. Zhang, H. de Waardt, and H. J. S. Dorren, “Scalable optical packet switches for multiple data formats and data rates packets,” IEEE Photonics Technol. Lett. 22(7), 483–485 (2010). [CrossRef]
3. D. J. Blumenthal, J. Laskar, R. Gaudino, S. Han, M. D. Shell, and M. D. Vaughn, “Fiber-optic links supporting baseband data and subcarrier-multiplexed control channels and the impact of MMIC photonic/microwave interfaces,” IEEE Trans. Microw. Theory Tech. 45(8), 1443–1452 (1997). [CrossRef]
4. N. Calabretta, “FPGA-based label processor for low latency and large port count optical packet switches,” J. Lightwave Technol. 30(19), 3173–3181 (2012). [CrossRef]
5. N. K. Fontaine, R. Ryf, H. Chen, A. V. Benitez, B. Guan, R. Scott, B. Ercan, S. J. B. Yoo, L. E. Grüner-Nielsen, Y. Sun, R. Lingle, E. Antonio-Lopez, and R. Amezcua-Correa, “30×30 MIMO transmission over 15 spatial modes,” in Proc. Optical Fiber Communication Conference and Exposition (OFC 2015), paper Th5C.1 (Los Angeles, CA, USA, 2015).
6. M. D. Feuer, L. E. Nelson, K. Abedin, X. Zhou, T. F. Taunay, J. F. Fini, B. Zhu, R. Isaac, R. Harel, G. Cohen, and D. M. Marom, “ROADM system for space division multiplexing with spatial superchannels,” in Proc. Optical Fiber Communication Conference and Exposition/National Fiber Optic Engineers Conference (OFC/NFOEC 2013), paper PDP5B.8 (Anaheim, CA, USA, 2013).
7. K. Kitayama, A. Maruta, Y. Yoshida, N. P. Diamantopoulos, Y.-C. Huang, M. Nakazawa, and T. Isoda, “Mode division multiplexing network: a deployment scenario in metro area network,” in Proc. IEEE Global Communications Conference (GLOBECOM 2014), ONS-4, 2154–2159 (Austin, TX, USA, 2014). [CrossRef]
8. K. Takenaga, H. Uemura, Y. Sasaki, S. Nishimoto, T. Uematsu, K. Omichi, R. Goto, S. Matsuo, and K. Saitoh, “Multicore fibre-based mode multiplexer/demultiplexer for three-mode operation of LP01, LP11a, and LP11b,” in Proc. 40th European Conference on Optical Communications (ECOC 2014), paper Tu.4.1.4 (Cannes, France, 2014).
9. M. Mizukami, J. Yamaguchi, N. Nemoto, Y. Kawajiri, H. Hirata, S. Uchiyama, M. Makihara, T. Sakata, N. Shimoyama, and K. Oda, “128×128 three-dimensional MEMS optical switch module with simultaneous optical path connection for optical cross-connect systems,” Appl. Opt. 50(21), 4037–4041 (2011). [CrossRef] [PubMed]
10. R. Maruyama, N. Kuwaki, S. Matsuo, and M. Ohashi, “Two mode optical fibers with low and flattened differential modal delay suitable for WDM-MIMO combined system,” Opt. Express 22(12), 14311–14321 (2014). [CrossRef] [PubMed]
11. ITU-T, “Forrward error correction for high bit-rate DWDM submarine systems,” Rec. G.975.1, Appendix I.9 (2014).
12. D. Boivin, M. Bigot-Astruc, M. Travagnin, and P. Sillard, “Weakly-coupled few-mode fibers for single-mode and mode-division-multiplexed transmissions,” in Proc. Optical Fiber Communication Conference and Exposition/National Fiber Optic Engineers Conference (OFC/NFOEC 2013), paper OTh3K.6 (Anaheim, CA, USA, 2013). [CrossRef]
