Submarine fiber optical networks represent a crucial backbone for the worldwide digital economy. Space division multiplexing (SDM) has the potential to solve the capacity crunch of traditional standard single-mode fiber (SSMF), with the merits of reducing network costs and enhancing power efficiency [1]. To satisfy the ever-increasing demand for higher bandwidth and faster connections across various multimedia and data services, such as big data, cloud computing, video streaming, the Internet of Things, machine-to-machine communication, and remote surgery, the first generation of SDM systems has been successfully introduced in submarine telecommunication networks, and the deployed SSMF is expected to be progressively replaced with multi-core fiber (MCF) by the end of 2026 [2]. Since optical fiber can be used as waveguide sensors, integrating distributed sensing into fiber optical communication systems has garnered significant research interest recently by transforming the submarine MCF infrastructure into a distributed sensing network that enables intelligent functionalities [3]. It is identified that both the state of polarization and optical phase obtained from telecom transponders can be employed for vibration sensing [4–6]. However, those techniques do not provide the required localization accuracy for distributed sensing applications, especially in more complex environments. Alternatively, distributed optical fiber sensing (DOFS) utilizing the backscattered light allows for measuring various physical parameters with exceptionally high spatial resolution. Among DOFS techniques, phase-sensitive optical time-domain reflectometry ($\mathrm{\Phi }$-OTDR) stands out as a promising method, due to its high spatial resolution and sensitivity for the distributed acoustic sensing (DAS) application. In particular, the notable resemblance between $\mathrm{\Phi }$-OTDR and optical communication systems, in terms of system architecture, signal modulation, and detection, makes them well-suited for integrated sensing and communication (ISAC).

Prior works have successfully demonstrated ISAC over a weakly coupled MCF. A joint optical fiber-based communication and sensing technology was deployed over a 1 km MCF in a harbor canal at the Red Sea, achieving a transmission capacity of 3.2 Gb/s and a vibration sensing frequency of 1 kHz with a spatial resolution of 5 m [7]. Furthermore, the distributed sensing capability based on $\mathrm{\Phi }$-OTDR over 16.5 km deploying seven-core fiber was demonstrated for the real-time traffic monitoring, together with the wavelength/space division multiplexed dual-polarization (DP)-16QAM transmission with a capacity of 200.88 Tb/s [8]. Although the above SDM-enabled ISAC solution minimizes the detrimental interactions, separating the communication and DAS signals into two independent cores significantly sacrifices the spectral efficiency (SE), reducing transmission capacity. As a result, the efficiency of integrating the DAS and communication signals for ISAC has been intensively investigated. For example, integrating DAS and communication signals within the same fiber was proposed under the condition of wavelength-division multiplexing (WDM). In Ref. [9], the field trial of DOFS and high-speed communication with the WDM technique was reported over an operated telecom network. A 36.8 Tb/s transmission with an SE of $8.28\mathrm{bit}{\mathrm{s}}^{-1}{\mathrm{Hz}}^{-1}$ was achieved, using probabilistic-shaped (PS) DP-144QAM over 110 km SSMF. Meanwhile, both vehicle speed and vehicle density were precepted with accuracies of 98.5% and 94.5%, respectively. In Ref. [10], a DAS experiment over $>1000\mathrm{km}$ SSMF link including a mixture of field and lab fibers was demonstrated when the bi-directional in-line Raman amplification was implemented after each about 80 km span transmission. By utilizing $20\times$ frequency diversity chirped-pulses for the probing, and recovering the Rayleigh backscattering with a coherent receiver and correlation detection, a sensitivity of $100\mathrm{n\epsilon }/\sqrt{\mathrm{Hz}}$@16 Hz with 20-m spatial resolution was achieved, together with 10 Tb/s transmission capacity. However, this WDM-enabled ISAC shares the fiber with two distinct transceivers, significantly enhancing the deployment cost. Additionally, a wavelength channel is sacrificed for sensing. Later, an ISAC solution employing a linear frequency modulation carrier as the traditional optical carrier was demonstrated under the co-wavelength channel [11]. Such an approach enables the transmission of a 56 Gb/s PAM4 signal over 24.5 km SSMF and the detection of vibration at 800 Hz. However, this method only involves the intensity modulation direct detection configuration with a limited reach and transmission capacity, compared to the commonly used coherent QAM transmission.

Digital subcarrier multiplexing (DSM) technology is a key driver for developing next-generation flexible and software-configurable optical networks [12–14]. Consequently, DSM finds extensive applications in high-speed commercial pluggable transceivers. For instance, Infinera has showcased its 800G transmission by utilizing 95.6 GBaud 8-subcarrier DSM signals. Because of its flexibility of spectral allocation, DSM is particularly suitable for integrating sensing and communication signals in different frequency bands but within the same wavelength channel, via frequency division multiplexing (FDM). Recently, an ISAC scheme integrating the digital linear chirped sensing signal into the DSM signal through a shared transmitter was demonstrated [15–17], achieving 200 Gb/s DP-16QAM signal transmission over the 10 km SSMF, with the capability of vibration sensing at a frequency of 500 Hz. Although this FDM solution for ISAC can minimize the transmitter complexity, the sensing and communication ability is significantly compromised by avoiding the optoelectrical devices operated at the nonlinear region. Moreover, according to the ocean soundscapes, geological processes and human activities in the deep sea mainly generate acoustic waves with a typical frequency range from 0.1 to 10 Hz [18]. Detection and acquisition of ultra-low-frequency acoustic waves are paramount for geological exploration, early warning of natural disasters, etc. Unfortunately, the current ISAC systems reported so far are mainly focused on detecting and acquiring acoustic waves with frequencies higher than 10 Hz, due to interference fading and inevitable frequency drift of lasers, as summarized in Table 1 [8–11,15,19]. As a result, it remains an open question on how to fully leverage a fiber to achieve both high-capacity transmission and ultra-low-frequency DAS.Table 1.

Recent Progress in ISAC in an Optical Fiber

This work proposes a seven-core fiber-based ISAC system utilizing the same wavelength channel for concurrent large-capacity data transmission and ultra-low-frequency DAS. Initially, the DSM and sensing signals are generated independently, but with a shared laser source, and then combined in the optical domain, instead of the digital domain [17]. Such a configuration significantly simplifies the performance optimization, as both sensing and DSM signals can be, respectively, optimized without the interference, ultimately enhancing the co-wavelength ISAC signal quality. Under the optimal protection interval (PI) bandwidth, the proposed ISAC system can fully utilize the communication and sensing capability. Consequently, we can successfully achieve a sensitivity of both $3.89\mathrm{n\epsilon }/\sqrt{\mathrm{Hz}}$ @0.1 Hz and $0.18\mathrm{n\epsilon }/\sqrt{\mathrm{Hz}}$ @10 Hz with a spatial resolution of 20 m and a record transmission capacity of 241.85 Tb/s for the ISAC when $96\times 7$ wavelength-space division multiplexed DP-16QAM DSM signals are transmitted over a 38 km seven-core fiber.

Figure 1 depicts the operation principle of the proposed ISAC system, which mainly comprises an integrated transmitter, a seven-core fiber link, and sensing and communication receivers. In the integrated transmitter, the continuous-wave (CW) light is split into two tributaries, which are separately modulated to generate the chirped pulse (CP) for sensing and DSM signals for communication. The CP and DSM signals can be, respectively, expressed as $$S_{\mathrm{CP}}(t)=e^{j2\pi \left(f_{1}t+\frac{1}{2}k{t}^2\right)}\mathrm{rect}\left(\frac{t}{{\tau }_p}\right)·\exp (j2\pi f_{c}t+j\theta (t)),$$(1)$$S_{\mathrm{DSM}}(t)=\sum _{k=1}^{N_{\mathrm{SC}}}{s}_k(t)\exp (-j2\pi f_{k}t)·\exp (j2\pi f_{c}t+j\theta (t)),$$(2)where $f_1$ is the initial frequency of the CP signal at the baseband, $k=B_{\mathrm{CP}}/{\tau }_p$ is the frequency slope, ${\tau }_p$ is the pulsewidth, and $B_{\mathrm{CP}}$ is the chirp-induced spectral bandwidth. $f_c$ and $\theta (t)$ are the central frequency of the optical carrier and phase noise, respectively. $s_k(t)$ is the baseband signal corresponding to the $k$th subcarrier for DSM, $f_k$ represents the central frequency of the $k$th subcarrier, and $N_{\mathrm{SC}}$ corresponds to the total number of subcarriers. For traditional DSM transmissions, there is no PI to maximize the SE, namely, $\mathrm{\Delta }{f}_{\mathrm{SC}}=f_k-f_{k-1}=B_{\mathrm{SC}}(1+{\alpha }_{\mathrm{RRC}})$, where $\mathrm{\Delta }{f}_{\mathrm{SC}}$ is the frequency spacing between adjacent subcarriers. $B_{\mathrm{SC}}$ is the symbol rate for each subcarrier, and ${\alpha }_{\mathrm{RRC}}$ is the roll-off factor of the rooted raised cosine (RRC) filter for the spectral shaping. When it comes to the ISAC system, PI is preserved between the two central subcarriers of the DSM signal in order to avoid the spectral overlapping between the DSM and CP signals, namely, $f_M-f_{M-1}=\mathrm{\Delta }{f}_{\mathrm{SC}}+{\mathrm{BW}}_{\mathrm{PI}}$, where $M$ and $M-1$ are the corresponding subcarrier indices. ${\mathrm{BW}}_{\mathrm{PI}}$ is the bandwidth of PI, which needs to be optimized. Please note that the frequency spacing between other subcarriers remains to be $\mathrm{\Delta }{f}_{\mathrm{SC}}$. Moreover, placing the PI in other spectral positions of the DSM signal is theoretically feasible but not practical, because it needs to shift the central frequency of the CP signal, which may potentially increase the complexity of the sensing transmitter and cause the optical filtering effect on the CP signal’s transmission over the fiber optical link. Thereafter, the CP and DSM signals are combined via FDM, and the obtained ISAC signals for a specific wavelength channel can be expressed as $$S_{\mathrm{ISAC}}(t)=S_{\mathrm{CP}}(t)+S_{\mathrm{DSM}}(t)\mathrm{.}$$(3)

To fully use spectral resources, other wavelength channels can be designed for communications solely by transmitting the DSM signals without PI, as shown in Fig. 1. However, some wavelength channels can be elaborately chosen to transmit the ISAC signal. In such a way, we can leverage the wavelength diversity technique to increase the DAS sensitivity [20], at the cost of sacrificing communication capacity. After wavelength division multiplexing (WDM), the generated optical signal is launched into the seven-core fiber through a fan-in/fan-out (FIFO) device. Similarly, the WDM signal for other cores can be generated and launched into the seven-core fiber. After the SDM transmission, the space division de-multiplexing is realized using another FIFO device, followed by wavelength division de-multiplexing. At the remote communication receiver, the forward propagated optical signal is coherently detected, with the help of another semiconductor laser acting as a local oscillator (LO). After analog-to-digital conversion (ADC), the removal of the CP can be easily achieved by using another RRC-matched filter during the subcarrier de-multiplexing process after the frequency offset compensation [21]. After that, the receiver-side digital signal processing (DSP) for the ISAC signals is the same as that of traditional DSM signals [22].

The Rayleigh backscattered ISAC signals from the seven-core fiber are first space division de-multiplexed and then amplified to compensate for the transmission loss. Thereafter, the backscattered ISAC signals are introduced to an optical filter, for wavelength division de-multiplexing, as well as rejecting DSM signals and out-of-band noise. If we reasonably ignore the fiber nonlinearity, the filtered backscattered signals can be represented as $$\begin{gathered} E_{\mathrm{CP}}(t)=\sum _{i=1}^{N_{\mathrm{CA}}}R({\tau }_i)e^{j2\pi (f_1(t-{\tau }_i)+\frac{1}{2}k{(t-{\tau }_i)}^2)+j{\phi }_i}·\mathrm{rect}\left(\frac{t-{\tau }_i}{{\tau }_p}\right) \\ ·\exp (j2\pi f_c(t-{\tau }_i)+j\theta (t-{\tau }_i))+N(t), \end{gathered}$$(4)where $N_{\mathrm{CA}}$ is the number of scatter points along the entire seven-core fiber, ${\tau }_i$ is the flight time of the $i$th scatter point, and $R({\tau }_i)$ is the scattering coefficient of the $i$th scatter point. ${\phi }_i$ is the phase from the vibration-induced refractive index change of the $i$th scatter point. $N(t)$ represents the additive noise as well as residual DSM signal after optical filtering, which can be regarded as noise to the CP signal [11]. For ease of discussion, we assume $N(t)$ is small enough with respect to the backscattered CP after the optical amplification; the received electrical signals after direct detection can be expressed as $$\begin{gathered} R(t)=E_{\mathrm{CP}}(t)·E_{\mathrm{CP}}^{*}(t) \\ =2\sum _{i=1}^{N_{\mathrm{CA}}}\sum _{j=1}^{N_{\mathrm{CA}}}R({\tau }_i)R({\tau }_j)\cos (2\pi ((f_1+f_c)({\tau }_i-{\tau }_j)+\frac{1}{2}k({\tau }_j-{\tau }_i)(2t-{\tau }_i-{\tau }_j))+\theta (t-{\tau }_i)-\theta (t-{\tau }_j) \\ +{\phi }_i-{\phi }_j)\mathrm{rect}\left(\frac{t-{\tau }_i}{{\tau }_p}\right)\mathrm{rect}\left(\frac{t-{\tau }_j}{{\tau }_p}\right)\mathrm{.} \end{gathered}$$(5)

Equation (5) implies that the received signals exhibit a random speckle-like pattern. The pattern remains unchanged if no external perturbation exists on the seven-core fiber. When the external vibration occurs, a phase difference of ${\phi }_i-{\phi }_j$ between scattering points happens. The phase variation can be compensated by the pulse frequency shifting, thus restoring the original speckle-like pattern. The frequency shift is linearly related to the vibration, due to frequency-to-time-mapping of the CP. Namely, the amount of local temporal delay in the trace linearly correlates with the vibration [23–25]. Therefore, vibration leads to a longitudinal shift of the local intensity trace. The temporal shift can be detected by trace-to-trace moving correlations, enabling both detection and quantification of the vibration, according to the following equation [26]: $$\left(\frac{\mathrm{\Delta }n}{n}\right)=\frac{\mathrm{\Delta }f}{f_c}=-\left(\frac{1}{f_c}\right)·\left(\frac{B_{\mathrm{CP}}}{{\tau }_p}\right)·\mathrm{\Delta }t\approx -0.78·\mathrm{\Delta }\epsilon ,$$(6)where $\mathrm{\Delta }n$ is the vibration induced refractive index change, $n$ is the fiber refractive index, $\mathrm{\Delta }f$ is the corresponding frequency shift, $\mathrm{\Delta }t$ is the temporal shift, and $\mathrm{\Delta }\epsilon$ is the applied strain. We should note that the spatial resolution of DAS refers to the minimum distance between two strain perturbation points to independently resolve variations in the optical fiber trace. Thus, it can be represented as $$\mathrm{\Delta }z=\frac{c{\tau }_p}{2n}\mathrm{.}$$(7)

The proof-of-concept experimental setup is shown in Fig. 3(a). A CW laser with a linewidth of 100 Hz and a central wavelength of 1550.12 nm is employed as the optical carrier to generate ISAC signal. After splitting the CW light into two copies, the light at the upper branch for sensing is first modulated into a linearly chirped light, when an in-phase quadrature (IQ) modulator is driven by a voltage-controlled oscillator (VCO) that is controlled by an arbitrary function generator (AFG). After the pre-amplification, in order to separate the scattered light returned from different positions of the fiber in time domain, the linearly chirped light is chopped into a 200-ns pulsed light via an acoustic-optic modulator, generating a CP with a bandwidth of 500 MHz. Consequently, the spatial resolution of DAS is 20 m, according to Eq. (6). Meanwhile, the light with the same wavelength at the lower branch for communication is introduced to a coherent driver modulator (CDM) module. The 36 GBaud DP-16QAM DSM electrical signal with four subcarriers is first generated offline with an RRC roll-off factor of 0.01. Then the DSM signals are loaded into the arbitrary waveform generator (AWG) with a sampling rate of 120 GSa/s and a resolution of 8 bits after resampling and clipping. A PI between central subcarriers of the DSM signals is reserved. Once the optical sensing and communication signals are generated, the ISAC signals are obtained via an optical coupler to combine both signals at the same wavelength channel.

To fully leverage the optical spectrum resource for the transmission capacity enhancement, the other 95 WDM channels from 1527.60 nm to 1565.50 nm with a channel spacing of 50 GHz are generated without inserting the probing signal due to the hardware constraints. Thus, we do not introduce the wavelength diversity in the experimental demonstration. Specifically, as for four WDM channels surrounding the 1550.12-nm channel, we use tunable external cavity lasers (ECLs) with a linewidth of 100 kHz as optical sources. Those optical carriers are modulated by the 45 GBaud DP-16QAM DSM signal with four subcarriers without the reserved PI, using another AWG and CDM. In addition, we utilize an amplified spontaneous emission (ASE) source shaped by a wave shaper (WSP, 4000B) to emulate the other 91 WDM channels. After the pre-amplification, all 96 WDM channels are combined, whose spectrum is shown in Fig. 3(b), and finally fed into the core-4 via a self-fabricated FIFO device. To further generate the communication signals for the other six cores, we first shape another ASE source utilizing another WSP to obtain 96 WDM channels with a 50-GHz grid, then spill the WDM signal into six copies, and finally introduce those copies into six cores via the same FIFO device after the pre-amplification.

After 38-km seven-core fiber transmission and space division de-multiplexing, a piezoelectric transducer (PZT) is applied to generate an ultra-low-frequency acoustic wave. Since the pigtail of used PZT is not compatible with the seven-core fiber, it is used after the transmission over 38 km seven-core fibers and the spatial division de-multiplexer. As a result, the space diversity is not applied in the current experimental demonstration. At the far-end communication receiver (Rx), the forward propagated signal is filtered for wavelength division de-multiplexing, amplified, and finally filtered to reject the out-of-band noise. Another tunable ECL with a linewidth of 100 kHz is used as the LO for coherent detection. After being sampled by an 80-GSa/s digital sampling oscillator (DSO), the Rx DSP for communication mainly includes chromatic dispersion (CD) compensation, coarse frequency offset compensation, subcarrier de-multiplexing along with CP signal removal utilizing a matched RRC filter, adaptive equalization, carrier phase recovery, and average BER calculation among all subcarriers. At the sensing Rx, the back-scattered ISAC signals are first passed through an optical filter (Alnair Labs, BVF-300CL) with a 3-dB bandwidth of 6 GHz, in order to remove the DSM signal and out-of-band noise, after the optical amplification. Then, a photodetector (PD) is used to detect the filtered signal, and the received signal is recorded by another 2-GSa/s DSO. As for the DSP for the sensing application, the spatial resolution of 20 m is used as the moving correlation window to determine local temporal delay. The strain temporal waveform is reconstructed by using the linear relationship between temporal delay and external strain. The key experimental parameters are summarized in Table 3 for convenience.Table 3.

Key Parameters of the Experiment

We first characterize the fabricated seven-core fiber, whose cross section is shown in Fig. 3(c). The measured inter-core crosstalk is as low as $-60\mathrm{dB}/100\mathrm{km}$ at 1550 nm, indicating negligible performance penalty during the SDM transmission, as shown in Fig. 4(a). Furthermore, as per Fig. 4(b), each core’s attenuation and CD parameters are around 0.18 dB/km and $20.05\mathrm{ps}{\mathrm{nm}}^{-1}{\mathrm{km}}^{-1}$ at 1550 nm, respectively.

We then study the interaction between the DAS and DSM signals. Since the DAS and DSM signals are combined in the optical domain, we can independently optimize each signal’s launch power into the fiber link in order to secure the best ISAC system performance. Afterwards, the PI bandwidth is optimized to minimize the nonlinear interactions between the DAS and DSM signals. The relationship between the achieved Q-factor of the DSM signals and the inserted PI bandwidth is presented in Fig. 5(a), with and without the presence of the CP signal. With the growing PI bandwidth, the achieved Q-factor gradually decreases, due to the bandwidth constraint of optoelectronics devices. Additionally, when the CP signal is co-propagated with the DSM signals, we observe a significant Q-factor penalty, because of the detrimental interaction between CP and DSM signals. Moreover, the Q-factor penalty is decreased with the growing PI bandwidth, indicating a reduced detrimental interaction. Meanwhile, the distributed SNR of the CP is improved, when the PI bandwidth is increased, as shown in Fig. 5(b). To secure both higher transmission capacity and better sensitivity for sensing ultra-low-frequency acoustic waves, the PI bandwidth is set as 8 GHz. We believe the PI bandwidth can be further reduced using a dedicated optical filter with a sharp response.

Next, the BER performance is characterized for all spectral and spatial channels, as shown in Fig. 6(a). Please note that we tune the central wavelengths of laser-2 to laser-5 to measure the BERs for all 95 WDM communication channels at each core. Overall, BER values of 96 WDM channels for all seven cores are below the 20% soft-decision feedforward correction coding threshold of $2\times {10}^{-2}$. Among those $672(7\times 96)$ wavelength channels, only one wavelength channel is transmitted with 36 GBaud DP-16QAM signal when CP is inserted for sensing, and the other 671 wavelength channels are transmitted with 45 GBaud DP-16QAM signals. Therefore, the aggregate capacity for communication is approximately 241.85 Tb/s $[(671\times 45\text{GBaud}+36\text{GBaud})\times 8\mathrm{bit}/\text{symbol}]$. Correspondingly, the net capacity, spectral efficiency, and net spectral efficiency can be calculated as 201.5 Tb/s, $50.4\mathrm{bit}{\mathrm{s}}^{-1}{\mathrm{Hz}}^{-1}$, and $42\mathrm{bit}{\mathrm{s}}^{-1}{\mathrm{Hz}}^{-1}$, respectively, by considering 20% overhead and the total occupied bandwidth of $96\times 50\mathrm{GHz}$. Using seven-core fiber with SDM communication technology can significantly enhance the transmission capacity, highlighting its strong potential for high-capacity data transmission within an single optical fiber.

Afterward, the time-distance mapping of the vibration is shown in Fig. 6(b) for core-4. The vibration can be discriminated around 38.72 km. Considering the experiment’s EDFA, circulator, and fiber patch cord, the location result agrees well with the expected distance of 38.72 km. The reflections due to the PC/APC facet mismatch can also be detected. In addition, two repeated periods of 10 s, corresponding to 0.1 Hz, are acquired over a 25-s duration, as shown in Fig. 6(c). The peak-to-peak value of the acoustic wave is $130\mathrm{n\epsilon }$, which agrees well with the strain generated by the PZT. Figure 6(d) displays the strain ASD. The observed vibration frequency is 0.1 Hz, consistent with the applied frequency of PZT. In addition, the calculated sensitivity by the ASD is $3.89\mathrm{\text{nε}}/\sqrt{\mathrm{Hz}}$, indicating the capability of accurately sensing the 0.1 Hz acoustic wave over the 38-km seven-core fiber. Please note that the sensing sensitivity can be further enhanced by combining the wavelength and spatial diversity technique as mentioned in Section 2 at the cost of slightly sacrificed transmission capacity. Because of the current hardware constraint, this remains our future work for experimental verification.

Furthermore, we investigate the ultra-low-frequency detection capability of the other six cores arising in the seven-core fiber. The frequency of all waveforms can be calculated as 0.1 Hz. Meanwhile, the strain sensitivities of six cores can be obtained as $4.94\mathrm{n\epsilon }/\sqrt{\mathrm{Hz}}@\text{core}\text{-}1$, $9.75\mathrm{n\epsilon }/\sqrt{\mathrm{Hz}}@\text{core}\text{-}2$, $6.10\mathrm{n\epsilon }/\sqrt{\mathrm{Hz}}@\text{core}\text{-}3$, $6.83\mathrm{n\epsilon }/\sqrt{\mathrm{Hz}}@\text{core}\text{-}5$, $5.91\mathrm{n\epsilon }/\sqrt{\mathrm{Hz}}@\text{core}\text{-}6$, and $6.61\mathrm{n\epsilon }/\sqrt{\mathrm{Hz}}@\text{core}\text{-}7$, respectively, as shown in Fig. 7(b). We should mention that the vibration detection below 0.1 Hz is theoretically feasible after the low-frequency noise is substantially suppressed [27].

Finally, we adjust the vibration frequency to 10 Hz and investigate the sensing performance for core-4 in the presence and absence of the communication signals. The sensing probe has a repetition rate of 1 kHz, and about 3000 trace periods are acquired. Under the condition of averaging eight times for demodulation, the highest detectable frequency during our investigation is 62.5 Hz. As shown in Fig. 8(a), four repeated periods of 0.1 s, corresponding to 10 Hz, are acquired over a 0.4 s duration, no matter whether the communication is powered on or off. As shown in Fig. 8(b), in the absence of a communication signal, the strain sensitivity is $0.11\mathrm{n\epsilon }/\sqrt{\mathrm{Hz}}@10\mathrm{Hz}$. When the communication signal is integrated, the strain sensitivity is degraded to $0.18\mathrm{n\epsilon }/\sqrt{\mathrm{Hz}}@10\mathrm{Hz}$, because of the detrimental interaction between DAS and DSM signals. In addition, a better strain sensitivity is guaranteed under the condition of 10 Hz vibration, compared to the case of 0.1 Hz vibration, due to the frequency-dependent noise distribution [31]. Therefore, the proposed ISAC system can detect and acquire acoustic waves with frequencies between 0.1 and 10 Hz.

We demonstrate the feasibility of integrating high-capacity optical communication with DAS in the seven-core fiber. By employing DSM and chirped-pulse through frequency division multiplexing at the same wavelength channel, the proposed ISAC system achieves a performance balance between the data transmission and DAS. With the optimal PI bandwidth, we successfully achieve a sensitivity of $3.89\mathrm{n\epsilon }/\sqrt{\mathrm{Hz}}@0.1\mathrm{Hz}$ and $0.18\mathrm{n\epsilon }/\sqrt{\mathrm{Hz}}@10\mathrm{Hz}$ with a spatial resolution of 20 m, together with a transmission capacity record of 241.85 Tb/s. Those results highlight the system’s potential to enable simultaneous high-capacity communication and distributed ultra-low-frequency vibration detection, paving the way for enhanced seismic monitoring and geological exploration through existing submarine telecommunication infrastructure.