Open Access
Add to BibSonomyAbstract
We propose a novel elastic optical path network where each specific bitrate signal uses its own dedicated fixed grid and one edge of its frequency grid is anchored at a specific frequency. Numerical evaluations using various bitrate signal patterns and network topologies show that the network proposal can almost match the performance of conventional flexible grid networks, while greatly mitigating the hardware requirements: it allows the use of the tunable filters for the fixed grid systems.
© 2014 Optical Society of America
1. Introduction
The rapid penetration of broadband access including ADSL (Asymmetric Digital Subscriber Line) and FTTH (Fiber-To-The-Home), and of high-speed mobile access is driving the exponential increase in the Internet traffic. In the backbone networks, point-to-point WDM (Wavelength Division Multiplexing) transmission systems and electrical forwarding and routing systems have been widely utilized. However, O-E-O (Optical-Electrical-Optical) conversion is needed at every node of these systems, and this will become a bottleneck that hinders the construction of large scale cost-effective networks given the growth in traffic volume. Recently, wavelength routing networks that utilize ROADMs (Reconfigurable Optical Add/Drop Multiplexers) [1, 2] have been extensively introduced. These ROADMs are equipped mostly with fixed add/drop capabilities. To support the future broadband services such as ultra-high definition TV, which requires at most 72 Gbps for uncompressed real-time transmission [3], and future advanced wavelength services [3, 4], the dynamic operation of wavelength paths is necessary. Enhanced optical layer flexibility is also critical to attain optical layer protection/restoration and to enable future advanced SDN (Software Defined Networks). As a result, ROADMs with so-called C/D or C/D/C (Colorless/Directionless/ Contentionless) add/drop capabilities are required; to this end, various add/drop architectures have been discussed [5–8].
Regarding fiber transmission capacity, maximizing the spectral utilization with minimum cost increment is an important issue. The recently proposed elastic optical path network (flexible grid network) uses a minimum frequency slot granularity (e.g., 12.5 GHz) [9] and can allocate multiple frequency slots as needed [10–13]. The elastic optical path network can be enhanced to maximize the spectral efficiency through the use of a set of sophisticated technologies such as optical coherent OFDM (Orthogonal Frequency Division Multiplexing) [14] or Nyquist WDM [15], an adaptive modulation technique to distance and bitrate. For distance adaptive modulation, the maximum number of bits per symbol is selected for each path, subject to the given transmission characteristic such as OSNR (Optical Signal-to-Noise Ratio) degradation [16, 17], while for bitrate adaptive modulation, the number of frequency slots assigned to each path is tailored to the service bitrate [18]. However, these ultimate technologies require significant technology advances (higher system cost and operation cost), while the available frequency utilization improvement (fiber cost reduction) is, say 30% [19–21], which can be further reduced by the defragmentation program. In particular for metro networks, node cost dominates and the relative link (fiber) cost can be much smaller than that of core network segment. Another practical issue comes from the network evolution scenario. With most large carriers, few kinds of transmission systems generally co-exist so optimization of overall network design and operation is possible [22]. In other words, old generation systems will be generally replaced fiber by fiber. Therefore, in this paper, we discuss flexible grid systems that can be most effectively applied widely, including the metro segment. More specifically, the bitrates considered here are a small set of those that have been standardized in ITU-T so far and future plausible higher bitrates; full coverage of highly granular bitrate signals is not considered.
As described so far, this paper assumes the use of flexible grid ROADMs with C/D or C/D/C function. At the drop side of the ROADM, to accommodate different bitrate signals, a tunable filter function is required [21]; the filter needs to tune both passband center frequency and passband bandwidth with a granularity of 6.25 GHz and 12.5 GHz, respectively. This tunable filter function is possible with coherent detection or with WSSs (Wavelength Selective Switches). Coherent detection is rather expensive and may take substantial time to be extensively deployed including metro area; WSSs are also expensive devices and the port counts of commercially available WSSs are still limited to 20 + and expansion will not be easy. Fortunately, the fixed grid ROADMs with C/D or C/D/C have much relaxed filter requirement compared with the flexible grid ones, and so are more cost-effective. The filter needs to tune its passband center frequency on the fixed grid and the passband bandwidth matches the bitrate of the client system (router) interface card
Recently, a very promising cost-effective tunable filter architecture has been proposed for fixed grid systems and its technical feasibility was verified using PLC (Planar Lightwave Circuit) technologies; a 192 channel tunable filter was fabricated on a single PLC chip (15 × 74 mm2) [23]; size and cost will be greatly reduced if we apply Silicon photonics technologies in the future. If we can match the fiber spectral utilization to that offered by the flexible grid networks, without increasing hardware complexity (e.g. utilizing the fixed grid based hardware), cost effective systems will become possible.
In this paper, we propose a novel elastic optical path network (the “Semi-flexible grid optical path network”), where each specific bitrate signal uses its own dedicated fixed grid and one edges of its frequency slot width is anchored at a specific frequency. Please note that in the flexible grid definition [9], frequency slots are defined with 12.5 GHz slot width granularity and 6.25 GHz central frequency granularity, instead of a grid. Since in the semi-flexible grid network, each bitrate signal is aligned with a fixed grid that is specific to each bitrate, we can utilize current cost-effective fixed grid based hardware. We evaluate the blocking performances of this new architecture in the dynamic path control scenario for different network topologies where various required slot widths patterns for different bitrate signals are tested. Numerical results demonstrate that the proposed network can attain almost the same performance as the conventional flexible grid network, while greatly reducing the complexity of devices, which will achieve cost-effective networks.
A preliminary version of this work was presented at an international conference [24]. In this paper, we detail extended investigations of the performance of the proposed semi-flexible grid optical path network. New material includes: 1) we test the proposed network under various types of connection demand patterns and slot width values allocated to different bitrate signals; 2) blocking characteristics among different bitrate signals are analyzed where blocking ratios and total blocking bandwidth ratio are evaluated; 3) we introduce an improved algorithm for the comparable flexible grid optical path network. The results conclusively confirm the effectiveness of our proposal, the semi-flexible grid network.
2. Proposed elastic optical path networks
Figure 1 shows channel frequency allocation examples for: (a) ITU-T fixed grid, (b) flexible grid, and (c) proposed semi-flexible grid. With the flexible grid [Fig. 1(b)], the channel central frequencies of different bitrate signals can be arbitrarily selected with a minimum granularity of 6.25 GHz provided no channels overlap.
The proposed semi-flexible grid network can attain the same frequency slot granularity as the flexible grid network, but each set of same bitrate channels is located in a fixed frequency slot width that is determined specifically to suit the channel bitrate (and modulation format) and one edge of the frequency slot width is anchored at a specific frequency [Fig. 1(c)]. Please note that this scheme is different from present MLR (Mixed Line Rate) systems where different line rate channels are accommodated on a single fixed grid (for example, 50 or 100 GHz spacing). The bitrate specific frequency slot width (frequency grid) enables us to use cost-effective tunable filters (such as the one shown in Fig. 2) at C/D or C/D/C ROADM drop part, which are the same as those used for fixed grid systems. Figure 3 explains a simplified model for the signal drop part of the fixed and flexible grid ROADM architectures that offer C/D or C/D/C function. A model of our proposed network architecture is shown in Fig. 3(a), where each client system (router) interface-card uses a fixed bitrate receiver, the same as that for fixed grid systems. When tunable lasers are used for the transponder, the tunability follows that of the fixed grid system. In the future, for flexible grid networks, variable bitrate transponders may be utilized that require fully flexible fine tunable lasers and tunable filters [see Fig. 3(b)] in combination with sophisticated router control, however, the interface card will cost a lot more. Our proposed semi-flexible grid networks can significantly reduce the hardware requirements since they can utilize transponders that are basically the same as fixed grid transponders (carrier frequency tunability is required only on the fixed grid); universal transponders are not required. As a result, cost effective semi-flexible networks can be created.
Fig. 3 Example model for (a) proposed semi-flexible grid and (b) conventional flexible grid ROADMs with C/D or C/D/C function.
The proposed semi-flexible network can be regarded as a subset of the flexible grid network, but new advantages are created with our proposed scheme: transponder simplification which includes filter simplification. The important task here is to evaluate the performance of the proposed network in terms of the amount of traffic that can be carried on the same fiber networks.
3. Dynamic optical path control scenario
In this work, we tackle a dynamic network design problem that assumes dynamic optical path setup/release requests; the objective here is to minimize the blocking ratio. In the dynamic path operation scenario, spectral fragmentation can substantially degrade the spectrum utilization efficiency. Hence, we adopt this scenario as a benchmark of the spectral utilization efficiency in assessing the proposed semi-flexible grid network. To this end, a dynamic path control algorithm is introduced. For a given topology and fibers, the algorithm accommodates a new path demand request by first calculating a set of route candidates for the source and destination node pair by using the k-shortest path algorithm. Then it assigns a route and slot pair set to the new path demand in lexicographical order of (“route priority”, “frequency slot index”); according to the route priority, we search for vacant consecutive 12.5 GHz slots and assign them to the selected path. The objective here is to minimize the blocking ratio in the networks. This simple algorithm has been proven to be effective in maximizing the spectrum utilization efficiency of conventional flexible grid networks [25, 17]. The algorithm is summarized as follows:
<Dynamic optical path control algorithm for elastic optical path networks>
Step1. For all connection demands whose holding time have expired, release the paths and free all occupied frequency slots on their routes.
Step2. Select one of the path set-up requests in order of arrival. For the new request, assign the first found pair of route and slot to the request, establish the path, and update the slot usage information on all fibers traversed by the new path. Otherwise block the request. Repeat this procedure until all set-up requests are processed. Go back to Step 1.
4. Numerical experiment
4.1 Conditions
We assume that the minimum frequency slot width is 12.5 GHz in the C-band (4400 GHz wide). All the connection demands are full-duplex, i.e., each demand requires a pair of bidirectional paths. The bitrates requested are set at 40 Gbps, 100 Gbps, 400 Gbps and 1 Tbps; the ratio of expected numbers of their connections is set at 1:1:1:1 (named “R1”), or the ratio corresponding to the inverse of capacity so that the total demand bandwidth for each bitrate signal is equal (named “R2”). We test 4 different slot width requirements (P1-P4, Table 1) for each bitrate. In this work, we do not consider finer bandwidth granularity or particular modulation formats like Nyquist WDM or CO-OFDM. Wavelength conversion is not assumed due to the high cost.
Table 1. Four different slot widths allocated in terms of m; multiple of 12.5 GHz for different bitrate signals
In order to construct a network that suits the traffic distribution considered, the initial stage applies a static flexible grid network design algorithm to determine the number of necessary fibers on each link. A full mesh and random traffic demand are assumed where each demanded bitrate is also randomly selected according the ratio shown in Table 1. The demands between node pairs are then accommodated one by one in descending order of the shortest hop count among node pairs so as to minimize the necessary number of fibers.
For dynamic path setup/release, path operation requests are processed using the algorithm described in the previous section (Section 3) for the semi-flexible grid network. On the other hand, for the flexible grid network, we introduce a slightly modified algorithm that improves the performance by trying to reduce the fragmentation. In the modified algorithm, we first calculate a set of route candidates for each node pair by using the k-shortest path algorithm and then search for the available spectrum segment with minimum contiguous available slot number to make the best use of contiguous available slots. We compare the performances of the improved algorithm and the conventional algorithm for the flexible grid optical path network in Section 4-3.
The dynamic connection demands are generated following a Poisson process and the source/destination nodes are assigned randomly to each dynamic connection demand and the demanded bitrate is distributed as mentioned before. The holding time of each connection follows a negative exponential distribution. The data obtained in the initial period, 10 times the average holding time, is not utilized in the blocking ratio calculation to ensure that the system had reached its steady state. The running time for each evaluation is 100 times the average holding time. For each traffic demand, 25 different traffic distribution patterns are generated and the results are the ensemble average of the obtained results. Physical network topologies tested are N × N (N = 5, 6, 7) polygrid networks, Telecom Italia backbone network [26] shown in Fig. 4(a), and COST266 pan European network [27] shown in Fig. 4(b).
4.2 Simulation criteria
In this work, we adopt 3 simulation criteria to evaluate the performance of the proposed semi-flexible grid optical path network. Here, we use an example shown in Table 2 to explain these criteria:
Table 2. Simulation results
Criterion 1 - Blocking ratio; the ratio of total number of blocked connection requests to total number of connection requests. For example in Table 2, the blocking ratio is (1 + 10) / (400 + 400 + 200 + 100) = 1%.
Criterion 2 - Blocking distribution per bitrate signal; the ratio of the total number of blocked connection requests of a certain bitrate signal to the total number of blocked connections requested. For the example in Table 2, the value for a 1 Tbps signal is 10 / (10 + 1) = 90.9%.
Criterion 3 - Blocking bandwidth ratio; the ratio of total blocked connection request bandwidth over total connection request bandwidth. For the example in Table 2, the blocking bandwidth ratio is (1 × 8 + 10 × 16) / (400 × 4 + 400 × 4 + 200 × 8 + 100 × 16) = 2.63%.
4.3 Results
Figure 5 compares the traffic volume that can be accommodated when the blocking ratio (Criterion 1) is at 1% for the improved algorithm and the conventional algorithm [25, 17] for the flexible grid optical path network. For dynamic services, the target blocking ratio is around 10−2 or less [28], which is commonly assumed value. Here, different network topologies with different patterns (P1-P4, Table 1) and connection demand ratios (R1 and R2) are used. The vertical axis plots the relative accepted traffic volume of “Improved” to “Conventional”, which is calculated by (DemImp – DemConv)/ DemConv, where DemImp and DemConv represents the accepted traffic volume of the proposed improved algorithm and the conventional algorithm for the flexible grid optical path network, respectively. Since our improved algorithm can perform better than the conventional algorithm [25, 17] over different frequency slot number patterns (Table 1) and network topologies, we use the results of the improved algorithm for the flexible grid network as the benchmark in evaluating the effectiveness of the semi-flexible grid network.
Fig. 5 Comparison of accepted traffic volume between different flexible grid optical path network algorithms for different networks with connection demand ratio (a) R1 and (b) R2.
Figure 6 and 7 shows the blocking ratios (Criteria 1) versus traffic volume for the 5x5 polygrid network and COST266 polygrid network, respectively, with slot assignment pattern P1 (Table 1) and the connection demand ratios of R1 and R2. The horizontal axis is determined by the averaged summation of SWsignal in the network, where SWsignal represents the slot width an arriving connection requires. Here, “Semi-flex” and “Flex” represent the results for the proposed semi-flexible grid network and conventional flexible grid network. The results show that our proposed semi-flexible grid network can attain almost the same blocking performance as the conventional flexible grid network over a broad blocking ratio area including the low blocking ratio one. For comparison, we also tested the equivalent fixed grid network. For the fixed grid network, the grid spacing is set at 200 GHz to accommodate the 1 Tbps signal. The results show that both flexible and semi-flexible grid networks can accommodate much larger traffic volume at the same blocking rate, which results in significantly improved spectral utilization efficiency.
Fig. 6 Comparison of blocking ratio for P1 for 5x5 polygrid network with connection demand ratio (a) R1 and (b) R2.
Fig. 7 Comparison of blocking ratio for P1 for COST266 pan European network with connection demand ratio (a) R1 and (b) R2.
Figure 8 compares the performance demonstrated in the experiments that used 200 different traffic distribution patterns and those that used 25; the parameter values used are same as those of Fig. 6(a). The results show that increasing the pattern number yielded virtually the same results; to reduce the computation burden, we used 25 different traffic distribution patterns in all subsequent evaluations.
Fig. 8 Performance comparison for different numbers of traffic distribution patterns for (a) flexible grid and (b) semi-flexible grid networks, tested using the same parameter values as for Fig. 6(a).
Figure 9 compares the traffic volume that can be accommodated when the blocking ratio (Criterion 1) is at 1% for different network topologies with different patterns (P1-P4, Table 1) and connection demand ratios (R1 and R2). The vertical axis plots the relative accepted traffic volume of “Semi-Flex” to “Flex”, which is calculated by (DemSemi-flex - DemFlex)/ DemFlex, where DemSemi-flex and DemFlex represent the accepted traffic volume of the proposed semi-flexible grid network and the conventional flexible grid network. Although there are some differences in the accommodated traffic volume, according to the necessary frequency slot number patterns (Table 1) and network topologies, the value ranges within a narrow area of between −1.6% to + 5.4%. Please note that plus values indicate that the semi-flexible grid network outperforms the conventional flexible grid network. This verifies that the proposed semi-flexible grid network can achieve almost the same blocking ratio as the conventional flexible grid network even with the increased restrictions placed on frequency slot assignment.
Fig. 9 Comparison of accepted traffic volume between Semi-flex and Flex for different networks with connection demand ratio (a) R1 and (b) R2.
Figures 10 and 11 compare blocking distribution for different bitrate signals (Criterion 2) for the semi-flexible grid network and the flexible grid network, under different connection demand ratios. The blocking ratio is 1% for 5x5 polygrid networks and for the COST266 pan European network. The results show that, for both the semi-flexible grid network and flexible grid network, path setup blocking of more than 90% is caused by the 1 Tbps signals, which occupies the broadest bandwidth among all types of bitrate signals. Therefore, to reduce slot fragmentation level in slot assignment, it is important to reserve contiguous available slots as long as possible for anticipation of high bitrate signals. For the semi-flexible grid network, the gap between each assigned bitrate signal allocated in the fiber can be relatively large due to the restrictions on their placement, and the anchored frequency slot assignment can reduce the slot fragmentation caused by iterative path setup/release operation.
Fig. 10 Comparison of blocking distribution per bitrate signal for Semi-flex and Flex. The connection demand ratio is (a) R1 and (b) R2 for 5x5 polygrid network.
Fig. 11 Comparison of blocking distribution per bitrate signal for Semi-flex and Flex. The connection demand ratio is (a) R1 and (b) R2 for the COST266 pan European network.
Next we evaluate the relative blocking bandwidth ratio (Criterion 3) between the semi-flexible grid network and the flexible grid network while the blocking ratio is at 1% for different network topologies with different patterns (P1-P4, Table 1). The vertical axis is given by (BBRSemi-flex - BBRFlex)/ BBRFlex, where BBRSemi-flex and BBRFlex represent the blocking bandwidth ratio values of the proposed semi-flexible grid network and the conventional flexible grid network. The results in Fig. 12 demonstrate that, even with increased restrictions on frequency slot assignment, the proposed semi-flexible grid network offers almost the same blocking bandwidth ratios (- 6.0% ~2.2%) as the conventional flexible grid network.
Fig. 12 Comparison of blocking bandwidth ratio for Semi-flex and Flex for different networks with connection demand ratio (a) R1 and (b) R2
5. Conclusion
We proposed a novel elastic optical path network that utilizes an anchored frequency slot assignment that is defined selectively for each bitrate signal, and evaluated its blocking performance. Numerical experiments proved that the proposed semi-flexible grid network achieves almost the same blocking performance as the conventional flexible grid network for various bitrate signal patterns and different network topologies, while significantly mitigating hardware requirements. The proposed approach yields flexible grid systems with much reduced hardware requirements and will be a viable approach to realizing flexible bandwidth networks cost-effectively.
Acknowledgment
This work was partly supported by NICT λ-reach project and KAKENHI (23246072).
References and links
1. K. Sato, Advances in Transport Network Technologies – Photonic Networks, ATM and SDH (Artech House, 1996).
2. T. S. El-Bawab, Optical Switching (Springer, 2006).
3. K. Sato and H. Hasegawa, “Optical networking technologies that will create future bandwidth-abundant networks,” J. Opt. Commun. Netw. 1(2), A81–A93 (2009). [CrossRef]
4. A. L. Chiu, G. Choudhury, G. Clapp, R. Doverspike, M. Feuer, J. W. Gannett, G. Kim, J. Klincewicz, T. Kwon, G. Li, P. Magill, J. M. Simmons, R. A. Skoog, J. Strand, A. Lehmen, B. J. Wilson, S. L. Woodward, and D. Xu, “Architectures and protocols for capacity efficient, highly dynamic and highly resilient core networks,” J. Opt. Commun. Netw. 4(1), 1–14 (2012). [CrossRef]
5. M. D. Feuer and S. L. Woodward, “Advanced ROADM networks,” in National Fiber Optic Engineers Conference, OSA Technical Digest (CD) (Optical Society of America, 2012), paper NW3F.3. [CrossRef]
6. R. Jensen, A. Lord, and N. Parsons, “Colourless, directionless, contentionless ROADM architecture using low-loss optical matrix switches,” in European Conference and Exhibition on Optical Communication, OSA Technical Digest (CD) (Optical Society of America, 2010), paper Mo.2.D.2. [CrossRef]
7. F. Naruse, Y. Yamada, H. Hasegawa, and K. Sato, “Evaluations of OXC hardware scale and Network Resource Requirements of Different Optical Path Add/Drop Ratio Restriction Schemes,” J. Opt. Commun. Netw. 4(11), B26–B34 (2012). [CrossRef]
8. Y. Iwai, H. Hasegawa, and K. Sato, “A large-scale photonic node architecture that utilizes interconnected OXC subsystems,” Opt. Express 21(1), 478–487 (2013). [CrossRef] [PubMed]
9. I. T. U.-T. Recommendations, “Spectral grids for WDM applications: DWDM frequency grid,” G.694.1 (2012), http://www.itu.int/rec/T-REC-G.694.1/
10. M. Jinno, H. Takara, B. Kozicki, Y. Tsukishima, Y. Sone, and S. Matuoka, “Spectrum-efficient and scalable elastic optical path network: architecture, benefits, and enabling technologies,” IEEE Commun. Mag. 47(11), 66–73 (2009). [CrossRef]
11. O. Gerstel, M. Jinno, A. Lord, and S. J. B. Yoo, “Elastic optical networking: a new dawn for the optical layer?” IEEE Commun. Mag. 50(2), s12–s20 (2012). [CrossRef]
12. 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]
13. A. Klekamp and U. Gebhard, “Benefits for mixed-line-rate (MLR) and elastic networks using flexible frequency grids,” in European Conference and Exhibition on Optical Communication, OSA Technical Digest (CD) (Optical Society of America, 2012), paper Mo.1.D.1. [CrossRef]
14. W. Shieh, H. Bao, and Y. Tang, “Coherent optical OFDM: theory and design,” Opt. Express 16(2), 841–859 (2008). [CrossRef] [PubMed]
15. E. Palkopoulou, G. Bosco, A. Carena, D. Klonidis, P. Poggiolini, and I. Tomkos, “Nyquist-WDM-based flexible optical networks: exploring physical layer design parameters,” J. Lightwave Technol. 31(14), 2332–2339 (2013). [CrossRef]
16. B. Kozicki, H. Takara, Y. Sone, A. Watanabe, and M. Jinno, “Distance-adaptive spectrum allocation in elastic optical path network (SLICE) with bit per symbol adjustment,” in Optical Fiber Communication Conference, OSA Technical Digest (CD) (Optical Society of America, 2010), paper OMU3.pdf. [CrossRef]
17. T. Takagi, H. Hasegawa, K. Sato, Y. Sone, B. Kozicki, A. Hirano, and M. Jinno, “Dynamic routing and frequency slot assignment for elastic optical path networks that adopt distance adaptive modulation,” in Optical Fiber Communication Conference, OSA Technical Digest (CD) (Optical Society of America, 2011), paper OTuI7. [CrossRef]
18. K. Christodoulopoulos, I. Tomkos, and E. Varvarigos, “Spectrally/bitrate flexible optical network planning,” in European Conference and Exhibition on Optical Communication, OSA Technical Digest (CD) (Optical Society of America, 2010), paper We.8.D.3.
19. T. Zami, “What is the benefit of elastic superchannel for WDM network? ” in European Conference and Exhibition on Optical Communication, OSA Technical Digest (CD) (Optical Society of America, 2013), paper We.2.E.1.
20. K. Sato, “Recent developments in and challenges of elastic optical path networking,” in European Conference and Exhibition on Optical Communication, OSA Technical Digest (CD) (Optical Society of America, 2011), paper Mo.2.K.1. [CrossRef]
21. S. L. Woodward and M. D. Feuer, “Benefits and requirements of flexible-grid ROADMs and networks,” J. Opt. Commun. Netw. 5(10), A19–A27 (2013). [CrossRef]
22. P. Magill, presented at the workshop on spectrally/bit-rate flexible optical network design and operation, in Optical Fiber Communication Conference, OSA Technical Digest (CD) (Optical Society of America, 2011).
23. T. Niwa, H. Hasegawa, K. Sato, T. Watanabe, H. Takahashi, and S. Soma, “Compact integrated tunable filter utilizing AWG routing function and small switches,” in Optical Fiber Communication Conference, OSA Technical Digest (CD) (Optical Society of America, 2013), paper OW1C.2. [CrossRef]
24. Z. Shen, H. Hasegawa, K. Sato, T. Tanaka, and A. Hirano, “A novel semi-flexible grid optical path network that utilizes aligned frequency slot arrangement,” in European Conference and Exhibition on Optical Communication, OSA Technical Digest (CD) (Optical Society of America, 2013), paper We.2.E.2.
25. G. Shen and Q. Yang, “From coarse grid to mini-grid to gridless: How much can gridless help contentionless?” in Optical Fiber Communication Conference, OSA Technical Digest (CD) (Optical Society of America, 2011), paper OTuI3. [CrossRef]
26. A. Allasia, V. Brizi, and M. Potenza, “Characteristics and trends of telecom Italia transport networks,” J. Fiber and Integrated Optics. 27(4), 183–193 (2008). [CrossRef]
27. R. Inkret, A. Kuchar, and B. Mikac, Advanced infrastructure for photonic networks – extended final report of COST 266 action (Faculty of Electrical Engineering and Computing, University of Zagreb, 2003), Chap.1.
28. T. Zami, “Illustration of the best synergy between grooming of static traffic and elastic spectral efficiency in the WDM networks,” in European Conference and Exhibition on Optical Communication, OSA Technical Digest (CD) (Optical Society of America, 2012), paper Mo.1.D.3. [CrossRef]
