Open Access
Add to BibSonomyAbstract
We propose a novel simple space division multiplexing (SDM) node which is rearrangeble nonblocking, and effectively utilizes enhanced network resources through SDM. The proposed node can reduce a number of ports of wavelength selective switches (WSSs) and a number of WSS modules by modifying conventional multi-stage switches and employing integrated multiple arrayed WSSs. We experimentally actualized the newly proposed node, and demonstrate wavelength, core, and direction switching functions based on 127-Gbps Dual Polarization Quadrature Phase Shift Keying (DP-QPSK) signals. We also confirm the feasibility of the proposed SDM node through SDM transmission experiments using a 40-km multicore fiber and a multicore amplifier.
© 2017 Optical Society of America
1. Introduction
In the forthcoming 5G mobile era, the amount of data traffic is expected to grow exponentially due to the rapid increase in the amount of mobile and video traffic. To meet this traffic demand, a space division multiplexing (SDM) system with multicores and multimodes has been extensively investigated [1]. Another marked trend is network virtualization employing software-defined networking (SDN)/network function virtualization (NFV) [2] that aims at the actualization of a programmable network. The next-generation network including optical transport will need to become simpler as well as more flexible.
In general, when M is the number of directions and K is the number of spatial modes, an SDM node can be constructed using a KM x KM switch [3]. Spatial channels can be transmitted using various optical fibers such as multicore fibers (MCFs), multiple single-mode fibers (SMFs), and multimode fibers (MMFs). These channels are multiplexed and switched using cores/modes/fibers. Two possible implementations have been conventionally employed: the use of a large matrix switch such as an optical cross-connect (OXC) and route-and-select (R&S) switching. R&S switching utilizes two-stage wavelength selective switches (WSSs), in which the signals are selectively routed to output WSSs by the first stage WSS, and selectively passed to output ports by the second stage WSS [3]. The OXC can be expensive as well as cause a single-point failure, but R&S switching is a reasonable solution. However, to achieve a large matrix switch using WSSs, very high port-count WSSs are required. It is known that we can reduce the port count of WSSs by employing a multi-stage switch configuration [4]. However, this configuration tends to increase the number of WSS modules and insertion loss (IL).
To simplify the configuration of SDM nodes, previous studies on SDM systems have tried to divide spatial modes and/or wavelengths into subgroups [3]. Optical paths are switched on a fiber basis [5,6] and/or on a wavelength group basis [7]. For a large-scale node architecture, hierarchical multi-granular routing using a fiber switch [8] and interconnected sub-system architecture [9] was proposed to avoid the use of high port-count switches. However, the same issues arise when the number of wavelengths and/or spatial modes becomes large. Therefore, full-functional SDM nodes that enable switching of wavelengths, cores, and directions have not yet been proposed, although such degrees of freedom are necessary to transport flexibly large-scale data traffic.
In this paper, we propose for the first time a novel and simple SDM node in which wavelengths, cores, and directions are switched to utilize enhanced network resources through SDM. The proposed SDM node consists of three-stage WSSs with a practical port size and therefore the number of switch elements is drastically reduced. We also experimentally show the SDM node using multiple-arrayed WSSs, and successfully show core/direction switching and a channel add/drop function by transmitting 127-Gbps Dual Polarization Quadrature Phase Shift Keying (DP-QPSK) signals.
2. SDM node architecture
2.1. Large scale node architecture
In this section, we consider an SDM node architecture. We focus here on SDM using MCFs and/or multiple SMFs so that wavelength channels and spatial channels can be independently utilized. Note that for the sake of simplicity, hereafter we describe only SDM nodes using MCFs while it is clear that the same discussion can be applied to those using multiple SMFs by viewing each SMF as each core in an MCF. Figure 1 shows an overview of SDM nodes. An SDM node consists of a wavelength cross-connect (WXC) function and an add/drop function. The WXC function supports the exchanging of optical paths and the amplification of transmitted signals. The add/drop function supports the assigning of an optical signal from an add/drop port to the WXC function, in which add/drop ports are typically connected to transponders. In an SDM network, the transmission capacity is enhanced by wavelength division multiplexing (WDM) and SDM, so the WXC and add/drop functions should have a switching function to utilize efficiently this enhanced capacity.
To achieve an SDM node that utilizes this enhanced capacity, there are two typical SDM node architectures: (1) a matrix-switch based configuration and (2) a multi-stage based switch configuration. Figure 2 shows an SDM node example based on a matrix switch. To assume K cores, K input/output ports of an SDM node are grouped as a port group as shown in Fig. 2. When an SDM node has M ports groups, a KM x KM WSS is required as shown in Fig. 2(a). In this figure, each of M’ port groups in M port groups connects to each direction, and the other L port groups are used for add/drop ports. Then this node has LK add/drop ports, and the add/drop ratio becomes LK/(M-L)KW, where W is the number of wavelengths. We can achieve this matrix switch by employing an R&S switch configuration, which comprises multiple 1 x KM WSSs as shown in Fig. 2(b). In a typical colorless, directionless, and contentionless reconfigurable optical add/drop multiplexer (CDC-ROADM), multicast switches (MCSs) and tunable filters are used for the add/drop function, which requires additional optical amplifiers to compensate for the large loss of MCSs. We assume that WSSs are used instead of MCSs and tunable filters to simplify the following discussion. In this configuration, the required number of 1 x KM WSSs becomes 2 KM. SDM nodes in this configuration require high port-count WSSs as M becomes large. For instance, when M is 8 and K is 20, 320 sets of 1 x 160 WSSs are required. It is difficult to actualize a high port-count WSS such as 1 x 160.
Fig. 2 (a) Concept of SDM node based on matrix switch. (b) Practical example of SDM node using R&S switch configuration.
Figure 3(a) shows a three-stage rearrangeable nonblocking SDM node as an example of the multi-stage based switch configuration [4]. The first and third stages have M sets of K x K WSSs, and the second stage has K sets of M x M WSSs. In this figure, each of M’ port groups connects to each direction, and the other L port groups are used for add/drop ports. Figure 3(b) shows a configuration that is changed compared to that in Fig. 3(a) in a manner similar to that in Fig. 2(b), in which K x K (M x M) WSSs are replaced with K x 1 (M x 1) WSSs and 1 x K (1 x M) WSSs. The required number of WSSs becomes 2MK at each stage, so the total required number of WSSs becomes 6MK. When we assume that M = 8 and K = 20, this configuration can be actualized by a 1 x 8 or 1 x 20 WSS, and 960 sets of WSSs. This configuration can be actualized by low port-count WSS; however, it requires a large number of WSSs.
Fig. 3 (a) Concept of SDM node based on multi-stage switches. (b) Practical example of SDM node employing R&S switch configuration at each stage.
The matrix switch based configuration in Fig. 2 is strictly nonblocking, that is, this node operates in a contentionless manner. On the other hand, the three-stage rearrangeable nonblocking SDM node in Fig. 3 is not strictly nonblocking. However, these two nodes exhibit nearly the same performance regarding the number of established optical paths when one of the directions is fully used and a new additional node must be installed, which is discussed in detail in Section 2.4. Moreover, even though blocking occurs in rearrangeable nonblocking SDM nodes, blocking can be addressed by rearranging fewer than M-1 optical paths [10]. This rearrangement can be actualized with hitless operation by switching a working path to a reserved path in a planned way in a packet-based network (such as a router network or L2 switch network). Therefore, contentionless operation is not necessary to achieve SDM nodes under the above condition.
These configurations have both advantages and disadvantages. Advantages and disadvantages for the matrix switch based SDM node shown in Fig. 2(b) are given below.
- (Adv-1) It has a simple two-stage configuration.
- (Adv-2) The same performance in terms of IL, filtering bandwidth, and crosstalk as that for conventional R&S configuration ROADMs is achieved if the high port-count WSSs have the same IL, filtering bandwidth, and crosstalk as those of conventional WSS.
- (DA-1) It requires high port-count WSSs.
- (DA-2) It requires complex control and/or adjustment to the optical signals due to the increase in the number of WSS ports.
The advantages and disadvantages for the three-stage SDM node shown in Fig. 3(b) are given below.
- (DA-3) It requires more modules (WSSs) than the matrix switch based SDM node.
- (DA-4) The IL, filtering bandwidth, and crosstalk tend to degrade due to the multi-stage configuration.
- (Adv-3) It is achieved with low port-count WSSs.
- (Adv-4) It facilitates easy control and/or adjustment of optical signals due to the reduced number of WSS ports.
Thus, both typical configurations face technical issues that must be addressed to actualize the SDM node.
2.2. Proposed SDM node
In this section, we describe the proposed SDM node that addresses the issues described in the previous section. First, we describe the desirable features of the SDM node that utilize efficiently the transmission capacity enhanced by WDM and SDM. They are listed below.
- (i) Optical signals can be switched to utilize efficiently the enhanced capacity.
- (ii) Optical signals can be added to any wavelength and direction. It is acceptable that cores cannot be selected as far as the enhanced capacity is efficiently utilized.
- (iii) Optical signals can be dropped from any wavelength, core, or direction.
- (iv) The switching function block in each direction is physically separated. The switching function block can be extended or replaced by a unit of direction so that it does not affect the signals in other directions [11].
Figure 4(a) shows the basic concept. The proposed SDM node is based on the three-stage rearrangeable nonblocking switch in Fig. 3(a). Compared to the configuration in Fig. 3(a), the WSSs in the third stage except for that to the drop ports are removed. We remove the third-stage WSSs because the third-stage WSSs and the first-stage WSSs at the next SDM node have the same switching function. By removing the third-stage WSSs, we can no longer select cores at the output of the SDM node. However, through a combination usage of the first and second-stage WSSs and the first-stage WSSs at the next node, we achieve a three-stage switch configuration. Therefore, the configuration can utilize efficiently resources enhanced through SDM.
Fig. 4 (a) Concept of proposed SDM node. (b) Example of SDM node employing R&S switch configuration at each stage. (c) Example of SDM node by interchanging the locations of WSSs compared to (b).
In a similar manner, for an add function, the configuration can efficiently utilize resources. For a drop function as in Fig. 4(a), the third-stage WSSs are retained, and then signals can be dropped from any core to any drop port. In these ways, this node satisfies desirable features (i) to (iii). We describe satisfying desirable feature (iv) later.
Figure 4(b) shows an example of the SDM node in Fig. 4(a) when K x K WSSs and M x M WSSs are replaced with R&S switches. As explained in relation to Fig. 3(b), K x K WSSs (M x M WSSs) comprise K x 1 WSSs and 1 x K WSSs (M x 1 WSSs and 1 x M WSSs) in an R&S configuration. We then transform the configuration as shown in Fig. 4(c) by interchanging the locations of the 1 x M WSSs and the M x 1 WSSs in the second stage in Fig. 4(b). In this figure, 1 x M WSSs in the second stage are arranged in the same order as the connected K x 1 WSSs in the first stage. In the same way, the M x 1 WSSs in the second stage are arranged in the same order as the output core. As a result, we can group the WSSs that are connected in the same direction, that is, the switching function in each direction can be separated. Through this transformation, we satisfy desirable feature (iv) as mentioned above.
Figure 5(a) depicts the final configuration of the proposed SDM node. In this figure, each of M’ port groups connects to each direction, and the other L port groups are used for add/drop ports. We consider two more ideas for the configuration in Fig. 5(a). One is to replace the WSSs connected back-to-back with a single module. Looking at the right side of the first stage and the left side of the second stage of Fig. 4(c), the K x 1 WSS and 1 x M WSS are connected back-to-back. We can replace these WSSs with a single K x M contention WSS. A K x M contention WSS can be actualized by changing the port usage of a 1 x N WSS with N≥ (K + M). Figure 5(b) shows the function of the K x M contention WSS, in which a K x 1 WSS and 1 x M WSS are connected back-to-back. Figure 5(c) shows an example of the 1 x N WSS configuration. Figure 5(d) shows a K x M contention WSS that can be actualized by changing some output ports to input ports using the same module as the 1 x N WSS. Then, the K x M contention WSS can switch different wavelength signals independently; however, it cannot switch more than one of the same wavelength signals (contention). The IL and passband of the K x M contention WSS are the same as those for the 1 x N WSS.
Fig. 5 (a) Proposed SDM node, (b) Function of K x M contention WSS. (c) Example of 1 x N WSS configuration. (d) Example of K x M contention WSS configuration.
The other idea is to employ multiple arrayed WSSs that integrate multiple WSSs into a single optics system [12,13] to decrease further the number of modules. Focusing on the switching function in each direction, K sets of 1 x K WSSs, K sets of K x M contention WSSs, and K sets of M x 1 WSSs are independently allocated in Fig. 5(a). Each K set of modules can be integrated into multiple arrayed WSSs. These integrated modules are shown in blue in Fig. 5. These two ideas drastically reduce the number of modules.
Here, we describe the IL of the proposed SDM nodes. The through signal passes through three WSSs, and the drop signal passes through four WSSs. When the IL of the WSSs is assumed to be 5 dB, the IL of the through path and that for the drop path become 15 dB and 20 dB, respectively. The IL of the conventional 8-degree ROADM is, for instance, 20 dB [14]. Therefore, we anticipate that the proposed SDM node will not require additional amplifiers.
2.3. Comparison of SDM nodes
In the previous section, we described that the proposed SDM node satisfies the desirable features and addresses the issues of the conventional SDM nodes. To confirm the discussion quantitatively, we compare the SDM nodes in terms of the numbers of total WSS ports and WSS modules. These numbers provide an indication of the complexity and cost of the SDM nodes. An increase in the number of WSS ports leads to difficult control and adjustment of optical signals. An increase in the number of WSS modules leads to an increase in the number of parts as well as that in the power consumption in SDM nodes. Note that a substantial amount of power is required to control the temperature of Liquid Crystal on Silicon (LCoS) based WSSs. These factors are directly linked to the cost of the SDM nodes.
Figure 6 shows the number of total WSS ports in each type of SDM node as a function of the number of cores. The matrix switch based SDM node shown in Fig. 2(b), the three-stage rearrangeable nonblocking SDM node shown in Fig. 3(b), and the proposed SDM node shown in Fig. 5(a) are considered. As described in Section 2.1, M is the number of directions and K is the number of cores. In the matrix switch based SDM node and the three-stage rearrangeable nonblocking SDM node, the add/drop ports have the same configuration as that for the input/output ports. On the other hand, for the proposed SDM node, the add ports have the same configuration as that for the input ports, but the drop ports have a different configuration from the output ports. For the drop ports, L sets of M x K contention WSSs and K x 1 WSSs are placed to select the required drop channels. The number of WSS ports is defined as follows. A 1 x K WSS has K ports, and a K x K contentionless WSS comprises K sets of 1 x K WSSs and K sets of K x 1 WSSs. Therefore, it has 2K2 ports. A K x M contention WSS comprises a K x 1 WSS and a 1 x M WSS. So it has K + M ports.
In Fig. 6, the total number of WSS ports for the matrix switch based SDM node is represented by the dotted line, that for the three-stage rearrangeable nonblocking SDM node is represented by the dashed line, and that for the proposed SDM node is represented by the solid line. The total number of WSS ports for the considered SDM nodes is also shown in the inset to Fig. 6. We assume here that M = 8 and L = 2. When K = 20, the total number of WSS ports for the matrix switch based SDM node is 51200, that for the 3-stage rearrangeable nonblocking SDM node is 15360, and that for the proposed SDM node is 10560. These results show that compared to the matrix switch based SDM node, the proposed SDM node yields the reduction of 79% in the total number of WSS ports.
Figure 7 illustrates the number of WSS modules for the considered SDM nodes as a function of the number of cores. Similarly, the dotted, dashed, and solid lines correspond to the number of WSS modules for the matrix switch based SDM node, the three-stage rearrangeable nonblocking SDM node, and the proposed SDM node, respectively. The numbers of WSS modules are shown in the inset to Fig. 7. Note that here we assume up to seven WSS modules can be integrated into a single module. An integer factor, α, is introduced and it increases by one as K increases by seven. We assume that M = 8 and L = 2 as well. When K = 20, the number of WSS modules for the matrix switch based SDM node is 320, that for the 3-stage rearrangeable nonblocking SDM node is 960, and that for the proposed SDM node is 78. As discussed in Section 2.1, the 3-stage rearrangeable nonblocking SDM node suppresses the total number of WSS ports but expands the number of WSS modules. On the other hand, the number of WSS modules for the proposed SDM mode decreases by 76% compared to the matrix switch based SDM node and by 92% compared to the 3-stage rearrangeable nonblocking SDM node. This shows that the proposed SDM node addresses well DA-3. We believe that the proposed SDM node represents a very practical solution.
The number of interconnection fibers also contributes to the cost. This corresponds to half of the total number of WSS ports because an interconnection fiber connects two WSS ports. Therefore, the number of interconnection fibers for the proposed SDM nodes is also fewer than that for other SDM nodes.
We can apply the same transformation that is shown in Fig. 4 to the three-stage SDM node in Fig. 5. We can employ multiple arrayed WSSs to the three-stage SDM node. Although this configuration increases the number of WSS modules compared to the proposed SDM node, it is advantageous in that the output core can be selected. The configuration will become an option for the proposed SDM node if the output core needs to be selected.
2.4. Utilization of resources in SDM nodes
In this section, we discuss the resource utilization of SDM nodes. Simple numerical simulation is carried out to evaluate the available number of established optical paths in the SDM nodes.
We evaluate the number of established connections for the proposed SDM node and a strictly nonblocking SDM node as shown in Figs. 8(b) and 8(a), respectively. We assume a single SDM node that supports K cores and M directions. For simplicity, add/drop ports are not considered here. Figure 8(c) shows a flowchart for this evaluation. Establishment of optical paths is sequentially requested in which the input and output directions are randomly selected. The output core of the least index is assigned to the requested optical path. Optical paths are established until optical path setup fails due to one of two conditions: (1) there is no available core either in the requested input direction or in the requested output direction (“Full usage of direction”), or (2) connection cannot be established in the node even if channels are available both in the requested input direction and in the requested output direction (“Blocking”). We evaluate this 10,000 times, and estimate the number of established optical paths. We evaluate the connections under the condition that K = 20 and M = 8.
Fig. 8 (a) Evaluation model of strictly nonblocking SDM node. (b) Evaluation model of the proposed SDM node. (c) Flow chart for evaluation.
Figure 9 shows a histogram of the number of established optical paths. This figure shows that the distributions of the two nodes are almost the same. The average of the established optical paths is 110.3 for the proposed SDM node, and is 110.9 for the strictly nonblocking node. Therefore, approximately 70% in average of the resources can be assigned in both SDM nodes. These results show that the proposed SDM node can efficiently utilize resources enhanced through SDM, and has almost the same performance as the strictly nonblocking SDM node. This means “blocking” of the proposed SDM node does not severely degrade the utilization efficiency. Therefore, the proposed SDM node efficiently utilizes the resources enhanced through SDM, and has almost the same performance as that of strictly nonblocking SDM node. As mentioned in the previous section, even if blocking occurs in the rearrangeable nonblocking SDM node, blocking can be addressed by rearranging fewer than M-1 optical paths. In this case, at most only seven paths are needed to address the blocking.
Fig. 9 Histogram of number of established optical paths in the proposed SDM node and in the strictly nonblocking node.
3. Experimental results
3.1. Switching characteristics of proposed SDM node
We conducted an experiment showing the functionality of the proposed SDM node. In particular, we confirmed the basic switching function of the SDM node using multiple-arrayed WSSs. In this section, we describe core/direction switching of the SDM node using multiple-arrayed WSSs. Figure 10(a) illustrates the target experimental configuration based on the configuration in Fig. 5(a). As shown in Fig. 5(a), the first stage consists of M sets of K-arrayed 1 x K WSSs and the second stage consists of M sets of K-arrayed K x M WSSs. The third stage has M’ sets of K-arrayed M x 1 WSSs for the output ports, and L sets of K-arrayed M x K WSSs and K-arrayed K x 1 WSSs for the drop ports to select the desired drop optical channel. We assume an 8-degree 7-core SDM node in the experiment, i.e., M = 8 and K = 7. Thus, for each direction, a 7-arrayed 1 x 7 WSS, a 7-arrayed 7 x 8 contention WSS, and a 7-arrayed 8 x 1 WSS are placed in each stage. For the drop ports, a 7-arrayed 8 x 7 contention WSS and a 7-arrayed 7 x 1 WSS are utilized. In the actual experiment, we modified the configuration due to constraints of the experimental equipment as indicated later in this section.
Figure 10(b) shows the experimental configuration. It consists of a transmitter, a 40-km 7-core multicore fiber, a 7-core multicore Erbium-doped fiber amplifier (EDFA), the proposed SDM node, and a receiver. At the transmitter, all the optical signals are generated using the 127-Gbps DP QPSK modulation format. As shown in Fig. 10(b), a 127-Gbps DP QPSK transmitter is employed to generate each of a switched, added, and dropped signal channel whose transmission characteristics are measured at the receiver. Depending on the wavelength of the above-mentioned channel, another 48 channels are simultaneously generated by another 127-Gbps DP QPSK transmitter so that 49 channels are emulated as a whole to verify the switching function of the proposed SDM node. The optical frequency spacing of the 49 channels is 50 GHz.
Due to constraints of the experimental equipment, we modified the configuration of the proposed SDM node as shown in Fig. 10(b). Part of the proposed SDM node is experimentally constructed. In the first stage, we placed a 7-arrayed 1 x 7 WSS (shown in blue) that was newly developed in our laboratories [12,13]. In the second stage, 8 x 8 optical couplers are used in place of the 7-arrayed 7 x 8 contention WSSs. In the third stage, the 7-arrayed 1 x 7 WSSs are replaced with 1 x 9 or 1 x 20 WSSs, and only the necessary ports are utilized. While the optical couplers cannot select the WSSs in the third stage, these WSSs can select output ports so that the same functionalities as those in Fig. 10(a) are achieved. For the add port, we replaced the 7-arrayed 1 x 7 WSS and 7-arrayed 7 x 8 contention WSS with a 1 x 8 optical selector and 8 x 8 optical couplers, respectively. Finally, for the drop port, an 8 x 13 WSS and an 8 x 1 optical selector are used in place of the 7-arrayed 8 x 7 contention WSS and 7-arrayed 7 x 1 WSS to select the required signal, respectively.
By replacing the second-stage WSSs with optical couplers, the filtering effect is relaxed, and the IL and crosstalk increase. However, we conduct the experiment on a single node, so we consider that the effect of filtering and crosstalk is small since the bandwidths of the WSSs are sufficiently wide. The 3-dB bandwidth of the 7-arrayed 1 x 7 WSS is approximately 44 GHz and those of the 1 x 9 WSS, the 1 x 20 WSS, and the 8 x 13 WSS are 47 GHz on average. Detailed analysis of the filtering effect and crosstalk in a multi-SDM node system is outside the scope of this paper.
The IL of the 7-arrayed 1 x 7 WSS is approximately 15 dB and those of the 1 x 9 WSS, the 1 x 20 WSS, and the 8 x 13 WSS are approximately 5 dB on average. Thus, the result of the total IL is approximately 30 dB for the entire SDM node. This value can be improved by adjusting the optical alignment of the 7-arrayed 1 x 7 WSS. We also expect that the node IL could be reduced to approximately 20 dB by replacing the optical couplers with WSSs in the second stage.
We verify the switching function of the proposed SDM node. The optical signal spectra at the input and output of the proposed SDM node are shown in Figs. 11(a) and 11(b), respectively. Here, we emulate two kinds of switching: (S1) switching from one core to another in the same direction and (S2) switching from one direction to another. In addition, we utilized all the cores to demonstrate the capability of the multicore fiber.
Fig. 11 Optical signal spectra. (a) Spectra at input of the proposed SDM node. (b) Spectra at output of the proposed SDM node.
We divide the 49 signal channels into 7 groups, each of which comprises 7 wavelengths in order. Each wavelength channel is assigned to a different core of input direction 1. For example, for the first group (CH 1 to CH 7), CH 1 is assigned to Core 1 of input port A1, CH 2 is assigned to Core 2 of input port A2, and so on. Wavelength channels for the other groups are assigned to the cores of the input port in the same manner as shown in Fig. 11(a).
To confirm S1, we switched 7 wavelength channels of each group, assigned to all the cores of the input port, to an identical core of an output port. For instance, CH 1 to CH 7 of the first group are switched to Core 1 of output port 1, i.e., (B1) in Fig. 11(b). Similarly, from CH 8 to CH 14 of the second group to CH 29 to CH 35 of the fifth group, 7 wavelength channels are switched to the identical core of output port 1, i.e., Core 2 (B2) to Core 5 (B5).
To confirm S2, for the sixth group and the seventh group, 7 wavelengths are switched to Core 1 of output port 2 (B6) and to Core 2 of output port 2 (B7), respectively. Output signals at B1 to B7 are shown in Fig. 11(b). The figure shows that both the core switching function and the direction switching function are successfully confirmed for each wavelength in each space (core).
3.2. Signal transmission characteristics for proposed SDM node
Next, we evaluated the transmission characteristics of typical channels. Bit error rates (BERs) of seven through, one add, and one drop channel were measured at the input and output of the proposed SDM node. Each of the 7 through channels was launched to each core of input port 1 so that all seven cores are utilized. The optical signal power is 6.6 dBm/ch at the input of the SDM node. Since the 7-arrayed 1 x 7 WSS is a laboratory prototype and has some polarization dependent loss (PDL) due to imperfect optical alignment, we select input and output ports with low PDLs as follows. Six of the 7 channels are switched to 3 cores of output port 1 (B3 to B5) and 1 channel is switched to output port 1 (B8). The add channel is directed to output port 1 (B1), and the channel from (A4) is used as the drop channel.
Figure 12 shows Q factors at the input and output of the proposed SDM node. Q factors are calculated using measured BERs of all the channels where the optical signal-to-noise ratio (OSNR) is set to 19 dB to assure a sufficiently high signal-to-noise ratio (SNR). The Q-factor penalty between the input and the output of the proposed SDM node is very low (typically less than 0.3 dB), except that Core 7 has a slightly higher Q-factor penalty of 0.9 dB due to a moderate PDL of 3.0 dB. We believe that the PDL can be suppressed by fine tuning the optical alignment of the 7-arrayed 1 x 7 WSS.
4. Conclusion
We proposed a novel simple SDM node that effectively utilizes enhanced network resources through SDM for the first time. It was shown that the proposed SDM node reduces the number of ports of WSSs and the number of WSS modules by modifying the conventional three-stage rearrangeable nonblocking switch architecture and employing multiple arrayed WSSs. Based on experiments the wavelength, core, and direction switching function of the proposed node was successfully confirmed using 127-Gbps DP-QPSK signals. We also confirmed that 127-Gbps DP-QPSK signals can feasibly be transmitted through the proposed SDM node and the SDM transmission line including a multicore fiber and a multicore amplifier.
Acknowledgments
Part of this research uses results from research commissioned by the National Institute of Information and Communications Technology (NICT) of Japan.
References and links
1. H. Takara, A. Sano, T. Kobayashi, H. Kubota, H. Kawakami, A. Matsuura, Y. Miyamoto, Y. Abe, H. Ono, K. Shikama, Y. Goto, K. Tsujikawa, Y. Sasaki, I. Ishida, K. Takenaga, S. Matsuo, K. Saitoh, M. Koshiba, and T. Morioka, “1.01-Pb/s (12 SDM/222 WDM/456 Gb/s) Crosstalk-managed transmission with 91.4 b/s/Hz aggregate spectral efficiency,” in ECOC 2012 (2012), paper Th.3.C.1.
2. Optical Transport Working Group, “Optical transport use cases,” Open Networking Foundation (2015).
3. D. M. Marom and M. Blau, “Switching solutions for WDM-SDM optical networks,” IEEE Commun. Mag. 53(Feb), 60–68 (2015). [CrossRef]
4. C. Clos, “A study of non-blocking switching networks,” Bell Syst. Tech. J. 32(2), 406–424 (1953). [CrossRef]
5. 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 OFC 2013 (2013), paper PDP5B.8.
6. N. K. Fontaine, T. Haramaty, R. Ryf, H. Chen, L. Miron, L. Pascar, M. Blau, B. Frenkel, L. Wang, Y. Messaddeq, S. LaRochelle, R. J. Essiambre, Y. Jung, Q. Kang, J. K. Sahu, S. U. Alam, D. J. Richardson, and D. M. Marom, “Heterogeneous space-division multiplexing and joint wavelength switching demonstration,” in OFC 2015 (2015), paper Th5C.5.
7. H. C. Le, H. Hasegawa, and K.-I. Sato, “Hierarchical optical path network design algorithm considering waveband add/drop ratio constraint,” J. Opt. Commun. Netw. 2(10), 872 (2010). [CrossRef]
8. V. Kamchevsk, A. K. Medhin, F. Da Ros, F. Ye, R. Asif, A. M. Fagertun, S. Ruepp, M. L. Berger, L. Dittmann, T. Morioka, L. K. Oxenlowe, and M. Galili, “Experimental demonstration of multidimensional switching nodes for all-optical data centre networks,” in ECOC 2015 (2015), paper Tu.1.2.2.
9. Y. Iwai, H. Hasegawa, and K.-I. Sato, “Large-scale photonic node architecture that utilizes interconnected small scale optical cross-connect sub-systems,” in ECOC 2012 (2012), paper We.3.D.3.
10. V. E. Beneš, “On rearrangeable three-stage connecting networks,” Bell Syst. Tech. J. 41(5), 1481–1492 (1962). [CrossRef]
11. E. B. Basch, R. Egorov, S. Gringeri, and S. Elby, “Architectural tradeoffs for reconfigurable dense wavelength-division multiplexing systems,” IEEE J. Sel. Top. Quantum Electron. 12(4), 615–626 (2006). [CrossRef]
12. Y. Ikuma, K. Suzuki, N. Nemoto, E. Tashimoto, O. Moriwaki, and T. Takahashi, “8 × 24 Wavelength selective switch for low-loss transponder aggregator,” in OFC 2015 (2015), paper Th5A.4.
13. K. Suzuki and Y. Ikuma, “Spatial and planar optical circuit,” in OFC 2016 (2016), paper Th3E.1.
14. S. Tibuleac and M. Filer, “Transmission impairments DWDM networks with reconfigurable optical add-drop multiplexers,” J. Lightwave Technol. 28(4), 557–598 (2010). [CrossRef]
