In recent years, to enlarge the single-mode fibers (SMFs) transmission capacity, researchers focused on the dimension of space, which is a new degree of freedom that is being considered for optical fiber communication beyond WDM. Space-division multiplexing (SDM), including mode-division multiplexing (MDM) using multimode fibers (MMFs) or few-mode fibers (FMFs), and core multiplexing using multicore fibers (MCFs), has attracted much recent attention. In an SDM system, high-density spatial channels are tightly packed into a single fiber, thus making crosstalk among cores or modes a critical challenge due to fiber imperfections, bending, and twisting. Previous studies have mostly been confined to the routing algorithms for crosstalk reduction but few focuses on the in-service crosstalk monitoring, tracing and quality-of-transmission (QoT)-oriented lightpath re-optimization. In this paper, we proposed novel in-service crosstalk monitoring and tracing (CMT) method and algorithm using fine-grained optical time slice monitoring channels for crosstalk reduction in SDM optical networks. Benefitting from the large amount of fine-grained channels provided by optical time slices, it becomes possible for every source node to allocate a dedicated monitoring time slice carrying the traffic and path information for each connection. Crosstalk monitoring and tracing can be realized by extracting the information contained in these monitoring time slices. Simulation results shows that the proposed CMT method and algorithm can obtain acceptable performance in large-scale network scenarios. Furthermore, we also proposed a quality-of-transmission (QoT)-oriented lightpath re-optimization mechanism based on in-service crosstalk monitoring and tracing to maintain a high level of QoT. Finally, we designed a prototype experiment to validate our proposed in-service crosstalk monitoring method. Results show that this method can realize in-service crosstalk monitoring, tracing and lightpath re-optimization over a seven-core fiber based transmission system, and the crosstalk with a minimum value of −37.9 dB can be monitored and successfully traced.
© 2017 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
With the rapid growth of Internet services and emerging bandwidth-hungry applications, the traffic of data center networks, access networks and optical backbones has sustained exponential growth in recent years, and this trend is expected to continue for the foreseeable future [1–4]. However, previous studies  have shown that the transmission capacity over single-mode fibers (SMFs) is rapidly approaching its fundamental Shannon limit. To break through the capacity limit of single-core single-mode fiber, researchers focused on the dimension of space and proposed the concept of space-division multiplexing (SDM) by introducing multi-core fiber (MCF) and few-mode fiber (FMF) into optical networks [6–13]. It is proved that MCF and FMF are able to enlarge the transmission capacity as well as reduce network CapEx .
However, the introduction of MCF and FMF into optical networks also brings new challenges in routing and resource allocation [6, 15]. Except the common challenges for traditional optical networks like chromatic dispersion and non-linear impairments, one of the biggest challenges in SDM networks is the inter-core/inter-mode crosstalk, which destroys orthogonality among spatial channels. In order to mitigate the side effects of crosstalk in transmission, a common solution is to use coherent detection and multiple-input multiple-output (MIMO) digital signal processing (DSP) to recover the transmitted signals [6, 16, 17]. However, in short-haul applications like passive optical networks (PONs), intra-datacenter networks (intra-DCNs) and high-performance computing (HPC) systems, it is possible to neglect the optical signal-to-noise ratio (OSNR) penalties caused by the coupling effect among spatial channels because of the short distance [6, 18–21]. In this case, MIMO DSPs is not preferred because of high CapEx and high computational complexity.
Previous works [18–21] have proposed uncoupled MCF/FMF transmission with ultra-low inter-core/mode crosstalk, in which signals on each core/mode are transmitted as independent channel, and low-complexity DSPs for wavelength-division multiplexing (WDM) channel equalization in current SMF transmission systems are applicable. In uncoupled SDM networks, lightpath crosstalk need to be optimized below a threshold which is related to the modulation format, baud rate, forward error correction (FEC) ability etc [22, 23], otherwise the OSNR penalties caused by the crosstalk will lead to a serious degradation on the Q-factor [24, 25], which makes the bit-error ratio (BER) unacceptable. However, the level of lightpath crosstalk can change during network operations including the setup and teardown of WDM channels or spatial channels transmitted through neighboring cores or different modes, and other environmental fluctuations . If the network controller cannot be aware of the changes of lightpath crosstalk and re-route some lightpaths for crosstalk reduction, the BER of some connections may be unacceptable. Therefore, in-service crosstalk monitoring and tracing is necessary in uncoupled SDM networks.
A method for in-service crosstalk monitoring in point-to-point transmission systems has been reported in literature . This method selects two “pilot tones” (signals close to the either sides of the WDM signal band) for each core to monitor and identify the source of inter-core crosstalk. It can work well in point-to-point MCF transmission systems, but may have problems when extended to be applied to the network scenarios, since two “pilot tones” are required for each connection. Figure 1 illustrates the minimum numbers of required “pilot tones” for crosstalk monitoring under a 20*20 2D-Torus 12-core MCF network under different network loads. It can be observed that hundreds of “pilot tones” are required for crosstalk monitoring and this number will rise with network load, which is difficult to provide. Furthermore, the complex operations on these “pilot tones” also make the method impractical in large-scale networks.
In SDM networks, during the processes of dynamic connection setup and teardown, the accumulated crosstalk of some already provisioned lightpaths may exceeds the threshold when the network state changes, which leads to the failure of traffic transmission. To maintain a high quality-of-transmission (QoT) of each connection, lightpath re-optimization [27, 28] is very important. Lightpath re-optimization was defined as a task consisting of setting a new configuration of lightpaths capable of routing the same existing client connections in the network using a lower (or the lowest) resource utilization . Literatures [28–30] have studied the lightpath re-optimization problem in WDM networks and literatures [31, 32] have studied the spectrum defragmentation problem which is a narrowly defined re-optimization problem in spectrum elastic optical networks (EON). Our previous work  have studied the off-line lightpath re-optimization problem in SDM networks and proposed the concept of fast parallel lightpath re-optimization for the first time to reduce overall crosstalk, but in-service crosstalk information acquisition and crosstalk source tracing were not involved and need further perfection.
In this paper, we proposed a novel in-service crosstalk monitoring method and algorithm using fine-grained optical time slice monitoring channels for crosstalk reduction in SDM optical networks. Benefitting from the large amount of fine-grained channels provided by optical time slices switching (OTSS) [33, 34], it becomes possible for every source node to allocate a dedicated monitoring time slice carrying the traffic and path information for each connection. Crosstalk monitoring and tracing can be realized by extracting the information contained in these monitoring time slices. Simulation results shows that the proposed crosstalk monitoring and tracing method and algorithm can obtain acceptable performance in large-scale network scenarios. Furthermore, we also proposed a QoT-oriented lightpath re-optimization mechanism based on in-service crosstalk monitoring and tracing to maintain a high level of QoT. Finally, we designed a prototype experiment to validate our proposed in-service crosstalk monitoring method. Results show that this method can realize in-service crosstalk monitoring, tracing and lightpath re-optimization over a seven-core fiber based transmission system, and the crosstalk with a minimum value of −37.9 dB can be monitored and successfully traced.
The rest of the paper is organized as follows. Section 2 explicates the concept and process of the proposed in-service crosstalk monitoring, tracing and lightpath re-optimization method. Section 3 introduces the proposed monitoring time slice allocation algorithm and presents the simulation results for the large-scale network scenario. Section 4 describes and discusses the setup and results of a prototype crosstalk monitoring experiment. Finally, Section 5 concludes the paper.
2. In-service crosstalk monitoring, tracing and lightpath re-optimization
To achieve in-service crosstalk monitoring and tracing without the use of an enormous number of “pilot tones”, we allocate a fine-grained dedicated monitoring time slice for each data channel to simulate the crosstalk generated between the monitored data channel and its “neighboring” spatial channels. Figure 2 illustrates the concept of the proposed in-service crosstalk monitoring and tracing method using fine-grained optical time slice monitoring channels in a 4-node SDM network.
A software-defined networking (SDN) controller is responsible for calculating and allocating available time slices for all the fine-grained monitoring channels. The monitoring time slices are transmitted periodically and their period is defined as the monitor period. The allocation of the monitoring time slices should obey the following three principles:
- 1) The monitoring channels and data channels are assigned different wavelengths.
- 2) The monitoring time slices for different traffics occupy different time slots.
- 3) The monitoring time slices are transmitted along the exact same paths (same fiber links, cores and modes, etc.) with their corresponding traffic data.
The monitoring channels can be established by switching the monitoring time slices at each traversing node coordinately at the pre-set time points. Notice that all nodes are required to be time synchronized in the system.
Crosstalk values of monitoring channels can be obtained by measuring the powers of the monitoring time slices. However, since the target of crosstalk monitoring is to measure the crosstalk values of the data channels, the crosstalk differences between monitoring channels and data channels need to be taken into account, which can be calculated according to the theory in work . For a specific MCF, the crosstalk () between core channels can be calculated as the following equation:
The monitoring time slices carry traffic and path information including traffic ID (Traf_ID), source node ID (Src_ID), destination node ID (Dst_ID), hop number (Hop_no), etc. By extracting the information contained in these monitoring time slices, in-service crosstalk monitoring and tracing can be realized.
For instance, Traffic #A and Traffic #B are assigned to Core 1 and Core 2 of the MCF between Node 1 and Node 2, respectively, as shown in Fig. 1. Monitoring time slices #A and #B are transmitted along exactly the same paths with their corresponding traffic data in the monitoring channels. Crosstalk will be generated between Core 1 and Core 2 in both of the monitoring and data channels on Link 1-2. For simplicity, we define “signal time slice” as the monitoring time slice of the traffic in the same spatial channel and “crosstalk time slice” as the monitoring time slices generated by the crosstalk effect. At the receivers of Node 3 (Port 1) and Node 4 (Port 2), “crosstalk time slices” (#B and #A) will be received and the “crosstalk traffic information” can be extracted, making it possible to accurately trace the crosstalk.
To maintain high-quality data transmission, lightpath re-optimization is necessary when the monitored crosstalk of a lightpath exceeds crosstalk threshold. Figure 3 illustrates the overall flow chart of our proposed lightpath re-optimization process.
Before the re-optimization process begins, all the new connection requests need to be rejected until re-optimization is finished. Then, the SDN controller gathers the power of every “crosstalk time slice” and the information contained in it. After that, the controller sorts the crosstalk values of the “crosstalk time slices”. If the largest crosstalk value exceeds the threshold, the corresponding traffic that causes this largest crosstalk will be re-routed following the procedures as below: (1) the SDN controller calculates a re-routing path for the traffic to be re-routed using the routing, wavelength and core allocation (RWCA) algorithm (Table 1, 2) it sends configuration commands to the related traversing nodes, and (3) these nodes execute the configuration commands at the pre-set time and achieve the lightpath re-routing coordinately (detailed re-routing mechanism have been proposed in our previous work ). Then, the SDN controller will allocate a new time slot for the monitoring time slice of the re-routed traffic and send a message which contains new path information to the source node. Upon receiving the message, the source node will send out a new monitoring time slice carrying the new path information along the re-routing path. After that, the SDN controller will update the crosstalk database and sorts the crosstalk values of the time slices again. The SDN controller will continue to trigger a new lightpath re-routing process until the largest crosstalk value is below the crosstalk threshold.
: The network topology, N, E, W and C represent the number of nodes, links wavelengths and cores respectively.
: A link occupying core/mode c and wavelength w.
: The lightpath which induces the largest crosstalk with the source and destination nodes of s and d.
: Inter-core crosstalk between core c1 and core c2 on link
: Inter-core crosstalk values of the network
: The lightpath after re-routing.
3. Monitoring time slice allocation and network simulation
3.1 Monitoring time slice allocation algorithm
To achieve the proposed in-service crosstalk monitoring and tracing mechanism in SDM networks, we propose a monitoring time slice allocation (MTSA) algorithm to allocate a monitoring time slice for each traffic in the network. Different from the traditional time slice allocation algorithms mentioned in [33, 34], the effect of “crosstalk time slices” are taken into consideration in our proposed algorithm.
To simplify the problem of MTSA, we introduce the concept of “crosstalk order”, which describes the number of cascading crosstalk effect that occurs. To make it clearer, an example of first- and second-order crosstalk is given in Figs. 4(a)-4(c) under a three-node MCF network. Traffic #B generates crosstalk with Traffic #A and Traffic #C on link 1-2 and link 2-3 respectively. At the receivers of node 3, there will be three time slices in the monitoring channel of Core 2 as shown in Fig. 4(b). Two first-order “crosstalk time slices” (#A and #C) are generated by the crosstalk with the monitoring time slices of Traffic #A and Traffic #C. The monitoring channel of Core 3 also has three time slices as shown in Fig. 4(c). The second-order “crosstalk time slice” #A is generated by the crosstalk with the first-order “crosstalk time slice” #A in the monitoring channel of Traffic #B.
In real SDM networks, the second- and higher-order crosstalk can be ignored compared with the first-order crosstalk (e.g. first-order crosstalk is −30dB and second-order crosstalk is −60dB). Based on this consideration, the proposed MTSA algorithm only considers about the signal and the first-order crosstalk time slices. The algorithm flowchart of MTSA is described in Fig. 4(d). Firstly, the MTSA algorithm seeks all the occupied time slots on the paths of the first-order crosstalk time slices. Then, it combines the occupied time slots on the paths of the signal and first-order crosstalk time slices. At last, it allocate a time slot for the monitoring time slice from the unoccupied time slots and update the monitoring time slice database.
Since propagation delay exists, the time slots allocated on different links will have different but closely related start time, which makes it necessary to consider the resource state of the whole path. To seek all the occupied time slots on a path, we propose a time slices combination algorithm (TSC) as shown in Table 2, and Table 3 presents the proposed MTSA algorithm based on the TSC algorithm.
: A link occupying core/mode c and wavelength w.
: Set of links along source and destination nodes, .
: A part of the lightpath, .
: The start time of the mth time slice on link .
: The end time of the mth time slice on link .
: The total number of time slices on link .
: The set of all time slices on link , .
: The propagation time of link .
: The network topology, N and E represent the number of nodes and links respectively.
: The set of all time slices on all links.
: The propagation time of all links.
: The path of the ith traffic, .
: The path of the target traffic which need to allocate monitoring time slice, .
: The path set of all the traffic except the target traffic, .
: The set of links which generate crosstalk with link ,
: Required time length of a monitoring time slice.
: All time slots in a monitor period.
3.2 Network simulations and discussions
To evaluate the proposed MTSA algorithm, we conduct simulations under a 20x20 2D-Torus topology  as shown in Fig. 5. The simulation is conducted on a Dell PowerEdge R730 server, which has two 2.5GHz Intel Xeon E5-2680v3 CPUs and 32GB RAM in total. The number of nodes and links are 400 and 1600, respectively. All of the nodes have a degree of 4 and the length of each link is set to 100km. The numbers of fibers, cores and wavelengths on each physical link are set to 1, 12, and 2, respectively. Monitoring channel and data channel occupy one wavelength each. The crosstalk between two adjacent cores is set to −46.92dB per 50km according to literature . The propagation delay of each link is 500μs. Traffic connections are distributed uniformly among all the node pairs with a fixed bandwidth demand of one wavelength. The length of a monitoring time slice is set to 100μs.
The number of required time slots will increase when the network load increases, which will lead to a longer monitor period and a lower monitor sampling frequency. Figure 6(a) presents the highest sampling frequency under different network loads. When the network load increases from 200 Erlang to 1000 Erlang, the highest monitor sampling frequency decreases from 55 Hz to 30Hz. It can be seen that the increasing of network load degrades the timeliness of in-service monitoring.
To promote the timeliness of in-services monitoring, an effect method is to set a crosstalk threshold in MTSA to improve the monitor sampling frequency by ignoring the impact of the transmission of weak crosstalk time slices and reducing their occupation of time slot resources. More specifically, only when the estimated crosstalk value of a crosstalk time slice exceeds the threshold, we allocate a time slot for it. In this way, the destination nodes can only receive crosstalk time slices which are beyond the threshold. Figure 6(b) illustrates the results of the highest monitor sampling frequencies that can be achieved at different crosstalk threshold values under the network load of 1000 Erlang. It can be seen that when the threshold is larger than −34.88dB (the sweet spot), the highest monitor sampling frequency almost cannot be improved any further. The sweet spot reflects the optimal balance between the monitoring timeliness and the monitoring capability. And we can see that the 102 Hz sampling frequency at the sweet spot is adequate for practical use.
4. Experimental setup, results and discussions
A testbed is setup to evaluate the performance of the proposed in-service crosstalk monitoring, tracing and lightpath re-optimization mechanism.
4.1 Experimental setup
As shown in Fig. 7, a 200-m high-crosstalk seven-core fiber is applied to simulate long-distance uncoupled MCF transmission. Three traffic (#TA ~#TC) are transmitted in 1546.4-nm data channel by 10Gb/s small form-factor pluggable (SFP + ) transceiver modules on three Dell PowerEdge R730 servers. Three monitoring time slices (#A ~#C) transmitted in 1561.8-nm monitoring channels are generated by 10G SFP + on the FPGA (Xilinx Virtex-7 VC709 evaluation board). The monitor period is set to 1 ms and the time length of each monitoring time slice is set to 100 μs. The threshold of lightpath accumulated crosstalk is set to −30dB according to literatures [24, 25]. An SDN controller is responsible for dispatching configuration commands to six magneto-optic switch (MO_SW) controllers, and switching operations are triggered by MO_SW controllers at the pre-set precise time. MO_SW1 is used to switch the three time slices into different cores, MO_SW2 ~MO_SW5 are used to switch the core channels and MO_SW6 is used to extract the desired time slice to obtain the crosstalk information of the corresponding monitoring channel (core). Besides, we also use a 1x4 switch (SW7) to select the channel (core) to be monitored. A FiberHome wavelength selective switch (WSS) is used to demultiplex data and monitoring channels. A power meter (HP 81531A) is deployed to measure the powers of the monitoring time slices and give feedback to the SDN controller. A digital storage oscilloscope (Agilent 86100C DSO) and an optical spectrum analyzer (Advantest Q8384 OSA) are used to measure signals at different ports.
The core allocation of the three traffic and their monitoring time slices are depicted in Fig. 7. Initially, the core allocation is in “State 1”, traffic #TA and time slice #A occupy Core 6 and the other two traffic together with their monitoring time slices occupy its two adjacent cores (Core 1 and Core 5). After lightpath re-optimization, traffic #TC switches to Core 2 and traffic #TB switches to Core 4 together with their monitoring time slices.
4.2 Crosstalk value compensation between monitor and data channels
Figure 8(a) presents the optical power spectrum at Port 1. The center wavelength of data channel and monitoring channel are 1546.4nm and 1561.8nm respectively. The bandwidth of the two channels are 0.8nm at −70dBm.
To compensate the crosstalk value differences between data and monitoring channels, we measured the crosstalk values between Core 1 and Core 6 under different ITU-T G.694.1 DWDM wavelengths as shown in Fig. 8(b). According to the theory of linear correlation between crosstalk and wavelength in literature , by linear fitting the measurement results, it can be seen that the linearly dependent coefficient (R) between crosstalk and wavelength is 0.992. Besides, we also measured the crosstalk values between any two cores of the MCF at the wavelength of 1546.4nm and 1561.8nm, the largest crosstalk difference is only 0.7dB. In this way, the crosstalk value differences between data and monitoring channels are compensated in our experiment.
4.3 In-service crosstalk monitoring and tracing
Figures 9(a)-9(g) present the DSO screen shots of the seven core monitoring channels at Port 2 and Fig. 9(h) gives the screen shot of time slice #C at Port 5 (after time slice extracted and amplified). The time length and power of the three monitoring time slices in Core 6 are measured by the DSO at Port 2 and the power meter at Port 4 (after time slice extraction) respectively. The main measured parameters of the received monitoring time slices in Core 6 are presented in Table 4. In this measurement, time slice #A is the signal and the other two (#B and #C) are the crosstalk. The power meter sends the power values of the three time slices to the SDN controller as the crosstalk power feedback.
These monitoring time slices have also been received successfully by the FPGA placed at Port 5. After packet parsing through the FPGA, traffic and path information contained in time slices #B and #C are extracted correctly and the results are shown in Fig. 10. The FPGA sends these information to the SDN controller as the crosstalk information feedback.
4.4 Lightpath re-optimization
After gathering the power and information feedbacks of the crosstalk time slices (#B and #C), the SDN controller calculates the crosstalks of traffic #TB and #TC, which are −10.6dB and −11.3dB respectively after crosstalk difference compensation between the data and monitoring channels. Since both of the crosstalks exceed the −30dB threshold, the SDN controller then calculates re-routing paths for traffic #TB and #TC, and triggers lightpath re-optimization process. Figure 11 presents the DSO screen shots of the data channels during lightpath re-optimization process. The green waveforms represent the control signal of MO_SWs, and the yellow waveforms represent the optical signal of different core channels at Port 3. The switching times of traffic #TB and #TC are 33.6μs and 35.9μs respectively.
The 10-s average traffic bitrates (TB) are measured by Iperf3 software on the servers. As shown in Fig. 12, none of the three traffic (#TA ~#TC) can be transmitted originally (core allocation as “State 1”), but all of them can reach maximum bitrates after lightpath re-optimization (core allocation as “State 2”).
4.5 Crosstalk monitoring and tracing after re-optimization
After lightpath re-optimization, the crosstalk values induced by time slice #B and #C are −31.5dB and −37.9dB, respectively. The information contained in these time slices are successfully extracted. Figure 13(a) presents the waveform of Core 6 at Port 2 and Fig. 13(b) presents the waveform of time slice #C after time slice extraction (Port 5). After packet parsing, the traffic and path information contained in time slice #C can be successfully extracted as shown in Fig. 13(c). This results would indicate that the small crosstalk of −37.9 dB (between Core 2 and Core 6) can be successfully traced.
In-service crosstalk monitoring and lightpath re-optimization are important for maintaining a high quality of traffic transmission and an efficient use of resources in SDM networks. In this paper, we propose a novel in-service crosstalk monitoring, tracing and lightpath re-optimization method for crosstalk reduction in SDM optical networks using fine-grained optical time slice monitoring channels. To achieve this, we allocate a dedicated monitoring time slice which contains the traffic and path information for each traffic, and send it along the exact same path (at different wavelength) with their corresponding traffic data. Crosstalk monitoring and tracing can be realized by extracting the information contained in these monitoring time slices. A novel lightpath re-optimization mechanism based on in-service crosstalk monitoring and tracing is also proposed to improve the quality of traffic transmission.
Besides, we propose a monitoring time slice allocation (MTSA) algorithm to achieve in-service crosstalk monitoring. To evaluate the performance of the MTSA algorithm, we conduct a large-scale network simulation and analyze two main impact factors, i.e., network load and crosstalk threshold for monitoring. The results indicate that we can set an appropriate crosstalk threshold for monitoring to improve the monitor sampling frequency under heavy network load.
Finally, we build a testbed to validate the in-service crosstalk monitoring, tracing and lightpath re-optimization method. The results show that our proposed method can realize in-service monitoring, tracing and lightpath re-optimization over a seven-core fiber based transmission system, and the crosstalk as low as −37.9 dB can be monitored and traced.
For future works, we will expand the proposed in-service crosstalk monitoring and tracing method to support spectrum the elastic optical network (EON), since it might replace the traditional WDM network in the future.
Program 973 (2014CB340104/05); National Natural Science Foundation of China (NSFC) (61621064, 61435006).
References and links
1. S. K. Korotky, “Price-points for components of multi-core fiber communication systems in backbone optical networks,” J. Opt. Commun. Netw. 4(5), 426–435 (2012).
2. R. W. Tkach, “Scaling optical communications for the next decade and beyond,” Bell Labs Tech. J. 14(4), 3–9 (2010).
3. Y. Lee, K. Tanaka, E. Nomoto, H. Arimoto, and T. Sugawara, “Multi-core fiber technology for optical-access and short-range links,” in Proceedings of 12th Int. Conf. Opt. Internet (IEEE, 2014), pp. 1–2.
4. M. Fiorani, M. Tornatore, J. Chen, L. Wosinska, and B. Mukherjee, “Spatial division multiplexing for high capacity optical interconnects in modular data centers,” J. Opt. Commun. Netw. 9(2), A143–A153 (2017).
5. R. J. Essiambre, G. Karmer, P. J. Winzer, G. J. Foschini, and B. Goebel, “Capacity limits of optical fiber networks,” J. Lightwave Technol. 28(4), 662–701 (2010).
6. G. Li, N. Bai, N. Zhao, and C. Xia, “Space-division multiplexing: the next frontier in optical communication,” Adv. Opt. Photonics 6(4), 413–487 (2014).
7. P. J. Winzer, “Spatial multiplexing: the next frontier in network capacity scaling,” in Proceedings of IET Conference, 2013, paper We.1.D.1.
8. P. J. Winzer, “Making spatial multiplexing a reality,” Nat. Photonics 8(5), 345–348 (2014).
9. L. Xiang, G. Shen, and M. Gao, “Nonlinear propagation in multicore fiber transmission link based on coupled mode analysis,” in Asia Communications and Photonics Conference (Optical Society of America, 2015), paper ASu2A.76.
10. J. Sakaguchi, B. J. Puttnam, W. Klaus, Y. Awaji, N. Wada, A. Kanno, T. Kawanishi, K. Imamura, H. Inaba, K. Mukasa, R. Sugizaki, T. Kobayashi, and M. Watanabe, “19-core fiber transmission of 19×100×172-Gb/s SDM-WDM-PDM-QPSK signals at 305Tb/s,” in Optical Fiber Communication Conference (Optical Society of America, 2012), paper PDP5C.1.
11. T. Mizuno, T. Kobayashi, H. Takara, A. Sano, H. Kawakami, T. Nakagawa, Y. Miyamoto, Y. Abe, T. Goh, M. Oguma, T. Sakamoto, Y. Sasaki, I. Ishida, K. Takenaga, S. Matsuo, K. Saitoh, and T. Morioka, “12-core x 3-mode dense space division multiplexed transmission over 40 km employing multi-carrier signals with parallel MIMO equalization,” in Optical Fiber Communication Conference (Optical Society of America, 2014), paper Th5B.2.
12. R. Ryf, S. Randel, A. H. Gnauck, C. Bolle, R. Essiambre, P. Winzer, D. W. Peckham, A. McCurdy, and R. Lingle, “Space-division multiplexing over 10 km of three-mode fiber using coherent 6 × 6 MIMO processing,” in Optical Fiber Communication Conference (Optical Society of America, 2011), paper PDPB10.
13. S. Randel, R. Ryf, A. Sierra, P. J. Winzer, A. H. Gnauck, C. A. Bolle, R. J. Essiambre, D. W. Peckham, A. McCurdy, and R. Lingle Jr., “6×56-Gb/s mode-division multiplexed transmission over 33-km few-mode fiber enabled by 6×6 MIMO equalization,” Opt. Express 19(17), 16697–16707 (2011). [PubMed]
14. Y. Li, N. Hua, and X. Zheng, “CapEx advantages of multi-core fiber networks,” Photonic Netw. Commun. 31(2), 228–238 (2016).
15. Y. Li, N. Hua, and X. Zheng, “Routing, wavelength and core allocation planning for multi-core fiber networks with MIMO-based crosstalk suppression,” in Proceedings of Opto-Electronics and Communications Conference (IEEE, 2015), pp. 1–3.
16. N. Cvijetic, E. Ip, N. Prasad, M. Li, and T. Wang, “Experimental time and frequency domain MIMO channel matrix characterization versus distance for 6×28Gbaud QPSK transmission over 40×25km few mode fiber,” in Optical Fiber Communication Conference (Optical Society of America, 2014), paper Th1J. 3.
17. R. Ryf, S. Randel, A. H. Gnauck, C. Bolle, A. Sierra, S. Mumtaz, M. Esmaeelpour, E. C. Burrows, R. Essiambre, P. J. Winzer, D. W. Peckham, A. McCurdy, and R. Lingle, “Mode-division multiplexing over 96 km of few-mode fiber using coherent 6×6 MIMO processing,” J. Lightwave Technol. 30(4), 521–531 (2012).
18. P. Sillard, M. Astruc, D. Boivin, H. Maerten, and L. Provost, “Few-mode fiber for uncoupled mode-division multiplexing transmissions,” in 37th European Conference and Exposition on Optical Communications (Optical Society of America, 2011), paper Tu.5.LeCervin.7.
19. E. Ip, G. Milione, M. J. Li, N. Cvijetic, K. Kanonakis, J. Stone, G. Peng, X. Prieto, C. Montero, V. Moreno, and J. Liñares, “SDM transmission of real-time 10GbE traffic using commercial SFP + transceivers over 0.5km elliptical-core few-mode fiber,” Opt. Express 23(13), 17120–17126 (2015). [PubMed]
20. T. Mizuno, K. Shibahara, H. Ono, Y. Abe, Y. Miyamoto, F. Ye, T. Morioka, Y. Sasaki, Y. Amma, K. Takenaga, S. Matsuo, K. Aikawa, K. Saitoh, Y. Jung, D. J. Richardson, K. Pulverer, M. Bohn, and M. Yamada, “32-core dense SDM unidirectional transmission of PDM-16QAM signals over 1600 km using crosstalk-managed single-mode heterogeneous multicore transmission line,” in Optical Fiber Communication Conference (Optical Society of America, 2016), paper Th5C.3.
21. T. Hu, J. Li, P. Zhu, Q. Mo, Y. Ke, C. Du, Z. Liu, Y. He, Z. Li, and Z. Chen, “Experimental demonstration of passive optical network based on mode-division-multiplexing,” in Optical Fiber Communication Conference (Optical Society of America, 2015), paper Th2A.63.
22. A. Sano, H. Takara, T. Kobayashi, and Y. Miyamoto, “Crosstalk-managed high capacity long haul multicore fiber transmission with propagation-direction interleaving,” J. Lightwave Technol. 32(16), 2771–2779 (2014).
23. P. Winzer, A. Gnauck, A. Konczykowska, F. Jorge, and J. Dupuy, “Penalties from in-band crosstalk for advanced optical modulation formats,” in 37th European Conference and Exposition on Optical Communications (Optical Society of America, 2011), paper Tu.5.B.7.
24. T. Hayashi, T. Taru, O. Shimakawa, T. Sasaki, and E. Sasaoka, “Ultra-low-crosstalk multi-core fiber feasible to ultra-long-haul transmission,” in Optical Fiber Communication Conference (Optical Society of America, 2011), paper PDPC2.
25. T. Hayashi, T. Taru, O. Shimakawa, T. Sasaki, and E. Sasaoka, “Design and fabrication of ultra-low crosstalk and low-loss multi-core fiber,” Opt. Express 19(17), 16576–16592 (2011). [PubMed]
26. T. Mizuno, A. Isoda, K. Shibahara, Y. Miyamoto, S. Jain, S. Alam, D. J. Richardson, C. Castro, K. Pulverer, Y. Sasaki, Y. Amma, K. Takenaga, K. Aikawa, and T. Morioka, “In-service crosstalk monitoring for dense space division multiplexed multi-core fiber transmission systems,” in Optical Fiber Communication Conference (Optical Society of America, 2017), paper M3J.2.
27. R. Luo, N. Hua, X. Zheng, and B. Zhou, “Fast parallel lightpath re-optimization for space-division multiplexing optical networks based on time synchronization,” J. Opt. Commun. Netw. 10(1), A8–A19 (2018).
28. F. Solano and M. Pióro, “Lightpath reconfiguration in WDM networks,” J. Opt. Commun. Netw. 2(12), 1010–1021 (2010).
29. E. Bouillet, J. F. Labourdette, R. Ramamurthy, and S. Chaudhuri, “Lightpath re-optimization in mesh optical networks,” IEEE/ACM Transactions on Networking (TON) 13(2), 437– 447 (2005).
30. J. Ahmed, F. Solano, P. Monti, and L. Wosinska, “Traffic re-optimization strategies for dynamically provisioned WDM networks,” in Preecdings of Optical Network Design and Modeling Conference (IEEE, 2011), pp. 1–6.
31. P. Zhu, J. Li, D. Wu, Y. Chen, Y. Tian, Z. Wu, D. Ge, X. Chen, Z. Chen, and Y. He, “Software-defined elastic optical network node supporting spectrum defragmentation,” J. Opt. Commun. Netw. 9(1), A63–A70 (2017).
32. T. Takagi, H. Hasegawa, K. Sato, Y. Sone, A. Hirano, and M. Jinno, “Disruption minimized spectrum defragmentation in elastic optical path networks that adopt distance adaptive modulation,” in 37th European Conference and Exposition on Optical Communications (Optical Society of America, 2011), paper Mo.2.K.3.
33. N. Hua and X. Zheng, “Optical time slice switching (OTSS): an all-optical sub-wavelength solution based on time synchronization,” in Asia Communications and Photonics Conference 2013 (Optical Society of America, 2013), paper AW3H.3.
34. Y. Li, N. Hua, and X. Zheng, “Fine-grained all-optical switching based on optical time slice switching for hybrid packet-OCS intra-data center networks,” in Optical Fiber Communication Conference (Optical Society of America, 2016), paper W3J.5.
35. N. Benzaoui, Y. Pointurier, and S. Bigo, “Latency in a 2D torus burst optical slot switching data center,” in Optical Fiber Communication Conference (Optical Society of America, 2017), paper Th2A.33.
36. F. Ye, J. Tu, K. Saitoh, K. Takenaga, S. Matsuo, and T. Morioka, “A new and simple method for crosstalk estimation in homogeneous trench-assisted multi-core fibers,” in Asia Communications and Photonics Conference 2014 (Optical Society of America, 2014), paper AW4C.3.