We investigate the capacity improvement achieved by bandwidth-variable transceivers (BVTs) in fiber links with cascaded reconfigurable optical add and drop multiplexers (ROADMs). It is shown that BVTs with flexibilities in both symbol rate and spectral efficiency enable one to optimize capacity by balancing optical filtering tolerance and signal required signal-to-noise ratio (SNR). Compared with a fixed symbol rate transceiver with standard quadrature amplitude modulations (QAMs), BVTs can increase average capacity by up to 17% in both simulations and experiments.
© 2017 Optical Society of America
Future high capacity elastic optical networks will be enabled by two key components : the first is a flex-grid reconfigurable optical add and drop multiplexer (ROADM) using wavelength selective switches (WSS), which provides reconfigurable optical routing with flexible bandwidth allocation [2,3]; the second is a bandwidth-variable transceiver (BVT) aided by advanced digital signal processing (DSP) techniques, which accommodates varying traffic demands and link conditions by modifying symbol rate and spectral efficiency (SE) [4,5]. In meshed optical networks with varying numbers of cascaded ROADMs for different links, the pass-band bandwidth will vary due to the effects of concatenated filtering caused by the ROADMs [6,7]. As a result, the filtering penalty and the accumulated link noise differ for different links. The conventional system is normally designed to tolerate the worst scenario, leading to an inefficient use of margins for other links. With BVTs, the symbol rate and SE of transmitted signals can be adapted according to certain link conditions. This enables full exploitation of the available capacity for each specific link in a meshed optical network.
In terms of flexible SE, transceivers with reconfigurable standard QAMs at a fixed symbol rate have been commercialized and deployed . Transceivers with a smaller SE granularity are being developed using advanced modulation formats such as time-domain hybrid QAM (TDHQ) [9–11] and multi-dimensional modulation formats . In meshed optical networks with cascaded ROADMs, it has been demonstrated that for fixed data rate (e.g. 100G/200G) systems the achievable transmission distance can be increased by varying the combination of symbol rate and SE using BVTs [13–15]. Besides transmission distance, a more important metric is the increase in network capacity using BVT with a variable symbol rate and/or modulation format. However, the flexibilities of BVTs are realized at extra costs and power consumption [16–18]. Therefore, it is essential to investigate the network values of BVTs with various degrees of flexibility.
In this work, we numerically and experimentally investigate the achievable capacities of BVTs with four different flexibility modes in symbol rate and SE. Note that this work is extended from our previous experimental work presented in  with more detailed analysis and extensive simulations. Particularly, the simulation results demonstrate that compared with a transceiver employing a fixed symbol rate and standard QAMs, the average capacity improvements achieved by BVTs with flexible TDHQ formats, flexible symbol rates or both of them can reach up to 11.3%, 14.9% and 17.0%. These results are in good agreement with the experimental results (reported in  and reviewed here), which are 12%, 14% and 17%, respectively. This paper is organized as follows: in Section 2, we extend the section 2 of  and describe the link scenario and define the four flexibility modes of BVT, which are studied in the following simulations and experiments; In Section 3, the average capacity improvement of the four BVT modes are evaluated in detail by extensive simulations. In Section 4, our experimental work reported in  is reviewed to verify the capacity improvement of BVTs. Finally, we conclude in Section 5.
2. Application of BVTs in fiber links with cascaded ROADMs
In meshed optical networks, the transmitted signal can be filtered and routed by multiple ROADMs before reaching the destination. For a certain pair of Tx and Rx in such networks, the transmitting link can be viewed as an equivalent straight line with cascaded ROADMs at different locations, as illustrated in Fig. 1(a). The ROADM distribution is determined by the network configuration. Since the transfer function of a ROADM is not ideally rectangular, cascading multiple ROADMs induces narrow pass-band filtering and thus reduces the available channel bandwidth. The filtering impact on signal quality depends on signal bandwidth, cascaded ROADM bandwidth and distribution of ROADMs in the link. In this paper, the transmitted signal has a root raised cosine (RRC) pulse shape with a roll-off factor of 0.1, and the signal bandwidth is proportional to symbol rate. For commercial ROADMs built on liquid crystal on silicon (LCOS)-based WSS, the filter transfer function S(f) can be accurately modeled using error functions [20,21]:Figure 1(b) depicts the evolution of the overall 3-dB bandwidth as the number of cascaded ROADMs increases for a legacy 50-GHz grid, where the 3-dB bandwidth of a single ROADM is set to 40 GHz to prevent channel crosstalk. The corresponding 10-dB bandwidth of a single ROADM for BOTF = 8.8 GHz and 14 GHz are 47.6 GHz and 51.8 GHz, respectively. These two values of BOTF as shown in Fig. 1(b) correspond to the value used in simulation and experiment, respectively. Note that the BOTF of 8.8 GHz used in simulation is closer to the specification of current commercial ROADMs . With a variable symbol rate and/or SE, BVTs can work at four different modes as listed in Table 1. Mode-1 corresponds to the current commercial transceiver with a fixed symbol rate and various standard QAMs, which sets the benchmark for the comparison. The reference symbol rate is 32 Gbaud, and the forward error correction (FEC) threshold is assumed to be at BER = 0.02. Mode-2 employs TDHQ to realize a quasi-continuously variable SE at a fixed symbol rate. The TDHQ signals are constructed by time-interleaving M1-QAM symbols with M2-QAM symbols on a symbol by symbol basis [10,22]. In this paper the hybrid signals are formed by two adjacent standard QAM formats. Therefore, any SE or bits per symbol (BpS) corresponds to a unique combination of (M1, M2) and symbol number (N1, N2) in one TDHQ frame. We use BpS per polarization to denote both TDHQs and standard QAMs. To emulate a quasi-continuously variable SE, the BpS resolution is set to 0.125 bit/symbol, meaning that there are seven TDHQ signals between two standard QAMs. To maintain the same BER for the two hybrid formats, the power ratio between the two formats is equal to the difference of the required signal to noise ratios (SNRs) for the target BER. The required SNRs for QPSK, 8QAM, 16QAM, 32QAM and 64QAM signals at BER = 0.02 are 6.3 dB, 10.4 dB, 12.7 dB, 15.7 dB and 18.4 dB, respectively, and the power ratio between each two adjacent formats is accordingly 4.1 dB, 2.3 dB, 3 dB and 2.7 dB, respectively. Figure 1(c) gives the required SNR versus BpS obtained in simulations where only additive white Gaussian noise (AWGN) is loaded. Mode-3 denotes the BVT with a variable symbol rate and standard QAMs. The flexibility in symbol rate provides a trade-off between filtering penalty and delivered capacity. Mode-4 denotes the BVT with both a variable symbol rate and a quasi-continuous SE using TDHQ, which enables to maximize the capacity by selecting an optimal combination of symbol rate and SE considering the filtering effect. The optimal combination is obtained by sweeping both the symbol rate and modulation format and choosing the combination that has the largest achievable capacity. In this study, each flexibility mode is assumed to be well implemented. Additional implementation complexities (and power consumption) will be induced to enable flexible modulation formats (e.g. TDHQ) and flexible symbol rate. However, the details of the implementation are beyond the scope of this paper.
3. Simulation and analysis of capacity improvement achieved by BVTs
In this section, we first describe the simulation setup. Then, we discuss about how the available margin of different link conditions can be converted into capacity with variable symbol rate and BpS, in which the link condition is defined by OSNR and number of cascaded ROADMs. Finally, more simulations were conducted to thoroughly investigate the capacity improvement under various conditions.
The simulation setup is shown in Fig. 2. Laser phase noise was modeled by a Wiener process with a linewidth of 20 kHz and applied at the Tx and Rx, separately . In addition, AWGN was equally loaded at the Tx and Rx to emulate an overall implementation noise of 20 dB SNR. The ROADM filter model was based on Eq. (1) with a BOTF of 8.8 GHz. The 3-dB bandwidth of the ROADM was set to 40 GHz. AWGN was loaded after each ROADM to emulate ASE noise. In the simulations, the link loss including both fiber attenuation and ROADM insertion loss was assumed to be fully compensated by gain-controlled erbium-doped fiber amplifiers (EDFA). The power spectral density of the loaded AWGN at each ROADM was independent of the symbol rate, mimicking the ASE noise generated by EDFAs. The noise variance added after each ROADM was identical, and the total noise accumulated over the link is quantified by optical signal-to-noise ratio (OSNR) with a 0.1 nm resolution. Fiber nonlinearity is not included in the simulations because its impact on our results can be neglected, which is verified in the following experiments. The receiver-side DSP algorithms were the same as those in , where a least mean square (LMS) butterfly filter and a phase lock loop (PLL) were employed to perform format-transparent channel equalization and carrier recovery. The pre-convergence was achieved using training symbols, and afterwards the LMS and PLL were switched to a decision-directed mode. Other format-transparent DSP algorithms could also be employed [24,25]. The symbol rate and BpS resolutions were 0.25 Gbaud and 0.125 bit/symbol, respectively. Chromatic dispersion (CD) was assumed to be completely compensated. Since the laser linewidth was only 20 kHz, the performance degradation induced by equalization-enhanced phase noise can be ignored .
Figure 3(a) shows the BER as a function of symbol rate with 5 cascaded ROADMs and an OSNR of 16.6 dB. The overall 3-dB bandwidth of the cascaded ROADMs is 32.9 GHz. The suitable standard QAM format at 32 Gbaud symbol rate is 8QAM, leading to a total capacity of 192 Gb/s. The same capacity can be obtained with different combinations of symbol rate and BpS, leading to different BER margins. Signals with a high symbol rate and a low BpS suffer from stronger filtering effects, whereas the performance of signals with a low symbol rate and a high BpS is limited by the high required SNR. In this simulation scenario, the capacity of Mode-4 can reach a capacity of 220 Gb/s with a symbol rate of 35.25 Gbaud and a BpS of 3.125 bit/symbol. Figure 3(b) shows another case with 15 cascaded ROADMs, in which the overall 3-dB bandwidth of the cascaded ROADMs is reduced to 29.1 GHz. Mode-1 has the same configuration but the BER margin is reduced. Compared with the 5-ROADM case in Fig. 3(a), the capacity of Mode-4 with 15 ROADMs is reduced to 206 Gb/s, with a symbol rate of 31.75 Gbaud and a BpS of 3.25 bit/symbol. The results suggest that for a certain link condition, the available margin can be converted into capacity by Mode-4 with an optimal combination of symbol rate and BpS.
The comparison result of the four modes depends on OSNR and the number of cascaded ROADMs in the link. First, we investigate the dependence of capacity on OSNR in Fig. 4 by varying the OSNR from 14 dB to 24 dB with 15 cascaded ROADMs. The SE gaps between standard QAMs of Mode-1 lead to discontinuities in the OSNR-Capacity and OSNR-BpS relations, whereas the TDHQ of Mode-2 enables a quasi-continuous tradeoff between OSNR and BpS, as shown in Figs. 4(a) and 4(b), respectively. For example, when the OSNR is reduced from 15.75 dB to 15.5 dB, the BpS of Mode-1 has to reduce from 3 bit/symbol to 2 bit/symbol. In contrast, the BpS of Mode-2 can be set to 2.875 bit/symbol, leading to a 43.7% capacity improvement. The variable symbol rate of Mode-3 can convert extra margin into capacity by balancing the filtering penalty and the required SNR of standard QAMs. For example, for 17 dB and 17.5 dB OSNRs, Mode-3 can be configured to 34.5 Gbaud/8QAM (207 Gb/s) and 27.75 Gbaud/16QAM (222 Gb/s), respectively, where Mode-1 has 32 Gbaud/8QAM (192 Gb/s) for both cases. Mode-4 allows the configuration of an optimal combination of symbol rate and BpS to maximize the capacity. As per Fig. 4(c), the variation of the optimal symbol rate of Mode-4 is smaller than that of Mode-3 attributed to the higher flexibility in BpS. In this investigation, the capacity improvements averaged over all of the OSNRs for Mode-2, −3 and −4 with respect to Mode-1 are 12.3%, 11.8% and 13.9%, respectively. It’s observed from Fig. 4(b) that the BpS of the modulation format in Mode-4 generally increases with a higher OSNR. But the BpS around 4 it is more likely to be configured at 16QAM (BpS = 4) other than the neighboring hybrid signals. This phenomenon is attributed to the non-smoothness in the required SNR-BpS relation around 16QAM signals shown Fig. 1(c).
Next, we show the dependence of capacity on the number of cascaded ROADMs by varying the ROADM number from 1 to 20 with OSNR = 21 dB in Fig. 5. The maximum capacity, and the corresponding BpS and symbol rate are shown Figs. 5(a)-5(c), respectively. For Mode-1, 16QAM (256 Gb/s) is suitable for all the cases. For Mode-2 and Mode-3, as the number of cascaded ROADMs increases, the BpS of Mode-2 can be reduced once the BER exceeds the corresponding threshold, and the symbol rate of Mode-3 can be reduced accordingly to keep the standard QAMs below the BER threshold, as shown in Figs. 5(b) and 5(c). Both of them increase capacity with respect to Mode-1 thanks to the additional flexibilities. Mode-4 achieves the highest capacity by selecting the optimal combination of symbol rate and BpS. For Mode-3 and Mode-4 with variable symbol rate, the capacity decreases rapidly after the first few ROADMs and then decreases more slowly as the number of cascaded ROADMs further increases, which is in good agreement with the evolution of 3-dB bandwidth shown in Fig. 1(a). Overall, the average capacity improvements for Mode-2, −3, and −4 with respect to Mode-1 are 12.5%, 16.7% and 19.0%, respectively.
More simulations were conducted to thoroughly investigate the capacity improvement under various conditions. We varied the OSNR from 14 dB to 24 dB with 1 dB per step, and for each OSNR the number of cascaded ROADMs was swept from 1 to 20. The capacities of the four modes were obtained in each case. The results are summarized in Table 2, where the average capacities are shown for each OSNR. The capacity improvements differ in these cases because the BER margins of Mode-1 are different. For Mode-2, −3 and −4, the average capacity improvements with respect to Mode-1 are 11.3%, 14.9% and 17.0%, respectively. Note that in this simulation we have assumed equal probabilities in both OSNR and the number of cascaded ROADMs. In reality, however, the benefit depends on the specific distributions of link conditions, as well as the bandwidth and filter shape of the ROADMs.
4. Experimental study of capacity improvement achieved by BVTs
In this section, we review the experimental work reported in  to demonstrate the capacity improvement using BVTs in links with cascaded ROADMs. Figure 6 depicts the experimental setup, where a dual-polarization (DP) single-channel coherent system was employed. The pseudorandom binary data was generated and mapped to standard or hybrid QAM symbols in MATLAB. After root-raised-cosine pulse shaping with a roll-off factor of 0.1, the samples were loaded to the transmitter module of a Ciena WaveLogic3 transceiver, which incorporated a low-linewidth external cavity laser (ECL), four high-speed digital-to-analog converters (DAC) and a DP IQ modulator. The maximum symbol rate in our experiment was 35 Gbaud. We mimicked the cascaded ROADM effects by putting a WSS within a 320-km re-circulating loop. The loop consisted of four spans of 80 km standard single mode fibers, each followed by an inline EDFA to compensate the loss. The WSS (Finisar Waveshaper) with a 3-dB bandwidth of 40 GHz was placed after the second inline EDFA. The WSS was measured to have a BOTF of 14 GHz. Before entering the loop, the signal launch power was controlled by an optical attenuator (VOA) following a booster EDFA. At the receiver side, another ECL was adopted as the local oscillator. After optical-to-electrical (OE) conversion by four balanced photodiodes, the waveform was captured by a four-channel real time oscilloscope with a sampling rate of 80 GSa/s. The waveform samples were processed offline in MATLAB. The DSP functions included front-end compensation, resampling to two samples per symbol, CD compensation, frequency offset compensation, matched filtering, cross-correlation based synchronization, decision-directed LMS equalization, PLL and symbol decision .
First, we transmit standard QAMs (Mode-1) and TDHQs (Mode-2) at 32 Gbaud. We focus on the distances up to 4800 km (15 loops). The maximum reaches of these two modes are given in Fig. 7(a), in which the solid line is the achievable number of loops and the dashed line is the interpolated maximum transmission reach. The interpolated maximum reach for a given BpS was obtained by measuring the BERs at different number of loops and then applying a linear interpolation at BER = 0.02. For the interpolated maximum reach versus BpS, a low-pass filter was further applied to remove small fluctuations. TDHQ realizes a continuous tradeoff between the SE and the achievable distances. For example, a maximum capacity improvement of 37% was obtained at 9 loops (2880 km) using TDHQ at 2.75 bit/symbol compared to QPSK at 2 bit/symbol. We then vary the symbol rate of standard QAMs (Mode-3), and the maximum reach is given in Fig. 7(b). The interpolation in Fig. 7(b) is similar with that in Fig. 6(a). Mode-3 can increase the capacity by using the extra margin to tolerate more filtering penalty at a higher symbol rate. For example, 32 Gbaud 8QAM can propagate 8 loops. Then at a distance of 6 loops with some extra margin for 8QAM, the symbol rate can be increased to 35 Gbaud to produce a gain of 9% capacity improvement.
The maximum capacities that Mode-1, −2, and −3 achieve at different numbers of loops can be obtained from Figs. 7(a) and 7(b). The maximum capacity of Mode-4 is obtained by further varying both the BpS and symbol rate accordingly. The results are shown in Fig. 8(a), while Figs. 8(b) and 8(c) show the corresponding BpS and symbol rate, respectively, for each mode. The improvements of Mode-2 and Mode-3 vary with the link length, and in some cases they arrive at the same improvement even with different symbol rates and BpSs. For example, when the transmission distance is increased from 5 loops to 6 loops, the format of Mode-1 has to be switched from 16QAM to 8QAM, which decreases the capacity by 25%. In this case, we can either reduce the BpS to 3.875 bits/symbol while keeping the same symbol rate in Mode-2 (shown in Fig. 8(b)), or lower the symbol rate to 31 Gbaud with 16QAM in Mode-3 to be below the FEC threshold (shown in Fig. 8(c)). Both configurations result in a capacity of 248 Gb/s with only 3% capacity decrease. Mode-4 balances the filtering penalty and the required SNR of signals by varying the symbol rate and SE together, in which case a larger capacity is achieved compared to other modes as demonstrated in Fig. 8. For example, at a distance of 11 loops, compared to Mode-1 (32 Gbaud QPSK), Mode-2 (32 Gbaud 2.375 bit/symbol TDHQ) and Mode-3 (35 Gbaud QPSK) increase the capacity by 18% and 9%, respectively, while Mode-4 can increase the capacity by 25% with 28 Gbaud 2.875 bit/symbol TDHQ. The capacity improvements are summarized in Table 3. Compared with Mode-1, the maximum capacity improvements achieved by Mode-2, −3 and −4 are 37%, 45%, and 47% for a specific distance, respectively. The average capacity improvements are 12%, 14% and 17% across all distances, respectively, and they are in good agreement with the simulation results.
We investigate the capacity improvement using bandwidth-variable transceivers with various flexibility features in the presence of the cascaded ROADMs induced optical filtering. Compared with the current commercial transceiver which employs variable standard QAMs at a fixed symbol rate, the average capacity improvements using BVTs with flexible time domain hybrid QAM formats, flexible symbol rates or both features can reach up to 11.3%, 14.9% and 17.0% in the simulations, and 12%, 14% and 17% in the experiments. These results underline the significance of using BVTs in cascaded ROADM links for optimizing network capacity.
China Scholarship Council (CSC) (201506070047).
The authors would like to thank Thang Minh Hoang, Mohammed Sowailem and Mathieu Chagnon for their support in building the experimental setup.
References and links
1. E. Agrell, M. Karlsson, A. Chraplyvy, D. J. Richardson, P. M. Krummrich, P. Winzer, K. Roberts, J. K. Fischer, S. J. Savory, B. J. Eggleton, M. Secondini, F. R. Kschischang, A. Lord, J. Prat, I. Tomkos, J. E. Bowers, S. Srinivasan, M. Brandt-Pearce, and N. Gisin, “Roadmap of optical communications,” J. Opt. 18(6), 063002 (2016). [CrossRef]
2. O. Gerstel, M. Jinno, A. Lord, and S. B. Yoo, “Elastic optical networking: A new dawn for the optical layer?” IEEE Commun. Mag. 50(2), 12–20 (2012). [CrossRef]
3. D. Klonidis, F. Cugini, O. Gerstel, M. Jinno, V. Lopez, E. Palkopoulou, M. Sekiya, D. Siracusa, G. Thouenon, and C. Betoule, “Spectrally and spatially flexible optical network planning and operations,” IEEE Commun. Mag. 53(2), 69–78 (2015). [CrossRef]
4. N. Sambo, P. Castoldi, A. D’Errico, E. Riccardi, A. Pagano, M. S. Moreolo, J. M. Fabrega, D. Rafique, A. Napoli, S. Frigerio, E. H. Salas, G. Zervas, M. Nolle, J. K. Fischer, A. Lord, and J. P. F.-P. Gimenez, “Next generation sliceable bandwidth variable transponders,” IEEE Commun. Mag. 53(2), 163–171 (2015). [CrossRef]
5. J. K. Fischer, S. Alreesh, R. Elschner, F. Frey, M. Nölle, C. Schmidt-Langhorst, and C. Schubert, “Bandwidth-variable transceivers based on four-dimensional modulation formats,” J. Lightwave Technol. 32(16), 2886–2895 (2014). [CrossRef]
6. Y.-T. Hsueh, A. Stark, C. Liu, T. Detwiler, S. Tibuleac, M. Filer, G.-K. Chang, and S. E. Ralph, “Passband narrowing and crosstalk impairments in ROADM-enabled 100G DWDM networks,” J. Lightwave Technol. 30(24), 3980–3986 (2012). [CrossRef]
7. D. Rafique and A. D. Ellis, “Nonlinear and ROADM induced penalties in 28 Gbaud dynamic optical mesh networks employing electronic signal processing,” Opt. Express 19(18), 16739–16748 (2011). [CrossRef] [PubMed]
8. K. Roberts and C. Laperle, “Flexible transceivers,” in European Conference on Optical Communication (Optical Society of America2012), We3A3. [CrossRef]
9. X. Zhou, L. E. Nelson, and P. Magill, “Rate-adaptable optics for next generation long-haul transport networks,” IEEE Commun. Mag. 51(3), 41–49 (2013). [CrossRef]
10. Q. Zhuge, M. Morsy-Osman, X. Xu, M. Chagnon, M. Qiu, and D. V. Plant, “Spectral efficiency-adaptive optical transmission using time domain hybrid QAM for agile optical networks,” J. Lightwave Technol. 31(15), 2621–2628 (2013). [CrossRef]
11. Q. Zhuge, X. Xu, M. Morsy-Osman, M. Chagnon, M. Qiu, and D. V. Plant, “Time domain hybrid QAM based rate-adaptive optical transmissions using high speed DACs,” in Optical Fiber Communication Conference, OSA Technical Digest (online) (Optical Society of America, 2013), paper OTh4E.6. [CrossRef]
12. D. S. Millar, T. Koike-Akino, S. Ö. Arık, K. Kojima, K. Parsons, T. Yoshida, and T. Sugihara, “High-dimensional modulation for coherent optical communications systems,” Opt. Express 22(7), 8798–8812 (2014). [CrossRef] [PubMed]
13. W. Wang, Q. Zhuge, Y. Gao, X. Xu, and D. Plant, “Performance optimization in ROADM-enabled DWDM systems using flexible modulation formats,” in Optical Fiber Communication Conference, OSA Technical Digest (online) (Optical Society of America, 2015), paper Th2A.27. [CrossRef]
14. W. Idler, F. Buchali, L. Schmalen, K. Schuh, and H. Buelow, “Hybrid modulation formats outperforming 16QAM and 8QAM in transmission distance and filtering with cascaded WSS,” in Optical Fiber Communication Conference, OSA Technical Digest (online) (Optical Society of America, 2015), paper M3G.4.
15. F. Buchali, W. Idler, L. Schmalen, and K. Schuh, “Performance and advantages of 100 Gb/s QPSK/8QAM hybrid modulation formats,” in Optical Fiber Communication Conference (Optical Society of America, 2015), Th2A. 16. [CrossRef]
16. C. Laperle and M. O’Sullivan, “Advances in high-speed DACs, ADCs, and DSP for optical coherent transceivers,” J. Lightwave Technol. 32(4), 629–643 (2014). [CrossRef]
17. X. Chen, S. Chandrasekhar, S. Randel, W. Gu, and P. Winzer, “Experimental quantification of implementation penalties from limited ADC resolution for Nyquist shaped higher-order QAM,” in Optical Fiber Communication Conference, OSA Technical Digest (online) (Optical Society of America, 2016), paper W4A.3. [CrossRef]
18. K. Roberts, S. H. Foo, M. Moyer, M. Hubbard, A. Sinclair, J. Gaudette, and C. Laperle, “High capacity transport-100G and beyond,” J. Lightwave Technol. 33(3), 563–578 (2015). [CrossRef]
19. X. Zhou, Q. Zhuge, M. Qiu, M. Xiang, F. Zhang, and D. V. Plant, “Capacity improvement using bandwidth-variable transceiver in meshed optical networks with cascaded ROADMs, ” in European Conference on Optical Communication (ECOC, 2016), paper W4P1SC4.42.
20. C. Pulikkaseril, L. A. Stewart, M. A. Roelens, G. W. Baxter, S. Poole, and S. Frisken, “Spectral modeling of channel band shapes in wavelength selective switches,” Opt. Express 19(9), 8458–8470 (2011). [CrossRef] [PubMed]
21. D. M. Marom, “Progress report on the design and development of flexible elements for the ‘Drop’ and ‘Add’ function,” http://cordis.europa.eu/docs/projects/cnect/5/318415/080/deliverables/001-FOXCD43Ares20142424820.pdf.
22. V. Curri, A. Carena, P. Poggiolini, R. Cigliutti, F. Forghieri, C. R. Fludger, and T. Kupfer, “Time-division hybrid modulation formats: Tx operation strategies and countermeasures to nonlinear propagation,” in Optical Fiber Communication Conference, OSA Technical Digest (online) (Optical Society of America, 2014), paper Tu3A.2. [CrossRef]
23. M. G. Taylor, “Phase estimation methods for optical coherent detection using digital signal processing,” J. Lightwave Technol. 27(7), 901–914 (2009). [CrossRef]
24. Y. Gao, Q. Zhuge, D. V. Plant, C. Lu, and A. P. T. Lau, “Blind and universal DSP for arbitrary modulation formats and time domain hybrid QAM transmissions,” in Optical Fiber Communication Conference, OSA Technical Digest (online) (Optical Society of America, 2014), paper Th3E.5. [CrossRef]
25. M. Qiu, Q. Zhuge, W. Wang, M. Chagnon, F. Zhang, and D. V. Plant, “Optimized superscalar parallelization-based carrier phase recovery for agile metro optical networks,” J. Lightwave Technol. 34(4), 1111–1119 (2016). [CrossRef]
26. Q. Zhuge, X. Xu, Z. A. El-Sahn, M. E. Mousa-Pasandi, M. Morsy-Osman, M. Chagnon, M. Qiu, and D. V. Plant, “Experimental investigation of the equalization-enhanced phase noise in long haul 56 Gbaud DP-QPSK systems,” Opt. Express 20(13), 13841–13846 (2012). [CrossRef] [PubMed]