Apodized fiber Bragg grating filters for efficient QKD channel separation in QKD-over-DWDM systems Authors: Namwook Joe, Taeho Woo, Suckwoo Shin, Kyungtaek Lee, Ju Han Lee Published: 2026 ABSTRACT This study theoretically investigates the optimal filter configuration for separating a quantum key distribution (QKD) channel from classical C-band communication channels in a discrete-variable (DV) QKD-over-dense wavelength division multiplexing (DWDM) communication system with an O-band QKD channel. Three filter configurations are analyzed and compared in terms of quantum bit error rate (QBER), secret key rate (SKR), and fiber link distance: two cascaded 1310-nm bandpass filters, a 1310-nm uniform fiber Bragg grating (FBG) with a 1310-nm circulator, and a 1310-nm apodized FBG with a 1310-nm circulator. The findings reveal that the optical filter configuration incorporating a Gaussian apodized FBG exhibits the best performance with an isolation of ~ 93 dB between 1310 nm and 1550 nm wavelengths. This optimal performance can be attributed to the Gaussian apodization profile, which effectively suppresses spectral side-lobes to enhance channel isolation while maintaining a lower insertion loss compared to the cascaded bandpass filter configuration. For 50 classical WDM channels, the QKD fiber link distance is 35.4 km with a Gaussian apodized FBG, while a configuration incorporating two cascaded 1310-nm bandpass filters yields only 18.3 km. Finally, the impact of adopting a raised-cosine apodized FBG to further increase filter channel isolation is examined. Despite a substantial increase in channel isolation by using a raised-cosine apodized FBG filter, QKD fiber link distance yields negligible improvement for this system with a QKD channel in O-band, although this filter may be highly useful for QKD channels in C- or L-bands. Introduction Quantum key distribution (QKD) has been extensively investigated over the past decades as it enables the exchange of cryptographic keys with security guaranteed by the uncertainty principle. Since Bennett and Brassard proposed the first QKD protocol in 1984,1 numerous implementations of QKD have been explored across various communication links, including fiber-optic links,2–4 ground-to-air free-space links,5 terrestrial free-space links,6,7 and ground-to-satellite links.8,9. Most fiber-based QKD demonstrations have relied on dedicated dark fibers, hindering their practicality. For scalable QKD deployments and expansion to point-to-point links to networked configurations, it is desirable to integrate QKD with existing fiber infrastructure alongside classical communication traffic.10 A major challenge for such implementation lies in separating QKD signals from classical wavelength division multiplexing (WDM) communication systems, a task that becomes particularly difficult in dense WDM (DWDM) topology.11–15 The state-of-the-art DWDM communication systems operate in the C-band with 100 GHz–50 GHz channel spacing on spectral grids for WDM applications of DWDM frequency grid.16 When QKD and DWDM classical channels coexist in a standard single-mode fiber (SMF28), the large difference between quantum and classical signals induces noise at the quantum receiver. First, the limited channel isolation performance of commercial WDM demultiplexers can cause a significant crosstalk noise from multiple classical DWDM channels.17 Second, Spontaneous Stokes and anti-Stokes Raman scattering from classical channels imposes considerable noise.18 Third, four-wave mixing (FWM) induced by DWDM classical channels further imposes significant noise on the quantum channel when the quantum channel wavelength is allocated near the classical channels.11 While FWM-induced noise is less invasive than that of Raman-induced noise, proper channel allocation is required to mitigate the impact of FWM secondary photons on the QKD channel. Fourth, erbium-doped fiber amplifiers (EDFAs) incorporated in practical DWDM communication systems generate substantial amplified spontaneous emission (ASE), which can be detrimental to QKD performance when QKD and classical channels share the same path. Therefore, a configuration of bypassing EDFA has been devised19; however, it introduces additional insertion loss and operational complexity to such an implementation. The commonly employed QKD-over-DWDM communication systems are implemented through three schemes: (1) discrete-variable (DV) QKD channels in the O-band,20 (2) DV QKD channels in the C-band,21,22 (3) continuous-variable (CV) QKD channels in the C-band.23,24 The O-band allocation of QKD channels simplifies the implementation of such integrated systems by reducing classical channel crosstalk through a ~ 240 nm spectral separation between the classical communication and QKD channels. However, this scheme has a fundamental issue of limited quantum key propagation distance due to the high propagation loss of standard single-mode optical fiber at 1300 nm. In contrast, the C-band allocation of QKD channels enables longer QKD transmission distances than O-band implementations. However, they face noise issues from classical channel crosstalk, ASE of EDFA, spontaneous Raman scattering, and four-wave mixing crosstalk. To mitigate these effects, a time-gated quantum key detection scheme for DV QKD channels needs to be employed,18 while a coherent quantum key detection scheme is required for CV QKD channels.25. Effective implementation of these systems requires high-performance optical filters at the receiver to ensure strong channel isolation, regardless of whether QKD channels are placed in the O- or C-band. Commonly used optical filters or demultiplexers for the separation of QKD channels from classical DWDM communication channels are based on arrayed waveguide gratings (AWGs)26 or liquid-crystal on silicon (LCoS) wavelength-selective switches (WSSs).27 However, the channel isolation of commercially available AWGs or LCoS-WSSs is only 25 ~ 50 dB,26 with a few AWG-based demultiplexers achieving up to ~ 80 dB.17 Recently, optical fiber Bragg gratings (FBGs) have been explored28 as a low-cost alternative to traditional AWG- or WSS-based filters. However, their limited isolation requires cascaded configurations of multiple bandpass filtering devices to sufficiently separate QKD and classical communication channels, which reduces QKD transmission distance and increases filter insertion loss, since the channel isolation values of the filtering devices are not high enough for successful recovery of transmitted quantum keys. It is well-known in the field of fiber-optic communication systems that the apodized FBG filter can provide high channel isolation and high side-lobe suppression, which are suitable for communication channel isolation in DWDM systems.29 However, apodized FBG filters have rarely been employed in practical DWDM communication systems since they do not allow for a multi-wavelength channel demultiplexing function, which is essential for DWDM systems. Note that AWGs or WSSs are commonly used as practical filtering devices for DWDM systems. The motivation of this work is that an apodized FBG filter with high channel isolation could be an excellent filtering device for such systems, in which the separation of a QKD channel does not require a multi-wavelength channel demultiplexing function. To the best of the authors’ knowledge there has been no investigation of using apodized FBG filters for this application. This study theoretically evaluates the potential of using a single filter based on apodized FBGs as an alternative to cascaded configurations of multiple filtering devices for separating QKD and DWDM classical communication channels. This analysis focuses on a DV QKD-over-DWDM communication system with a QKD channel in the O-band, comparing three filter configurations: two cascaded 1310-nm bandpass filters, a 1310-nm uniform FBG with a 1310-nm fiber-optic circulator, and a 1310-nm apodized FBG with a 1310-nm fiber-optic circulator. The system performance of the filter configurations is assessed in terms of quantum bit error rate (QBER), secret key rate (SKR), and the corresponding fiber link distance. Results reveal that a Gaussian apodized FBG with a fiber-optic circulator achieves a channel isolation of ~ 93 dB between 1310 nm and 1550 nm wavelengths, sufficient for reliable QKD channel detection. This design outperforms both configurations of two cascaded 1310-nm bandpass filters and a 1310-nm uniform FBG with a 1310-nm fiber-optic circulator. Finally, this study also investigates the impact of a further increase in FBG filter channel isolation between QKD and DWDM classical channels by adopting a raised-cosine apodization. System configuration Schematic of QKD-over-DWDM Systems This study considers a BB84 QKD protocol integrated into a DWDM communication system with a QKD channel in the O-band (Fig. 1). The QKD channel is assigned a wavelength of 1310 nm, while the classical DWDM communication channels follow the wavelengths satisfying the DWDM 100 GHz ITU Grid, in which the channel separation wavelength interval is 0.8 nm. The classical communication channels are in the C-band (1528.77 ~ 1568.36 nm), including two distillation channels, two data encryption channels, and additional classical communication channels. To better reflect practical WDM telecommunication conditions, the average optical power of each classical WDM channel is set to 0 dBm at the input end of a fiber link, unlike previous demonstrations for such hybrid quantum-classical architectures22 that reduced the average optical power of each C-band WDM channel to below − 20 dBm at the transmitter. In the proposed system configuration (Fig. 1), the average optical power of each C-band WDM channel at the transmitter is 5.7 dBm. Classical DWDM communication channels are multiplexed and demultiplexed at the Alice and Bob sides using a single AWG, with the number of downstream and upstream channels N varying from 2 to 25 to evaluate the performance of our proposed QKD channel filter. The QKD channel is assumed to have only a downstream transmission, with the classical DWDM communication channels and the downstream QKD channel multiplexed and demultiplexed through a 1310/1550-nm wavelength division multiplexer (WDM). On the Bob side, the QKD channel is separated from classical DWDM communication channels using a high-isolation filter after the WDM demultiplexer (DEMUX). The filtered QKD signal is detected with a QKD receiver based on a single photon avalanche photodiode (SAPD). The quantum signal detection efficiency η is 7%, the gate time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}{t}_{gate}$$\end{document} is 1.5 ns, the detector dead time is 10 µs, the dark count probability rate \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{p}_{dc}^{{\prime\:}}$$\end{document} is \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:5\times\:{10}^{-6}\:{\mathrm{n}\mathrm{s}}^{-1}$$\end{document}, and the guaranteed bit error rate (BER) is less than 10⁻¹². The QKD receiver requires a channel isolation exceeding 100 dB to effectively suppress the channel crosstalk caused by classical DWDM communication channels, even though the impact of spontaneous Raman scattering and FWM effects on the QKD channel decreases compared to similar implementations with QKD channels in the C-band. All system parameters in the DV QKD-over-DWDM communication systems with an O-band QKD channel are listed in Table 1. Fig. 1Schematic of a DV QKD-over-DWDM communication systems with a QKD channel in the O-band, used for this investigation. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\lambda\:}_{Q}$$\end{document}=1310 nm, Quantum key distribution channel; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\lambda\:}_{D}$$\end{document}, Downstream at classical communication channels; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\lambda\:}_{U}$$\end{document}, Upstream classical communication channels. The classical communication channels are in the C-band (1528.77 ~ 1568.36 nm), including two distillation channels, two data encryption channels, and additional classical communication channels. Single detector dark count probability: 1.5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\times\:$$\end{document}10–6 ns–1. Schematic of a DV QKD-over-DWDM communication systems with a QKD channel in the O-band, used for this investigation. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\lambda\:}_{Q}$$\end{document}=1310 nm, Quantum key distribution channel; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\lambda\:}_{D}$$\end{document}, Downstream at classical communication channels; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\lambda\:}_{U}$$\end{document}, Upstream classical communication channels. The classical communication channels are in the C-band (1528.77 ~ 1568.36 nm), including two distillation channels, two data encryption channels, and additional classical communication channels. Single detector dark count probability: 1.5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\times\:$$\end{document}10–6 ns–1. Table 1Parameters in the DV QKD-over-DWDM communication systems with a QKD channel in the O-band, used in this investigation.Variable [unit]SymbolUsed valueFiber length [km] L 1-100Signal Wavelength [nm] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\lambda\:$$\end{document} 1550 @ Classical channel1310 @ QKD channelFiber attenuation [dB/km] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\alpha\:$$\end{document} 0.21 @ Classical channel0.35 @ QKD channelQuantum detection efficiency \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\eta\:$$\end{document} 0.07Effective Raman cross- section [nm−1km− 1] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\rho\:\left(\lambda\:\right)$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:3.75\times\:{10}^{-13}\:$$\end{document} [22]AWG insertion loss [dB] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{AWG}$$\end{document} 4.5Loss of internal receiver components [dB] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{B}$$\end{document} 2.65WDM insertion Loss [dB] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{WDM}$$\end{document} 1.2Circulator insertion Loss [dB] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{Cir}$$\end{document} 1.2Circulator isolation [dB] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{i,Cir}$$\end{document} 8 @ Port 1→2, 8 @ Port 2→3Bandpass filter insertion Loss [dB] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{BPS}$$\end{document} 1.2Suppression of Bandpass filter [dB] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{F}$$\end{document} 40Designed wavelength of FBG filter [nm] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\lambda\:}_{D}$$\end{document} 1310Repetition rate [Hz] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{f}_{rep}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:5\times\:{10}^{6}$$\end{document} Detector Dead time [s] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\tau\:}_{dead}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:10\times\:{10}^{-6}$$\end{document} Quantum receiver Bandwidth [nm] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}{\uplambda\:}$$\end{document} 0.8Visibility V 0.997Error correction Efficiency \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\eta\:}_{ec}$$\end{document} 1.2Detector Dark count ratio [ns− 1] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{p}_{dc}{\prime\:}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:5\times\:{10}^{-6}$$\end{document} Detector gate-time [ns] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}{t}_{gate}$$\end{document} 1.5FBG length [mm] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{L}_{g}$$\end{document} 14Refractive index modulation of the FBG \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}n$$\end{document} 0-\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:3\times\:{10}^{-4}$$\end{document}QKD protocol coefficient \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\beta\:$$\end{document} 1 @ BB84Number of upstream Channels \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{N}_{U}$$\end{document} 2–25 (1 Distillation channel,1 encryption channel are fixed)Number of downstream Channels \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{N}_{D}$$\end{document} 2–25 (1 Distillation channel,1 encryption channel are fixed) Table 1Parameters in the DV QKD-over-DWDM communication systems with a QKD channel in the O-band, used in this investigation.Variable [unit]SymbolUsed valueFiber length [km] L 1-100Signal Wavelength [nm] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\lambda\:$$\end{document} 1550 @ Classical channel1310 @ QKD channelFiber attenuation [dB/km] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\alpha\:$$\end{document} 0.21 @ Classical channel0.35 @ QKD channelQuantum detection efficiency \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\eta\:$$\end{document} 0.07Effective Raman cross- section [nm−1km− 1] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\rho\:\left(\lambda\:\right)$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:3.75\times\:{10}^{-13}\:$$\end{document} [22]AWG insertion loss [dB] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{AWG}$$\end{document} 4.5Loss of internal receiver components [dB] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{B}$$\end{document} 2.65WDM insertion Loss [dB] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{WDM}$$\end{document} 1.2Circulator insertion Loss [dB] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{Cir}$$\end{document} 1.2Circulator isolation [dB] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{i,Cir}$$\end{document} 8 @ Port 1→2, 8 @ Port 2→3Bandpass filter insertion Loss [dB] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{BPS}$$\end{document} 1.2Suppression of Bandpass filter [dB] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{F}$$\end{document} 40Designed wavelength of FBG filter [nm] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\lambda\:}_{D}$$\end{document} 1310Repetition rate [Hz] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{f}_{rep}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:5\times\:{10}^{6}$$\end{document} Detector Dead time [s] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\tau\:}_{dead}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:10\times\:{10}^{-6}$$\end{document} Quantum receiver Bandwidth [nm] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}{\uplambda\:}$$\end{document} 0.8Visibility V 0.997Error correction Efficiency \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\eta\:}_{ec}$$\end{document} 1.2Detector Dark count ratio [ns− 1] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{p}_{dc}{\prime\:}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:5\times\:{10}^{-6}$$\end{document} Detector gate-time [ns] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}{t}_{gate}$$\end{document} 1.5FBG length [mm] \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{L}_{g}$$\end{document} 14Refractive index modulation of the FBG \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}n$$\end{document} 0-\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:3\times\:{10}^{-4}$$\end{document}QKD protocol coefficient \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\beta\:$$\end{document} 1 @ BB84Number of upstream Channels \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{N}_{U}$$\end{document} 2–25 (1 Distillation channel,1 encryption channel are fixed)Number of downstream Channels \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{N}_{D}$$\end{document} 2–25 (1 Distillation channel,1 encryption channel are fixed) Parameters in the DV QKD-over-DWDM communication systems with a QKD channel in the O-band, used in this investigation. 1550 @ Classical channel 1310 @ QKD channel 0.21 @ Classical channel 0.35 @ QKD channel 2–25 (1 Distillation channel, 1 encryption channel are fixed) 2–25 (1 Distillation channel, 1 encryption channel are fixed) Channel crosstalk For the discussed architecture with a QKD channel in the O-band (Fig. 1), the channel crosstalk noise detection probability (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{p}_{CXT}$$\end{document}) at the QKD receiver is expressed as follows221\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{p}_{CXT}=\eta\:\cdot\:{n}_{d}\left[n{s}^{-1}\right]\cdot\:{10}^{-{t}_{i}/10}\cdot\:\varDelta\:{t}_{gate}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\eta\:$$\end{document} denotes the detection efficiency, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{i}$$\end{document} represents the channel isolation, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}{t}_{gate}$$\end{document} signifies the gate time, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{n}_{d}\left[n{s}^{-1}\right]$$\end{document} corresponds to the minimum number of photons per nanosecond required to guarantee a BER < 10⁻¹², and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{n}_{d}\left[n{s}^{-1}\right]$$\end{document} can be calculated using the receiver sensitivity and the photon energy \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{E}_{photon}=\frac{hc}{\lambda\:}$$\end{document}, as follows222\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{n}_{d}\left[n{s}^{-1}\right]=\frac{{10}^{{R}_{x}\left[dBm\right]/10}}{{E}_{photon}\left[J\right]}\cdot\:{10}^{-12}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{R}_{x}\left[dBm\right]$$\end{document} denotes the power of the classical channel accepted by the QKD receiver, which takes into account the propagation loss at each fiber-optic component. Since the QKD channel is located in the O-band, the FWM-induced crosstalk to QKD channel is neglected in this investigation. Raman scattering effect The effect of spontaneous Raman scattering is considered using the effective Raman cross-section, ρ(λ), which incorporates both backward spontaneous Raman scattering power from the fiber input and forward Raman scattering power from the fiber output. Experiments report that C-band WDM channel-induced spontaneous Raman scattering noise at 1310 nm is 4000 times smaller than at 1550 nm22. Therefore, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\rho\:\left(\lambda\:\right)=3.75\times\:{10}^{-13}$$\end{document} [nm−1km− 1] is approximated accordingly. The forward spontaneous Raman scattering power \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{P}_{Ram,f}$$\end{document} is determined as follows223\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{P}_{Ram,f}={N}_{U}\cdot\:{P}_{out}\cdot\:L\cdot\:\rho\:\left(\lambda\:\right)\cdot\:{\Delta\:}\lambda\:$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{N}_{U}$$\end{document} denotes the number of classical channels at upstream, L represents the fiber length, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}\lambda\:$$\end{document} signifies the quantum receiver bandwidth, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{P}_{out}$$\end{document} refers to the power of each classical WDM channel at the detector after propagation over a fiber link and is expressed simply as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{P}_{in}\cdot\:\mathrm{e}\mathrm{x}\mathrm{p}(-\alpha\:L)$$\end{document}. As aforementioned, the power of each classical WDM channel is set at 0 dBm at the input end of the fiber link after WDM MUX, yielding an average optical power of each classical WDM channel at the transmitter to be 5.7 dBm, considering the insertion losses of the AWG MUX and 1310/1550-nm WDM. The backward spontaneous Raman scattering power affected by the downstream WDM channel is given by224\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{P}_{Ram,b}={N}_{D}\cdot\:{P}_{out}\cdot\:\frac{\mathrm{s}\mathrm{i}\mathrm{n}\mathrm{h}\left(\alpha\:L\right)}{\alpha\:}\cdot\:\rho\:\left(\lambda\:\right)\cdot\:{\Delta\:}\lambda\:$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{N}_{D}$$\end{document} denotes the number of downstream WDM channels. The forward (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{p}_{Ram,f}$$\end{document}) and backward (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{p}_{Ram,b}$$\end{document}) Raman scattering detection probabilities can be obtained using the aforementioned Raman scattering power as follows5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{p}_{Ram,f}=\frac{{P}_{Ram,f}}{{E}_{photon}}\cdot\:\eta\:\cdot\:{\Delta\:}{t}_{gate}$$\end{document}6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{p}_{Ram,b}=\frac{{P}_{Ram,b}}{{E}_{photon}}\cdot\:\eta\:\cdot\:{\Delta\:}{t}_{gate}$$\end{document} The total Raman scattering detection probability is defined by the sum of the forward and backward components as follows:7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{p}_{Ram}{=p}_{Ram,f}+{p}_{Ram,b}$$\end{document} Optical filter configurations for QKD channel Effective QKD channel separation in such an integrated system requires a high-isolation bandpass filter. Since commercially available optical thin-film bandpass filters exhibit a channel isolation of approximately 30 ~ 40 dB, even though a 1310-nm QKD channel is located ~ 240 nm away from the C-band WDM channels, using a single, commercial bandpass filter cannot provide a channel isolation sufficient to effectively suppress the classical WDM channel-induced crosstalks. To address this limitation, we propose the use of an apodized FBG. The fabrication of FBGs is a well-established technique in optical communications. In particular, Gaussian apodized FBGs are currently available as commercial products from various manufacturers. This study evaluates three optical filter configurations for QKD channel separation: two cascaded bandpass filters (Fig. 2(a)), a uniform 1310-nm FBG with a 1310-nm centered fiber-optic circulator (Fig. 2(b)), and the proposed apodized 1310-nm FBG with a 1310-nm centered fiber-optic circulator (Fig. 2(c)). The insertion loss of a bandpass filter is assumed to be 1.2 dB at 1310 nm, considering the commercially available bandpass filter parameters. The insertion losses of a fiber-optic circulator are set at 1.2 dB. The channel isolation between the 1310-nm and 1550-nm wavelengths is set to be 8 dB between Ports 1 and 2 or between Ports 2 and 3. The insertion loss and channel isolation values of the fiber-optic circulator are confirmed by experimentally measuring the parameters of a commercially available 1310-nm fiber-optic circulator. Fig. 2Three different optical filter configurations for QKD channel separation include (a) a conventional two cascaded bandpass filters; (b) uniform 1310-nm FBG with a 1310-nm centered fiber-optic circulator; (c) an apodized 1310-nm FBG with a 1310-nm centered fiber-optic circulator. Three different optical filter configurations for QKD channel separation include (a) a conventional two cascaded bandpass filters; (b) uniform 1310-nm FBG with a 1310-nm centered fiber-optic circulator; (c) an apodized 1310-nm FBG with a 1310-nm centered fiber-optic circulator. FBG modeling and apodization Unlike a uniform grating, an apodized FBG minimizes the refractive index modulation amplitude at the start of the grating and gradually increases toward the center, reaching a maximum at the grating midpoint, before gradually decreasing again toward the end of the grating. This structural design effectively suppresses the sidelobes characterized by a Sinc function pattern characteristic of uniform FBGs. As the index profile changes inside the optical fiber, the way light propagates was modeled through the following process. The schematic of an apodized FBG is illustrated in Fig. 3. The effective refractive index profile (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{n}_{eff}$$\end{document}) of an apodized FBG along the length (z) is given by:30 Fig. 3Schematic of an apodized FBG. 8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{n}_{eff}\left(z\right)={n}_{0}+{\Delta\:}n\:g\left(z\right)\mathrm{c}\mathrm{o}\mathrm{s}\left(\frac{2\pi\:}{{\Lambda\:}}z+\phi\:\left(z\right)\right)$$\end{document} Schematic of an apodized FBG. where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{n}_{0}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}n$$\end{document} represent the refractive index of the optical fiber core and the refractive index modulation of the FBG; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:g\left(z\right)$$\end{document} denotes a normalized apodization function for the FBG; for uniform FBGs, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:g\left(z\right)$$\end{document} = 1; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Lambda\:}$$\end{document} represents the period of the gratings. For this investigation, we considered two different apodization functions (Fig. 4) as follows31. Fig. 4Profile of two apodization functions: Gaussian and raised-cosine. Profile of two apodization functions: Gaussian and raised-cosine. Gaussian Apodization Function9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:g\left(z\right)={\mathrm{e}\mathrm{x}\mathrm{p}[-{\left(\frac{z-{L}_{g}/2}{{L}_{g}/2}\right)}^{2}\:]}^{\:\:\:}$$\end{document} raised-cosine Apodization Function10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:g\left(z\right)=\frac{1}{2}\cdot\:\left[1+cos(\pi\:+\frac{2\pi\:z}{{L}_{g}})\right]$$\end{document} For a refractive index profile with the aforementioned apodization profiles, the electric field amplitudes \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:u\left(z\right)$$\end{document} (forward) and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:v\left(z\right)$$\end{document} (backward) propagating waves satisfy the following coupled-mode theory29,3211\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{du}{dz}=i\stackrel{\sim}{\sigma\:}\left(z\right)u\left(z\right)+i\kappa\:\left(z\right)v\left(z\right)$$\end{document}12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{dv}{dz}=-i\stackrel{\sim}{\sigma\:}\left(z\right)v\left(z\right)-i\kappa\:\left(z\right)u\left(z\right)$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\kappa\:\left(z\right)$$\end{document} signifies the AC coupling coefficient as follows13\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\kappa\:\left(z\right)=\frac{\pi\:\gamma\:}{\lambda\:}\delta\:{\Delta\:}n\cdot\:g\left(z\right)$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\stackrel{\sim}{\sigma\:}\left(z\right)$$\end{document} denotes the generalized DC self-coupling coefficient.14\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\stackrel{\sim}{\sigma\:}\left(z\right)=2\pi\:{n}_{eff}\left(\frac{1}{\lambda\:}-\frac{1}{{\lambda\:}_{B}}\right)+{\sigma\:}_{dc}\left(z\right)$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\lambda}_{\mathrm{B}}$$\end{document} is the designed wavelength (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\lambda\:}_{B}=2{n}_{eff}{\Lambda\:}$$\end{document}); \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\sigma\:}_{dc}$$\end{document} is the DC coupling coefficient; for apodization, it is set to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\sigma\:}_{dc}$$\end{document} = 0 to eliminate self-chirping by keeping the average refractive index constant along the grating length. In the numerical model, the chromatic dispersion of the background fiber was neglected. This approximation is valid because the phase error induced by material dispersion over the short grating length is negligible compared to the structural dispersion dominated by the grating coupling. With all necessary information specified, the reflectivity of the designed FBG can be obtained by solving the transfer matrix \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\mathrm{F}}_{\mathrm{i}}$$\end{document}. If the entire grating length \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\mathrm{L}}_{\mathrm{g}}$$\end{document} is divided into M uniform sections, and the coupling coefficient of each section \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{i}$$\end{document} is evaluated to solve the transfer matrix.3315\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{F}_{i}=\left[\begin{array}{cc}{cos}h\left({\gamma\:}_{B}\varDelta\:z\right)-j\frac{{\stackrel{\sim}{\sigma\:}}_{i}}{{\gamma\:}_{B}}{sin}h\left({\gamma\:}_{B}\varDelta\:z\right)&\:-j\frac{{\kappa\:}_{i}}{{\gamma\:}_{B}}{sin}h\left({\gamma\:}_{B}\varDelta\:z\right)\\\:-j\frac{{\kappa\:}_{i}}{{\gamma\:}_{B}}{sin}h\left({\gamma\:}_{B}\varDelta\:z\right)&\:{cos}h\left({\gamma\:}_{B}\varDelta\:z\right)+j\frac{{\stackrel{\sim}{\sigma\:}}_{i}}{{\gamma\:}_{B}}{sin}h\left({\gamma\:}_{B}\varDelta\:z\right)\end{array}\right]$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\gamma\:}_{B}$$\end{document} is the coupling parameter defined as16\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\gamma\:}_{B}=\sqrt{{\kappa\:}_{i}^{2}-{\stackrel{\sim}{\sigma\:}}_{i}^{2}}$$\end{document} Here, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\gamma\:}_{B}$$\end{document} characterizes the rate of amplitude change of the propagating modes within the grating structure derived from the coupled-mode equations. If \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\gamma\:}_{B}$$\end{document} is real (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\kappa\:}_{i}^{2}-{\stackrel{\sim}{\sigma\:}}_{i}^{2}>0$$\end{document}), it represents the decay rate of the evanescent field in the bandgap; if imaginary (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\kappa\:}_{i}^{2}-{\stackrel{\sim}{\sigma\:}}_{i}^{2}<0$$\end{document}), it describes the phase variation of the propagating mode.32 The overall transfer matrix \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:\mathrm{F}$$\end{document} of the FBG is then obtained as the following product of these sectional matrices17\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:F={F}_{1}\cdot\:{F}_{2}\cdot\:\dots\:\cdot\:{F}_{M}$$\end{document} The reflectivity r can be obtained by calculating a reflection coefficient with a simple relation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\rho\:}_{r}={F}_{21}/{F}_{11}$$\end{document} as18\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:r={\left|{\rho\:}_{r}\right|}^{2}$$\end{document} For theoretical performance analysis, we used our made MATLAB code rather than a commercial simulation tool. The FBG length, peak reflectivity, center wavelength, and 3-dB bandwidth are set at 14 mm, 99.9%, 1310 nm, and ~ 0.56 nm, respectively. The FBG length of 14 mm was selected based on practical fabrication constraints provided by commercial manufacturers. To ensure a saturated reflectivity of over 99.9%, a grating length of at least 10 mm is typically required. Considering that a typical phase mask has a length of 15 mm with a maximum effective grating length of approximately 14 mm, we adopted 14 mm as the upper limit for the design. The refractive index modulation \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}n$$\end{document} of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:3\times\:{10}^{-4}$$\end{document} is sufficient to achieve high reflectivity and isolation, and is readily attainable in high-germanium fibers, boron-germanium-codoped fibers, or two mentioned fibers hydrogenated. Regarding sensitivity, our analysis indicates that for a design target of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}n=3\times\:\:{10}^{-4}$$\end{document}, a fabrication tolerance of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\pm\:\:5\%$$\end{document} results in a maximum channel isolation deviation of only 0.5 dB. However, if the error increases to 10% (specifically a reduction in modulation), the reflectivity may decrease for the fixed 14 mm grating length, which could potentially lead to increased signal insertion loss. However, FBGs designed for high refractive index modulation—typically using high-germanium doped fibers—exhibit higher thermal sensitivity compared to those fabricated with standard SMF. Assuming a thermal sensitivity of 9.7 pm/°C for a standard SMF-based FBG,34 a temperature rise of 50 °C induces a center wavelength shift of approximately 0.48 nm. In contrast, high germanium-doped fibers experience a larger wavelength shift under the same thermal conditions due to their increased thermo-optic coefficient. Consequently, considering the limited FBG bandwidth, the use of high-germanium fibers implies that signal misalignment could potentially occur at temperature variations approaching 50 °C, suggesting that thermal stability is a factor to be considered for optimal operation. In our proposed system, the spectral separation between the target QKD channel (1310 nm) and the classical channel (1550 nm) is approximately 240 nm. Due to this large channel spacing, variations in channel isolation caused by strain induced wavelength shifts are negligible. When evaluating the impact of fabrication tolerance on system performance, decreasing the FBG length from 14 mm to 10 mm (at a fixed \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}n$$\end{document} of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:3\times\:{10}^{-4}$$\end{document}) maintained the reflectivity above 99.9% and resulted in a negligible isolation drop of only ~ 0.2 dB, leaving the QBER and SKR unaffected. However, variations in the refractive index modulation (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}n$$\end{document}) exhibited a more distinct impact. Increasing \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}n$$\end{document} from \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:3\times\:{10}^{-4}$$\end{document} to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:5\times\:{10}^{-4}$$\end{document} (at a fixed length of 14 mm) in the Gaussian and uniform profiles led to an approximate 4 dB reduction in channel isolation due to increased sidelobes. Consequently, while the raised-cosine apodization remained robust with no change in the maximum transmission distance, the Gaussian and uniform profiles experienced a reach penalty of approximately 3 km. When the external environment requires such high channel isolation safety, the use of the raised-cosine apodization filter may be considered. Figure 4 presents our calculated reflectivity spectra of three different FBGs: a uniform FBG, an FBG with a Gaussian apodization function, and an FBG with a raised-cosine apodization function. Figures 5(a) and 5(b) depict the spectra from 1306 to 1320 nm and from 1306 to 1560 nm, respectively. The calculated spectra indicate that the application of apodization functions to FBGs significantly enhances the QKD channel isolation between 1306 and 1550 nm, with isolation values listed in Table 2. In general, apodization reduces the effective grating length and effective coupling coefficient, which can narrow the passband and reduce peak reflectivity, leading to higher insertion loss for the QKD signal. However, our design utilized a sufficiently large refractive index modulation (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\Delta\:}n\ge\:3\times\:{10}^{-4}$$\end{document}) and grating length (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{L}_{g}$$\end{document} = 14 mm) to operate in the saturated reflectivity regime. As shown in Fig. 5, our simulation confirmed that even with Gaussian or raised-cosine apodization, the peak reflectivity remained above 99.9%. Consequently, the excessive insertion loss caused by apodization is negligible (< 0.04 dB). Moreover, although the passband width narrows slightly due to apodization, it remains sufficient (~0.56 nm) to accommodate the QKD signal linewidth. Fig. 5Calculated reflectivity spectra of three FBGs: uniform FBG, Gaussian apodized FBG, and raised-cosine apodized FBG. Spectra over a (a) wavelength range of 1306 ~ 1320 nm and (b) a wide spectral bandwidth of 1306 ~ 1560 nm. Calculated reflectivity spectra of three FBGs: uniform FBG, Gaussian apodized FBG, and raised-cosine apodized FBG. Spectra over a (a) wavelength range of 1306 ~ 1320 nm and (b) a wide spectral bandwidth of 1306 ~ 1560 nm. Table 2Channel isolation performance comparison for various FBG designs.Apodization typeChannel Isolation between 1310 and 1550 nm[dB]Uniform FBG58FBG with a Gaussian apodization function93FBG with a raised-cosine apodization function210 Table 2Channel isolation performance comparison for various FBG designs.Apodization typeChannel Isolation between 1310 and 1550 nm[dB]Uniform FBG58FBG with a Gaussian apodization function93FBG with a raised-cosine apodization function210 Channel isolation performance comparison for various FBG designs. System performance comparison Finally, the system performance of the designed FBGs is computed using the QBER, and the SKR is calculated using the BB84 QKD protocol. The QBER is defined as follows, where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\beta\:$$\end{document} represents the protocol coefficient (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\beta\:=1$$\end{document} for the BB84 protocol):2219\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:QBE{R}_{BB84}=\frac{1}{2}\frac{{p}_{\mu\:}\left(1-V\right)+{N}_{d}{p}_{dc}+{p}_{AP}+{p}_{Ram}+{p}_{CXT}}{\beta\:{p}_{\mu\:}+{N}_{d}{p}_{dc}+{p}_{AP}+{p}_{Ram}+{p}_{CXT}}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\mathrm{p}}_{\mathrm{μ}}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\mathrm{p}}_{\mathrm{dc}}$$\end{document} denote the signal detection probability and dark count probability, respectively; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\mathrm{p}}_{\mathrm{AP}}$$\end{document} represents the after-pulse probability; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\mathrm{p}}_{\mathrm{Ram}}$$\end{document} signifies the Raman noise detection probability; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\mathrm{p}}_{\mathrm{CXT}}$$\end{document} refers to the crosstalk photon detection probability. The signal detection probability \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\mathrm{p}}_{\mathrm{μ}}$$\end{document} is given by3520\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{p}_{\mu\:}={\mu\:\cdot\:t}_{F}\cdot\:{t}_{IL}\cdot\:\eta\:\cdot\:{t}_{B}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\mathrm{t}}_{\mathrm{B}}$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\mathrm{t}}_{\mathrm{F}}$$\end{document}, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\mathrm{t}}_{\mathrm{IL}}$$\end{document} denote the internal receiver component loss, fiber transmission loss, and WDM insertion loss, respectively. The dark count detection probability \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\mathrm{p}}_{\mathrm{dc}}$$\end{document} and after-pulse detection probability \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\mathrm{p}}_{\mathrm{AP}}$$\end{document} can be approximated as2221\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{p}_{dc}={p}_{dc}^{{\prime\:}}\cdot\:{\Delta\:}{t}_{gate}$$\end{document}22\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{p}_{AP}=0.008\left({p}_{\mu\:}+{N}_{d}{p}_{dc}+{p}_{AP}+{p}_{Ram}+{p}_{CXT}\right)$$\end{document} The secret key rate (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{R}_{sec}$$\end{document}) can be obtained by starting with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{R}_{sift}$$\end{document} through the following process.23\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{R}_{sift}=\frac{1}{2}\left(\beta\:{p}_{\mu\:}+{N}_{d}{p}_{dc}+{p}_{AP}+{p}_{Ram}\right){f}_{rep}{\eta\:}_{duty}{\eta\:}_{dead}$$\end{document}24\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{R}_{sec}={R}_{sift}({I}_{AB}-{I}_{AE})$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\:R}_{sift}$$\end{document}denote the post-sifting detection rates; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\mathrm{I}}_{\mathrm{AB}}\text{}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\it \:{\mathrm{I}}_{\mathrm{AE}}$$\end{document}represent the mutual information between Alice and Bob and between Alice and potential eavesdroppers; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\eta\:}_{duty}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\eta\:}_{dead}$$\end{document} represent duty cycle efficiency and detector dead time efficiency, respectively. 22,34 However, in this study, duty cycle efficiency and detector dead time efficiency were not considered to compare only the performance of each filter structure.25\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{I}_{AB}=1-{\eta\:}_{ec}H\left(QBER\right)$$\end{document} where the binary entropy function \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:H\left(p\right)$$\end{document} is given by26\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:H\left(p\right)\equiv\:\:-plo{g}_{2}p-\left(1-p\right){\mathrm{log}}_{2}\left(1-p\right)$$\end{document}27\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{I}_{AE}=\frac{\left(1-\frac{\mu\:}{2t}\right)\left(1-H\left(P\right)\right)+\frac{\mu\:\:}{2t}}{1+\frac{2{p}_{dc}}{\mu\:t\eta\:}}$$\end{document} 28\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:P=\frac{1}{2}+\sqrt{d\left(1-d\right)}, \:d=\frac{1-V}{2-\frac{\mu\:}{{t}_{F}}}$$\end{document} The system performance of each filter is determined by applying its spectral characteristics to the aforementioned equations. For the configuration with two cascaded bandpass filters (Fig. 2(a)), the channel isolation between 1310 and 1550 nm is set to be 80 dB, considering the experimentally measured channel isolation (~ 40 dB) of a single, commercially available. 1310-nm bandpass filter. For the configuration using a uniform 1310-nm FBG and a 1310-nm centered fiber-optic circulator (Fig. 2(b)), the total channel isolation is 74 dB, considering an FBG isolation of 58 dB and a fiber-optic circulator isolation of 16 dB. For the configuration using an apodized 1310-nm FBG and a 1310-nm centered fiber-optic circulator (Fig. 2(c)), the total channel isolations are 93 dB and 210 dB for Gaussian and raised-cosine apodization functions, respectively, including a fiber-optic circulator isolation of 16 dB. Theoretically, the maximum allowable QBER for secret key distillation in the BB84 QKD protocol is 11% with consideration of error correction and privacy amplification.22,36 In practice, this value is lower than the theoretical value due to finite-size effects and non-ideal error-correction efficiency. A practical value of 9% is adopted, consistent with the experimental setup in [22]. The distillation limit (dashed line in figure) indicates the maximum QBER that can be distilled, and serves as a benchmark for various filter configurations. For secret keys that are essential for QKD channel encryption, a 256-bit secret key is assumed to be updated every 60 s to divide the symmetric key between Alice and Bob, requiring a SKR of 256 bits/60 seconds = 4.27 bps per channel. Note that the following calculation results were obtained, based on the aforementioned equations using our made MATLAB code. In this calculation, the parameters were carefully selected to reflect realistic conditions. The insertion loss and channel isolation of the circulator were set to experimentally measured values of 1.2 dB and 8 dB, respectively. Other component parameters were derived from manufacturer specifications and Ref. [22]. Two cascaded bandpass filters Figure 6 presents our calculated (a) QBER and (b) SKR using a configuration of two cascaded bandpass filters with varying numbers of classical C-band WDM channels. Since the assumed system employs two encryption channels, a minimum SKR of 8.6 bps is required, indicated in the figure as the AES-256 limit (dashed line). Our calculations reveal that the fiber link distance of the QKD channel reaches 31 km with four classical C-band WDM channels only, which can be used for data encryption and distillation in the current setup. However, the QKD fiber link distance significantly decreases when the number of classical WDM channels is increased up to 50. With 50 classical WDM channels, the QKD fiber link distance is only 18.3 km due to WDM channel crosstalk and spontaneous Raman scattering noise. Fig. 6Calculated (a) QBER and (b) SKR for two cascaded bandpass filters with varying numbers of classical C-band WDM channels. Calculated (a) QBER and (b) SKR for two cascaded bandpass filters with varying numbers of classical C-band WDM channels. A uniform FBG and a fiber-optic circulator Figure 7 presents our calculated (a) QBER and (b) SKR using a configuration of a 1310-nm uniform FBG and a 1310-nm fiber-optic circulator for various numbers of classical C-band WDM channels. The fiber link distance of the QKD channel reaches 27.1 km in the case with only four classical C-band WDM channels, which can be used for data encryption and distillation in the current setup. However, the QKD fiber link distance significantly decreases when the number of classical WDM channels is increased up to 50. With 50 classical WDM channels, the QKD fiber link distance is only 11 km due to WDM channel crosstalk and spontaneous Raman scattering noise. The findings indicate that the 1310-nm uniform FBG and the 1310-nm fiber-optic circulator offer poorer QKD channel separation performance than a conventional configuration of two cascaded bandpass filters. Fig. 7Calculated (a) QBER and (b) SKR of a 1310-nm uniform FBG with a 1310-nm fiber-optic circulator for various numbers of classical C-band WDM channels. Calculated (a) QBER and (b) SKR of a 1310-nm uniform FBG with a 1310-nm fiber-optic circulator for various numbers of classical C-band WDM channels. A Gaussian apodized FBG and a fiber-optic circulator Figure 8 presents our calculated (a) QBER and (b) SKR using a configuration of a 1310-nm Gaussian apodized FBG and a 1310-nm fiber-optic circulator for various numbers of classical C-band WDM channels. The fiber link distance of the QKD channel reaches 36.6 km in four classical C-band WDM channels only, which can be used for data encryption and distillation in the proposed system. Using the 1310-nm Gaussian apodized FBG and 1310-nm fiber-optic circulator effectively suppresses classical C-band WDM channels even though the number of WDM channels is increased to 50 due to such a high channel isolation of 93 dB between 1310 and 1550 nm. With 50 classical WDM channels, the QKD fiber link distance is 35.4 km, which is very close to the distance achievable with only four classical C-band WDM channels, which can be used for data encryption and distillation exist for the proposed system. Fig. 8Calculated (a) QBER and (b) SKR of a 1310-nm Gaussian apodized FBG with a 1310-nm fiber-optic circulator for various numbers of classical C-band WDM channels. Calculated (a) QBER and (b) SKR of a 1310-nm Gaussian apodized FBG with a 1310-nm fiber-optic circulator for various numbers of classical C-band WDM channels. A raised-cosine apodized FBG and a fiber-optic circulator As illustrated in Fig. 5, the calculated reflectivity spectra of three different FBGs indicate that applying a raised-cosine apodization function significantly improves channel isolation between 1310 and 1550 nm. In our FBG design, this function yields a channel isolation of 210 dB. Figure 9 presents our calculation of QBER and SKR as a function of fiber link distance. The QKD channel achieves a fiber link distance of 36.6 km with four classical C-band WDM channels only, which can be used for data encryption and distillation in the proposed system. With 50 classical WDM channels, the QKD fiber link distance is 35.5 km. Compared to the 1310-nm Gaussian apodized FBG with a 1310-nm fiber-optic circulator, a negligible enhancement of fiber link distance is observed. This finding indicates that the 93 dB channel isolation provided by the 1310-nm Gaussian apodized FBG with a 1310-nm fiber-optic circulator is high enough to effectively suppress the classical WDM channel crosstalks. Fig. 9Calculated (a) QBER and (b) SKR of a 1310-nm raised-cosine apodized FBG with a 1310-nm fiber-optic circulator for various numbers of classical C-band WDM channels. Calculated (a) QBER and (b) SKR of a 1310-nm raised-cosine apodized FBG with a 1310-nm fiber-optic circulator for various numbers of classical C-band WDM channels. Finally, Table 3; Fig. 10 compare the system performance across various filter configurations in terms of QKD channel fiber link distance according to the number of classical C-band WDM channels. Figure 10(a) and 10(b) present the simulation results under the coexistence of 4 and 50 classical channels, respectively. Since the performance difference between the Gaussian and raised-cosine apodized filters was negligible, we compared three representative architectures: the cascade Bandpass Filter, the uniform FBG and circulator, and the Gaussian apodized FBG and circulator. The results indicate that while the conventional cascade bandpass filter suffers from high insertion loss, the uniform FBG approach—intended to reduce this loss—fails to provide sufficient channel isolation, leading to severe degradation in QKD performance. In contrast, the Gaussian apodized FBG structure successfully overcomes these trade-offs by enhancing channel isolation, resulting in a significant improvement in QKD performance. Furthermore, comparing Figs. 10(a) and 10(b) reveals that the Gaussian apodized FBG maintains its performance superiority even as the number of classical channels increases, demonstrating greater robustness against high-capacity WDM environments compared to other structures. The filter configuration using a 1310-nm Gaussian apodized FBG with a 1310-nm fiber-optic circulator, providing a 93 dB channel isolation, is identified as the optimal filter option for the investigated DV QKD-over-DWDM communication system with a QKD channel in the O-band. Table 3System performance comparison of various filter configurations in terms of QKD channel fiber link distance according to the number of classical C-band WDM channels.Filter ConfigurationNumber of classical C-band WDM channelsQKD channel fiber link distance [km]Two cascaded Bandpass filters431.01027.42023.85018.3A 1310-nm uniform FBG and a 1310-nm fiber-optic circulator427.11021.72017.15011.0A 1310-nm Gaussian apodized FBG and a 1310-nm fiber-optic circulator436.61036.42036.25035.4A 1310-nm raised-cosine apodized FBG and a 1310-nm fiber-optic circulator436.61036.42036.25035.5 Table 3System performance comparison of various filter configurations in terms of QKD channel fiber link distance according to the number of classical C-band WDM channels.Filter ConfigurationNumber of classical C-band WDM channelsQKD channel fiber link distance [km]Two cascaded Bandpass filters431.01027.42023.85018.3A 1310-nm uniform FBG and a 1310-nm fiber-optic circulator427.11021.72017.15011.0A 1310-nm Gaussian apodized FBG and a 1310-nm fiber-optic circulator436.61036.42036.25035.4A 1310-nm raised-cosine apodized FBG and a 1310-nm fiber-optic circulator436.61036.42036.25035.5 System performance comparison of various filter configurations in terms of QKD channel fiber link distance according to the number of classical C-band WDM channels. Fig. 10Calculation results of QBER(dashed) and SKR(solid) versus fiber length under varying channel conditions: (a) 4 WDM Channels (b) 50 WDM channels. Calculation results of QBER(dashed) and SKR(solid) versus fiber length under varying channel conditions: (a) 4 WDM Channels (b) 50 WDM channels. Conclusion This study theoretically investigated the optimum filter configuration for separating the QKD channel from classical C-band WDM channels in a DV QKD-over-DWDM communication system with a QKD channel in the O-band. Specifically, this work considered a standard optical communication environment where the optical power of classical channels are not attenuated purposely, aiming to address a more realistic implementation scenario. Three filter configurations are evaluated in terms of QBER, SKR, and fiber link distance: two cascaded 1310-nm bandpass filters, a 1310-nm uniform FBG and a 1310-nm fiber-optic circulator, and a 1310-nm apodized FBG and a 1310-nm fiber-optic circulator. The optical filter configuration composed of a Gaussian apodized FBG and a 1310-nm fiber-optic circulator achieves channel isolation of ~ 93 dB between 1310 nm and 1550 nm wavelengths, sufficient for reliable detection of QKD channels with multiple classical WDM channels. With 50 classical WDM channels, the QKD fiber link distance is 35.4 km, nearly matching the 36.6 km distance achieved with only four classical C-band WDM channels, which can be used for data encryption and distillation in our proposed architecture. This fiber link distance represents a substantial enhancement over the conventional filter configuration of two cascaded 1310-nm bandpass filters, which yield 18.3 km with 50 WDM channels and 31.0 km with four WDM channels. Finally, the impact of further increasing FBG filter channel isolation by adopting a raised-cosine function was also investigated. For the investigated system configuration with a QKD channel in the O-band, no significant improvement in QKD fiber link distance is observed, despite its high channel isolation performance (210 dB). However, such a high isolation filter from raised-cosine apodized FBG would be highly useful for QKD channel separation in QKD-over-WDM systems with QKD channels in the C- or L-bands. In such cases, a Gaussian apodized FBG cannot provide sufficient channel isolation when the wavelength separation between QKD and classical WDM channels is very small.