Coexistence of Commercial CV-QKD and DWDM 100G/400G Transmission in Amplified FOADM-based Metro Links

Coexistence of Commercial CV-QKD and DWDM 100G/400G Transmission in Amplified FOADM-based Metro Links Antonio Melgar(1), Masab Iqbal(2), José Manuel Rivas-Moscoso(1), Jeison Tabares(3), Michela Svaluto Moreolo(2), Borja Villanueva(3), Sebastián Etcheverry(3), Pablo Armingol(1), Jesús Folgueira(1) (1)Telefónica CTIO, Rda. Comunicación, 28050 Madrid, Spain, [email protected] (2)CTTC/CERCA, Av. C. F. Gauss, 7, 08860 Castelldefels, Barcelona, Spain. (3)LuxQuanta Technologies S.L., Esteve Terradas 1, 08860 Castelldefels, Barcelona, Spain. Abstract We assess the viability of CV-QKD system integration into amplified FOADM-based metro links deployed conforming to network operators standards, allowing fiber infrastructure sharing. Through experiments and simulations, we determine classical-channel power bounds and quantum-channel frequency allocation enabling QKD and full C-band DWDM channel coexistence. ©2024 The Author(s) Introduction Network operators are faced with the challenge of establishing quantum safe communications over their optical transport networks [1]. Amid several approaches attempting to fulfil this objective, quantum key distribution (QKD) emerges as a strong contender, which, combined with classical cryptographic techniques and post-quantum cryptography (PQC), constitutes a defence-indepth strategy enabling hybrid authenticated key exchange while overcoming the inherent singular vulnerabilities of each technology [2]. Among the various QKD protocols, continuous-variable (CV) QKD [3] is of particular interest for telecom operators due to its robustness against interfering classical channels (CC) within the same transmission band, its seamless integration with current telecom industry infrastructure, and the quantum-channel (QC) frequency tuning flexibility it accords [4]. However, the application of commercially available CV-QKD systems featuring true local oscillators is restricted to links of <50 km, in contrast with research-oriented CV-QKD and commercial discrete-variable (DV) QKD systems, for which distances >100 km have been reported [3, 5-7]. Previous research has examined the coexistence of CV-QKD and classical communications in amplified links [8-10]. These studies evaluated CV-QKD research prototypes with discrete QPSK [8-9] or Gaussian modulation of coherent states (GMCS) [10] and their performance with different QC-CC guard-bands (GB), thus laying the groundwork for understanding the coexistence of CV-QKD systems and traditional communications over the same fiber-optic infrastructure. However, to the best of our knowledge, no research to date has tackled the question of whether commercial CV-QKD systems can be feasibly deployed over existing links conforming to telco production network standards, or provided guidelines for the network design to enable the integration of both technologies. Given the specificities of commercial implementations of CV-QKD systems, in this paper we target metro network links  20 km. In typical dense wavelength division multiplexing (DWDM) systems, irrespective of link lengths, erbium-doped fiber amplifiers (EDFA) are commonly rolled out as part of the switching nodes equipped with optical add/drop multiplexers (OADM) placed on both ends of the link, while variable optical attenuators (VOA) are utilized to adjust the input power to the EDFAs. This holds particularly true for network topologies relying on reconfigurable OADM (ROADM) architectures, but it is also commonplace in fixed OADM (FOADM)-based topologies. In today’s metro networks, 100G and 200G channels are transmitted on the 50-GHz grid, alongside 10G signals in low-capacity links, while 400G signals are beginning to be deployed, requiring 75-100 GHz channel spacing. The power per 50-GHz channel is usually set to ~0 dBm and therefore the EDFA on the transmit side can reach total output powers (Pout_EDFA) of ~20 dBm for 80/96 channels in the C-band, while on the 100-GHz grid (40 channels), Pout_EDFA <17 dBm. These power levels exceed the capabilities of commercial CV-QKD systems and consequently new engineering rules for metro networks need to be put in place to enable coexistence of quantum and classical technologies within the same transmission band. This work presents practical considerations for the deployment of commercial CV-QKD systems in amplified FOADM-based metro networks from a telecom operator’s perspective. Through experiment and simulations, we assess the total power bounds that the DWDM and QKD systems can withstand in coexistence scenarios with up to 39 100G/400G CCs and the optimal placement of the QC in the C-band and the necessary GB, concluding that links of lengths in the range of 10 km can be designed satisfactorily with powers between -8 and -6 dBm/channel. The upper bound can be extended (~18%) by allocating the QC toward the upper frequency end of the C-band. Experimental and Simulation Setups Fig. 1 shows the experimental setup. The CVQKD system is implemented with LuxQuanta’s NOVA LQ® commercial platform, utilizing a true local oscillator and C-band tunable lasers to realize the GMCS protocol. The system features realtime digital signal processing in FPGA, while post-processing software (parameter estimation, error correction) is run in a GPU-powered computer server (Alice or Bob). The system requires transmission of two channels: a QC, supporting up to 8 dB total loss, and a CC intended for key reconciliation, implemented through a TCP/IP link in the setup. The FOADM nodes comprise carrier-grade equipment, including 40-channel 100-GHz MUX/DEMUX with measured insertion loss (IL) between 4.6 and 5.2 dB; and EDFAs with operating wavelength range of 1529 to 1561 nm, noise figure (NF)  6.0 dB, 23 dB gain, and Pout_EDFA (max) of 20 dBm, integrating variable optical attenuators (VOA) used to adjust the input power to the EDFA ( 20 dB). Current FOADMbased systems are designed to support transmission of 40 100G signals (or 80 signals if using an interleaver). In this work, we also allow for 400G transmission as a network evolution and select the 100-GHz grid to accommodate 100G/400G signals. 400G coherent pluggable transceivers in QSFP-DD form factor, enabling IP over DWDM [11], are of great interest to telecom operators. They show great potential for cost and power consumption reduction and CapEx optimization. In the setup, we considered 8 100G channels from OTN cards with fixed optics provided by the same vendor as the optical line system (OLS), 9 100G channels from OTN cards with CFP2 modules (both low and high power) from a different provider, 8 400G channels from CFP2 modules plugged into whiteboxes from two different vendors compliant with the Telecom Infra Project (TIP) Phoenix specifications [12], and 12 400G channels from QSFP-DD modules, with varying output power, from four manufactures, plugged into routers from two vendors, interoperable through the 400G Open ZR+ [13] multi-source agreement (MSA). The power of the CCs was partially equalized to avoid degrading the performance of the high-power channels, with lowpower module outputs kept lower, as observed in Fig. 1. The CCs are multiplexed, amplified and passed through several stages of notch filters, with combined IL of 3.5 dB, to eliminate the amplified spontaneous emission (ASE) noise from channel 193.2 THz. The filtered signal is then combined with the QC (at 193.2 THz) from the QKD Tx module at an OADM (IL of 0.4 dB for CCs; 0.7 dB for QC) and transmitted over a 10km ITU-T G.652.D single-mode fiber (SMF), after which another OADM separates the CCs and the QC. The measured total loss experienced by the QC, due to OADMs and fiber propagation, was 3.4 dB. The QKD Rx detects the QC, and the CCs traverse a VOA, EDFA, and DEMUX, with each channel being detected by the corresponding module through an optical loopback circuit. The simulation setup (implemented with VPIphotonics [14]) replicates the experimental setup. The QKD Tx module generates GMCSs using a laser source operating at 193.2 THz, with 10-kHz linewidth. The laser source carrier is modulated by an amplitude modulator and a phase modulator, and a VOA at the QKD Tx output adjusts the modulation variance (Vmod). To simulate the channel profile in Fig. 1, 37 dummy CCs from 192.1 THz to 193.0 THz, and from 193.4 THz to 196.0 THz are generated and 100-GHz multiplexed. The multiplexed signal is then combined with the QC at an OADM and launched into the 10-km SMF. After propagation, the QC is dropped at the Rx OADM and the QKD Rx module performs phase-diversity detection using a 90o optical hybrid and two pairs of balanced photodetectors. For simulations, the local oscillator is derived from the Tx laser using a beam splitter [15]. A parameter estimation process is conducted on the recovered quantum data to determine transmittance T, excess noise ξ, and asymptotic secret fraction r, considering a reconciliation efficiency β=0.95. T reflects the channel losses of the system (represented by Tch) and considers the detection efficiency ηde (T=Tchηde). Average T=0.1898 is taken to match the experimental setup. For the performance evaluation, we consider the ξ and asymptotic secure key rate (SKR) metrics. The SKR is given by 0.5×R×r, where 0.5 indicates half of the quantum symbols are consumed in parameter estimation and the symbol rate (R) of quantum symbols is assumed to be 31.25 MHz. In the coexistence scenarios, ξ Fig. 1: Experimental setup, and OSA measurement (1%) of 100G/400G CC profile before Tx OADM (Pout_EDFA = 12 dBm). FOADM Node #1 EDFA Mon VOA Notch Filter QKD Tx 1% VOA 1% 17 100 Gb/s 20 400 Gb/s FOADM Node #2 OADM OADM 100GHz MUX 100G 400G Rx Tx OSA #1 OSA #2 Alice Server QKD Rx Bob Server 10 km SMF QC 193.20 THz OSA #3 EDFA Mon 100GHz DEMUX OSA #4 196 195 194193192 -30 -50 -70 dBm THz 400G CFP2 400G QSFP-DD 100G CFP2 400G CFP2 400G QDFP-DD 100G (Fixed) 100G CFP2 OSA #2 Guardband (100 GHz on each side) can be expressed as ξ = ξo + ξSpRS + ξNLI, where ξo represents the intrinsic noise of the CV-QKD system, taken as 0.01 SNU to be aligned to LuxQuanta’s system [4], ξSpRS is due to the spontaneous Raman scattering (SpRS), and ξNLI accounts for the fiber nonlinearity noise, including self-phase modulation (SPM), cross-phase modulation (XPM), and four-wave mixing (FWM). For DWDM compatibility of QKD in the C-band, ξNLI plays a significant role when high-power CCs are placed close to the QC, but SpRS was identified as the main impairment when channel spacing between CCs and QC is sufficiently large [10]. Results and Conclusion The experimental QKD system performance is illustrated in Fig. 2(a). Quantum key-exchange metrics SKR and ξ behave similarly for the coexistence scenario shown in Fig. 1 (37 CCs) and a scenario in which 10G channels are added at 193.1 and 193.3 THz (39 CCs), demonstrating the channel spacing on the 100-GHz grid is sufficient to enable QKD-DWDM transmission, without the requirement of additional GB. This statement is valid when the power is distributed across the C-band. In similar scenarios with power concentrated among a reduced number of channels, [16] estimated through simulations a minimum spacing of 135 GHz for a scenario with 8 200G channels. Fig. 2(a) shows that the maximum power input to the Tx OADM is 9.8 dBm, with SKR of ~0.5 kb/s, for both 37 and 39 CCs, decreasing ~90% vs. a CC power of -1.3 dBm. Fig. 2(b) compares experimental and simulation asymptotic SKR. The maximum attainable power is 11 dBm in simulation and 9.8 dBm in experiments, beyond which asymptotic SKR is negative, not allowing secure key generation. Performance is slightly better in simulation because of the dummy channels vs. modulated channels. Fig. 2(c) explores through simulations [14] the optimal QC frequency allocation, considering a total power of 8.51 dBm after the Tx OADM. Three scenarios are evaluated: QC at 192.2, 193.2, and 195.9 THz, leaving the adjacent 100GHz channels empty. With sufficient GB, the dominant noise source is the ξSpRS and placing the QC at the anti-Stokes region of SpRS yields minimal noise and the highest SKR. This is demonstrated in Fig. 2(c), where QC at 195.9 THz, in the anti-Stokes region, exhibits superior performance (18.26%) than at 192.2 THz (-23.66%), located in the Stokes region when compared with 193.2 THz. Fig. 3 illustrates the performance of a selection of 100G and 400G CCs (to achieve -6 dBm per channel after the Rx EDFA) in terms of preforward error correction (pre-FEC) bit error rate (BER) vs. the channel power. The 100G CCs run across the entire power range considered, but one of the 400G QSFP-DD falls below the preFEC BER for channel powers <-8.3 dBm (7.2 dBm total power before the Tx OADM), and all QSFP-DD are under this threshold for <-10.1 dBm (5.5 dBm total power). 400G CFP2 exhibit a better performance, withstanding power levels down to -10.1 dBm. Thus, 7.2 dBm of total power represents a critical point from the telecom operator's perspective, marking the minimum power for the operation of all CCs corresponding to ~2.4 kb/s SKR. For a 10-km link, the power bounds for both QKD and DWDM systems to be mutually operative spans 7.2 to 9.8 dBm (-6.2 dBm/channel). In summary, in this work we evaluated the copropagation of up to 39 100G/400G CCs with a CV-QKD QC, demonstrating that coexistence is possible through proper system engineering, that all channels on the 100-GHz can be utilized and that the QC must be preferably allocated toward the upper frequency end of the C-band. Fig. 2: (a) SKR and excess noise vs. CC total power for 37 and 39 CCs. (b) Comparison of asymptotic SKR obtained from experiments and simulations vs. CC total power. (c) QC allocation at three frequencies. (a) -15.4% -23.6% (c) Fig. 3: Pre-FEC BER vs. power per CC before the Tx OADM for 3 400G QSFP-DD (C49, C53, C55), 3 400G CFP2 (C37, C38, C39) and 6 100G CFP2 (C41-C46). QSFP-DD Pre-FEC BER threshold CFP2 Pre-FEC BER threshold 400G QSFP-DD (C49, C53 & C55) 400G CFP2 (C39) 100G CFP2 (C43 & C45) 100G CFP2 (C41 & C46) 100G CFP2 (C42) 100G CFP2 (C44) 400G CFP2 (C37 & C38) Acknowledgements Work funded by EU’s Horizon Europe research and innovation program, HE ALLEGRO (101092766) project and EC Digital QUARTER (101091588) project, and by the "Ministerio de Asuntos Económicos y Transformación Digital" and the European Union-NextGenerationEU in the frameworks of the "Plan de Recuperación, Transformación y Resiliencia" and of the "Mecanismo de Recuperación y Resiliencia" under reference 6G-OPENSEC-KEYS (TSI-0630002021-61). References [1] M. Mosca, M. Piani, “2021 Quantum Threat Timeline Report”, Global Risk Institute, 2022. [2] J. Prisco, “Achieving Quantum-Level Security With Hybrid Networks Using PQC, QKD And Quantum Internet”, Forbes, 10 February 2023. [3] Y. Zhang, Y. Bian, Z. Li, S. Yu, and H. Guo, “Continuous-variable quantum key distribution system: Past, present, and future”, Applied Physics Reviews, 11, 011318, 2024, doi: 10.1063/5.0179566. [4] M. Svaluto Moreolo, M. Iqbal, L. Nadal, R. Muñoz, J. Morales, A. Pastor, R. Canto, S. Etcheverry, B. Villanueva, I. Núñez, J. Cunha, “SDN-enabled CV-QKD for Quantum Secure Communication in Open and Disaggregated 6G Networks,” in SPIE Photonics West 2024, San Francisco, USA, 2024. Paper 1289432. [5] A. Hajomer, I. Derkach, N. Jain, H. Chin, U. Andersen, T. Gehring, "Long-distance continuous-variable quantum key distribution over 100-km fiber with local local oscillator," in Sciences Advances, vol. 10, no.1, 2024, doi: 10.1126/sciadv.adi9474. [6] V. Martin, J.P. Brito, L. Ortiz, R.B. Mendez, J.S. Buruaga, R.J. Vicente, A. Sebastián-Lombraña, D. Rincon, F. Perez, C. Sanchez, M. Peev, H. H. Brunner, F. Fung, A. Poppe, F. Fröwis, A.J. Shields, R.I. Woodward, H. Griesser, S. Roehrich, F. De La Iglesia, C. Abellan, M. Hentschel, J.M. Rivas-Moscoso, A. Pastor, J. Folgueira, D. R. Lopez, “MadQCI: a heterogeneous and scalable SDN QKD network deployed in production facilities,” arXiv:2311.12791v2 [quant-ph], doi: 10.48550/arXiv.2311.12791 [7] Y. Pi, H. Wang, Y. Pan, Y. Shao, Y. Li, J. Yang, Y. Zhang, W. Huang, B. Xu, “Sub-Mbps key-rate continuous-variable quantum key distribution with local local oscillator over 100-km fiber,” in Optics Letters, vol. 48, no. 7, pp. 1766-1769, 1 April 2023, doi: 10.1364/OL.485913. [8] F. Karinou, H. H. Brunner, C.-H. F. Fung, L. C. Comandar, S. Bettelli, D. Hillerkuss, M. Kuschnerov, S. Mikroulis, D. Wang, C. Xie, M. Peev, and A. Poppe, "Toward the Integration of CV Quantum Key Distribution in Deployed Optical Networks," in IEEE Photonics Technology Letters, vol. 30, no. 7, pp. 650-653, 1 April1, 2018, doi: 10.1109/LPT.2018.2810334. [9] T. Kawakami, H. Kawahara, T. Okamura, K. Yoshino, and W. Maeda, “No-guard-band integration of digital coherent CV-QKD system into 400 Gbit/s 75 GHz grid DWDM systems”, OFC 2024, paper M3H.3. [10] X. Wang, J. Wang, Y. Chen, Y. Xiao, Z. Li, Z. Chen, Z. Li, D. Wang, "Field Trial of 7 89λ 256 Gb/s CBand Classical / CVQKD Co-Existence Transmission over 7-Core Fiber," 2023 Asia Communications and Photonics Conference/2023 International Photonics and Optoelectronics Meetings (ACP/POEM), Wuhan, China, 2023, pp. 1-4, doi: 10.1109/ACP/POEM59049.2023.10369872. [11] Ó. González de Dios, J. P. Fernández-Palacios Giménez, S. Melin, R. Diaz, J.-F. Bouquier, J. A. Gómez, D. Breuer, S. Hill, E. Lerouzic, K. Coskun, R. Bozaci, A. Mayoral, “IPoWDM convergent SDN architecture - Motivation, technical definition & challenge”, Telecom Infra Project MANTRA Whitepaper, v. 1.0, August 2022. [12] J. A. Gómez, A. Lindgren, J. Hjortas, V. López, A. Gladisch, J. Westphal, S. Moro, D. M. Moreton, V. Cardoso, “Phoenix – Technical Requirements V1.0”, Telecom Infra Project, 2019. [13] B. Zhang (Ed.), Open ZR+ MSA Technical Specification, v3.0, 28 July 2023, https://openzrplus.org/documents/ [14] VPIphotonics. https://www.vpiphotonics.com/index.php [15] M. Iqbal, A. Villegas, L. Nadal, M. Svaluto Moreolo, R. Muñoz, P. Adillon, S. Sarmiento, J. Tabares, S. Etcheverry, “SDN-Enabled Continuous-Variable QKD in Coexistence with 8×200 Gb/s 16-QAM Classical Channels,” in 28th International Conference on Optical Network Design and Modelling (ONDM 2024), Madrid, Spain. [16] M. Svaluto Moreolo, M. Iqbal, A. Villegas, L. Nadal, R. Casellas, and R. Muñoz, “Continuous-Variable Quantum Key Distribution for Enabling Sustainable Secure 6G Networks,” 28th International Conference on Optical Network Design and Modelling (ONDM 2024), Madrid, Spain, 2024.