The Sensing Attack: Mechanism and Deployment in Submarine Cable Systems

Abstract

Submarine cable systems, serving as the critical backbone of global communications, face evolving resilience threats. This work proposes a novel sensing attack that utilizes ultra-narrow-linewidth lasers to surveil these infrastructures. First, the Narrowband Jamming Attack (NJA) is introduced as an evolution of conventional physical-layer jamming. NJA is divided into three categories according to the spectral position, and the non-overlapping class represents the proposed sensing attack. Its operational principles and the key parameters determining its efficacy are analyzed, along with its deployment strategy in submarine cable systems. Finally, the sensing capability is validated via OptiSystem simulations. Results demonstrate successful localization of vibrations within the 50–200 Hz range on a 1 km fiber, achieving a spatial resolution of 1 m, and confirm the influence of vibration parameters on sensing performance. This work reveals that the proposed sensing attack has the potential to covertly monitor environmental data, thereby posing a threat to information security in submarine cable systems.

1. Introduction

International submarine cable networks, serving as the critical information infrastructure, carry over 99% of global intercontinental data traffic. The resilience of submarine cable systems is a worldwide concern, with frequent incidents of intentional destruction and infrastructure interference [1]. In 2024, the International Telecommunication Union (ITU) and the International Cable Protection Committee (ICPC) jointly established an international advisory body on submarine cable resilience to strengthen resilience of submarine cable networks [2]. However, advanced attack techniques targeting submarine cables have also evolved, posing potential threats to information security. Conventional jamming attacks represent one of the classical physical-layer attacks documented in the literature [3]. The development of advanced narrowband laser sources has further refined this threat, leading to the Narrowband Jamming Attack. By leveraging Distributed Optical Fiber Sensing (DOFS) technology, this evolved attack exhibits sensing capabilities, making it more hazardous than conventional jamming attacks. This enhanced threat is therefore referred to as a sensing attack.

DOFS technology provides a powerful solution for continuous, real-time monitoring of optical communication networks, enabling the measurement of various parameters such as temperature, strain, vibration, and acoustic signals [4,5]. Bell Laboratories proposed the integration of Distributed Acoustic Sensing (DAS) systems into Wavelength Division Multiplexing (WDM) systems [6], making it feasible to deploy DAS in spectrum-constrained submarine cable systems. DAS technology has been employed in various scenarios within submarine cable systems, including submarine cable location [7], monitoring of geological activities and structures [8,9], monitoring volcanic activity [10], and tracking of whales [11]. Nevertheless, as DAS technology matures in submarine cable systems, its potential as an attack vector increases accordingly. Sensing attacks based on DAS technology aim to conduct surveillance of the environment around submarine cables. Therefore, it is crucial to investigate such threats and manage the associated risks.

This paper proposes a Narrowband Jamming Attack (NJA), an evolution of conventional jamming attacks. The proposed NJA is categorized into three types according to its consequences, and the concept of the new sensing attack is clarified. The principles and critical parameters of the sensing attack are analyzed, and a deployment strategy within submarine cable systems is presented. Finally, multi-point sensing of the proposed attack was validated through OptiSystem simulations. The optimal spatial sampling interval is identified, and the impact of vibration frequency and vibration amplitude on sensing performance is discussed. In a 1 km test system, the sensing attack successfully detected vibrations within the 50–200 Hz range, achieving a spatial resolution of 1 m.

2. Principle and Key Parameters of the Sensing Attack

2.1. Attack Modes: From Jamming to Sensing

A jamming attack is an active attack performed by introducing a jamming signal into the transmission to partially or completely disrupt it. The disturbance may be of low magnitude, merely increasing the error rate in the demodulated signal. It may also take the form of an unmodulated high-power signal, the purpose of which is to saturate the receiver in the central node [12]. In jamming attacks, the spectral position of the jamming signal is critical, and all such attacks are categorized into two types based on their spectral relationship to the useful signal. Out-of-band attacks occur when their bands differ, whereas in-band attacks arise when both occupy the same band [12,13]. The NJA represents an evolution of conventional physical-layer jamming techniques. NJA injects a highly coherent, narrowband signal at a specific wavelength into the system, which can lead to the leakage of environmental data. It is noteworthy that if such an attack occurs within a submarine cable system, it could pose a potential threat in the field of information security.

For the NJA, the spectral position of the jamming signal remains a crucial consideration. Submarine cable systems carry normal communication traffic as well as periodic probe signals used to verify system status (further details are provided in Section 3). Consequently, NJAs can be categorized into three types based on the spectral position of the jamming signal, as illustrated in Figure 1. If the band of the narrowband jamming signal overlaps with the traffic signals, it disrupts communications and is thus termed communication interference (this case is analogous to conventional in-band jamming). If the band overlaps with the probe signal, it affects the system’s monitoring results and is termed monitoring interference. Finally, if the jamming signal is out-of-band, the system can detect the backscattered Rayleigh light generated by the input optical signal. This enables the measurement of weak ambient vibration. Hence it is named the sensing attack. Unlike most jamming attacks that solely disrupt communication (a denial-of-service attack), the sensing attack represents a more sophisticated threat. By exploiting the high coherence of a narrowband laser, it enables the extraction of ambient vibrational information surrounding the cable. This transforms the infrastructure from a mere transmission medium into a sensor, thereby compromising information security and making it more hazardous.

2.2. Principle of the Sensing Attack

The sensing attack fundamentally leverages DAS technology to achieve sensing of environmental data around submarine cables. The principle is illustrated in the lower right part of Figure 2. Pulsed light is inserted into one end of the optical fiber, and distributed information along the fiber is obtained through backscattered Rayleigh light. Since an ultra-narrow-linewidth laser with strong coherence is inserted at the terminal station, the backscattered light results from the interference of all backscattered Rayleigh signals generated within the pulse width of the input light. When a vibration occurs around the fiber, the change in the refractive index of the optical fiber causes a corresponding phase change in the backscattered Rayleigh light, thereby altering the interference signal. Thus, multi-point sensing can be achieved simultaneously. Theoretically, the phase difference of Rayleigh backscattered light generated at both ends of the disturbed fiber segment has a linear relationship with the vibration. Therefore, quantitative measurement of the vibration amplitude can be realized by demodulating this phase difference. This constitutes the principle of the sensing attack.

2.3. Key Parameters of the Sensing Attack

All commercially deployed DAS systems in current engineering practice are physically based on φ-OTDR [14]. The key parameters of φ-OTDR are closely related to the effectiveness of the sensing attack.

1.

Sensing Range (L)

The sensing range is proportional to the energy of the pulsed light. The primary method to improve the sensing range is to increase the energy of the pulsed light, which can be achieved by either increasing the peak power of the pulsed light (P_peak) or widening the pulse width (τ). However, an excessively high P_peak can induce nonlinear effects such as stimulated Brillouin scattering [15]. Therefore, increasing the pulse width τ is a more commonly used approach to enhance the pulsed light energy.

2.

Spatial Resolution

Spatial resolution refers to the ability to distinguish two vibrations, which affects positioning accuracy. As shown in Equation (1), the spatial resolution Δz is determined by the pulse width τ of the pulsed light:

Δz = cτ/2n,

(1)

A smaller τ results in a smaller Δz, meaning higher spatial resolution. Thus, the setting of τ requires a trade-off between high spatial resolution and long sensing range.

3.

Frequency Measurement Range

To avoid overlapping of consecutive pulses in the optical fiber, the upper frequency detectable by the system f_{BW_max} is constrained by the pulse repetition frequency f_rep, which depends on the sensing range L, as shown in Equations (2) and (3):

f_rep ≤ c/(2nL),

(2)

f_{BW_max} ≤ f_rep/2

(3)

Consequently, a higher pulse repetition frequency allows the system to measure higher frequencies, but simultaneously reduces the achievable sensing range.

3. Deployment Strategy of the Sensing Attack in Submarine Cable Systems

Figure 2 shows schematic diagrams of the spectrum allocation and the deployment of the sensing attack in both unrepeatered and repeatered submarine cable systems. In repeatered systems, the available frequency band is determined by the gain bandwidth of the repeaters. The central portion of this band is allocated to Dense Wavelength Division Multiplexing (DWDM) service channels, while the edges are reserved as monitoring channels for legitimate periodic probe signals. Current monitoring in submarine cable systems primarily utilizes Coherent Optical Time-Domain Reflectometry (C-OTDR) technology [16,17]. In unrepeatered systems, while there is no amplifier bandwidth limitation, the available bandwidth is still finite and is entirely occupied by DWDM service channels.

To launch the attack, a DAS system must be deployed at a cable terminal station to insert jamming light near the DWDM channel guard bands or the monitoring channels at the spectrum edges. The phase-sensitive Optical Time-Domain Reflectometry (φ-OTDR) system employs an ultra-narrow-linewidth, highly coherent laser to achieve high sensitivity. This enables the sensing of environmental data around the cable, such as vessel movements, fish activities, or anomalous anchoring events. Specifically, the bandwidth of the jamming light needs to be less than 3 kHz. Long-range DAS systems impose even stricter requirements on laser linewidth; for instance, a 50 km system typically requires a bandwidth of less than 1 kHz. Existing submarine cable systems primarily use DWDM channel spacings of 50 GHz and 100 GHz [18]. Therefore, the narrowband light can be inserted into these channel spacings (guard bands) to realize the sensing attack.

4. Simulation of the Sensing Attack Based on OptiSystem

This section presents simulation results based on the OptiSystem software. First, the computational model employed in the simulations is introduced, followed by a discussion on the selection of the spatial sampling interval. Subsequently, multi-point sensing capability is validated and then discussed the impact of vibration frequency and amplitude on the sensing performance. It should be noted that the φ-OTDR component used in this simulation is a newly added module in the software. Currently, it only supports basic unamplified sensing. Therefore, all simulation results presented hereafter are derived from short-range validation and do not involve optical repeaters.

4.1. Discrete Scattering Model

A discrete model is utilized in the simulation to describe Rayleigh backscattering. In this model, an optical fiber with a total length L is divided into N segments based on the distribution of scattering centers. Each segment has a length of ΔL = L/N. The Rayleigh backscattered signal acquired by the system is formed by the superposition of scattered signals from these N scattering centers [19].

4.2. Optimal Spatial Sampling Interval

The value of N is critically important in the process of discretizing the optical fiber. According to the spatial sampling theorem, if N is too small, high-frequency vibrations cannot be accurately reconstructed. Conversely, if N is too large, the fiber is segmented too finely, drastically reducing the number of scattering sources per segment and causing the scattered signal to be degraded by coherent fading noise. Therefore, the value of N must be moderate and determined based on the vibration frequency. For a 1000 m link, a 100 Hz vibration was applied at the 200 m point, and the sensing results under different spatial sampling intervals were observed.

As shown in Figure 3, when N is 40 (the minimum value satisfying the spatial sampling theorem), a vibration at 200 m can be observed, but the noise level is high, resulting in mediocre overall performance. When N is set to 150, the system’s quantization error decreases, the Signal-to-Noise Ratio (SNR) increases, and the sensing effectiveness becomes the best among the three cases. When N is further increased to 500, the sensing performance degrades due to coherent fading noise.

4.3. Multi-Point Sensing Performance

This section verifies the multi-point sensing capability and discusses the impact of vibration frequency and amplitude on the sensing performance. Figure 4 shows three sets of sensing results. For a 1 km link, two vibrations are introduced at 200 m and 400 m, respectively. The vibration frequency and amplitude are modified to observe different sensing results. First, all three sensing results successfully located two vibrations and accurately identified their frequencies. Specifically, comparing Figure 4b,d, when the vibration frequency changes, the peak position in the frequency result shifts accordingly. Comparing Figure 4a,e, as well as Figure 4b,f, when the vibration amplitude changes, the locations of the vibration positioning peaks and the frequency peaks remain unchanged, but their amplitudes vary. These observations illustrate the influence of vibration frequency and amplitude on the sensing performance. The default system parameters are listed in

Table 1. Default parameters of the simulation system.

Parameters	Value	Units
Reference wavelength	1550	nm
Laser linewidth	3	kHz
Fiber attenuation	0.2	dB/km
Fiber core refractive index	1.5	/
Fiber effective area	80	µm2
Rayleigh backscattering	50 × 10−6	1/km
Pulse duration	10	ns
Pulse repetition rate	10	kHz

.

5. Conclusions

This work establishes the feasibility and outlines the mechanisms of a novel sensing attack that exploits φ-OTDR-based DAS technology. First, NJA and its clear taxonomy are presented, introducing the concept of the sensing attack. Building upon this foundation, the underlying principles of the sensing attack and the critical parameters governing its performance are thoroughly analyzed. Finally, simulation results successfully validate the efficacy of the attack, demonstrating multi-point vibration sensing with high spatial resolution and the influence of vibration frequency and amplitude on the sensing effect. This work validated the sensing attack through simulation. While the results conclusively demonstrate the theoretical feasibility and potential impact of the sensing attack, experimental validation remains an essential next step in future work. These findings expose a latent risk associated with the increasing integration of DAS into critical infrastructure, underscoring the urgent need to develop robust countermeasures and defense strategies to mitigate this emerging threat and secure global data transmission networks.