Open Access
Add to BibSonomy
Abstract
A newly designed phase-locked (PL) Φ-OTDR system was proposed and instrumented. Field tests of water impact, anchor damage towing and tide diagnosing were carried out in a natural freshwater lake as well as the East China Sea. Personnel movement trajectory monitoring and ship flow monitoring were carried out by a buried cable along the floodplain of the Yangtze River. It proved that the proposed system can monitor the real-time status and sense the surrounding environment of existing underwater communication cables, which could be helpful for the maintenance of the cable itself as well as underwater information collection.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
With the rapid development of marine economy, the number of underwater cables increases rapidly [1]. According to the report in 2019, a total of $\$ 50$ billion has been invested in submarine cables alone globally with a total length of over 1.4 million kilometers [2]. More than $\$ 10$ trillion worth of communications traffic is traded every day [3]. However, due to potential threats such as man-made destruction, ship anchoring, biological gnawing, etc., the cable transmission system is vulnerable to failure and cause communication interruption. In recent years, underwater cable accidents have occurred frequently worldwide, which have seriously affected the safe and stable operation of cross-sea interconnected smart power grids and information communication networks. The anchor of a cargo ship led to the break of the submarine cable in Pingtan City, Fujian Province, China, resulting in power outages of the entire island [4]. Canvas fishing led to the destruction of the Trans-Pacific China-US Cable TV Network (CUCN), making domestic users unable to access foreign websites. The berthing of the cargo ship cut the Asia-Europe submarine cable, and the communication signals were interrupted for more than 10 days [5]. The M7.5 earthquake broke submarine cables off the coast of Southeast Alaska, which took several months to repair [6]. A report of the Asia-Pacific Economic Cooperation states that, with respect to hazards to cables: “The overall minimum-approach should be to build up the capacities to monitor the situation, obtain as much information about the status as possible, and prevent undesirable developments as soon as possible.” And three UN specialized agencies International Telecommunication Union (ITU), World Meteorological Organization (WMO), and Intergovernmental Oceanographic Commission (IOC) of UNESCO were jointly sponsoring an international Joint Task Force (JTF) to integrate scientific sensors into trans-ocean submarine cables to measure key ocean seafloor observables [7]. In 2016, the concept of Scientific Monitoring and Reliable Telecommunications (SMART) cable system has been proposed as a means to provide early warning of earthquake and tsunami events and expand our understanding of the ocean’s role in Earth’s climate [8,9]. Therefore, monitoring the operating status of underwater cables and timely predicting potential threats is of great significance.
Traditional diagnostic methods of submarine cables include optical time domain reflectometry (OTDR), coherent OTDR (COTDR), Raman OTDR (ROTDR), and Brillouin OTDR (BOTDR) [10]. OTDR is the first generation of optical fiber distributed sensor, where Rayleigh scattering is used to measure the attenuation profiles of optical fibers. Since OTDR cannot distinguish the amplified spontaneous emission (ASE) noise generated by repeaters and the weak Rayleigh backscattered signal (RBS), it is usually used to monitor cables without repeaters. Unlike OTDR, COTDR converts RBS into electrical signals with one specific intermediate frequency by using heterodyne detection. Most of the ASE noise can be inhibited by a band-pass filter, enabling it to cover ultra-long distance cables with cascaded repeaters [11]. ROTDR is based on the Raman scattering phenomenon, which is only capable of measuring temperature rather than strain [12]. While for BOTDR or Brillouin Optical Time domain Analysis (BOTDA), cross-sensitivity of Brillouin frequency shift enables the sensing system to measure both temperature and strain, but difficult to distinguish between them [10,13]. All above techniques can only make alarm after the underwater cable faults happened. Recent advances in distributed vibration sensors make the phase-sensitive optical time-domain reflectometry (Φ-OTDR) an ideal tool for condition monitoring and fault warning of underwater cables for its high sensitivity and low noise equivalent power (NEP) for weak vibration signal detection [11]. Researcher have carried out validation experiments for dynamic monitoring such as anchor dropping, anchor dragging and impacting based on the amplitude-discrimination Φ-OTDR system [10,14–17]. However, the external vibration in these experiments was direct impact on the cable, and the perception of the surrounding environment of the cable is rarely reported. Different from the amplitude-discrimination structure, the phase-discrimination structure can significantly improve the signal reconstruction quality of Φ-OTDR system and provide full vector information of vibration field [18]. Reference [19] realized motion trajectory tracking of underwater acoustical signal with high accuracy based on the phase-discrimination Φ-OTDR system, but the results depended on additional customized cables, which obviously could not meet the “plug and play” monitoring requirements for in-service underwater optical cables. Moreover, the existing underwater optical cable is affected by armor protection and surface deposits, and its internal optical fiber is not well coupled to the external vibration, which makes the requirement on the sensitivity of the Φ-OTDR system to be higher. When using traditional heterodyne coherent detection structure to realize phase identification Φ-OTDR system, the beat signal is usually mixed with the local electric signal of the same frequency to achieve demodulation. However, the frequency drift of the modulator and the instability of the local electric signal will cause the clocks to be out of sync between the electrical signals, resulting in residual frequency, thereby reducing the demodulation accuracy and signal-to-noise ratio. Therefore, the residual frequency is not negligible and must be suppressed [18,20]. Otherwise, the target signal may fall into the same frequency band which would be difficult for subsequent pattern recognition.
In this paper, we present a novel phase-discrimination Φ-OTDR system based on phase-locked (PL) structure. This system provides clock homologous carrier signal, modulation signal and data acquisition card (DAQ) trigger signal, which can provide higher accuracy for vibration signal reconstruction. Field tests of water impact, anchor damage towing and tide diagnosing were carried out in a natural freshwater lake as well as the East China Sea to simulate and verify the direct effect of external vibration on the existing underwater cable. Personnel movement trajectory monitoring and ship flow monitoring by a buried cable along the floodplain of the Yangtze River were carried out to verify the influence of indirectly applied vibration on the cable.
2. Principle of the PL Φ-OTDR system
2.1 Vibration signal reconstruction
In a phase-discrimination Φ-OTDR system, the photocurrent ihet of a photodetector after band-pass filtering can be expressed as [21]
Figure 1 shows how an external vibration that induces extra stress on the fiber and results in a change in optical path length (OPL), where two segments of fiber A and B with a length of L are selected as the reference regions [22]. Regardless of the noise, the backscattered electric field from A and B at photodetector is given by
From Eq. (3), the axial strain of the cable can be calculated as follows
2.2 PL Φ-OTDR system
In Φ-OTDR system, acoustic optical modulator (AOM) is a common modulator to shape CW light into optical pulses, which consists of driving source and acousto-optic crystal, as shown in Fig. 2. A radio frequency (RF) carrier signal launched by radio frequency (RF) source inside the AOM driver is multiplied with external modulation pulse signal through mixer to output the electrical acousto-optic modulated driving signal, which is the chopper for sinusoidal signal. By applying a periodical on/off external modulation pulse signal to the AOM driving source, an amplitude modulated (AM) RF signal with pulse envelope would be generated and then amplified by an RF driver to increase its power level to drive the AOM crystal. However, since the carrier signal, external modulation pulse signal and data acquisition (DAQ) trigger signal come from independently oscillation source, each optical pulse has a random initial phase bias so that every IF signal trace also has the different initial phase bias that varies over time. The initial phase bias change continuously so that the correlation between several consecutive intermediate frequency (IF) traces obtained by the photodetector decrease with time [24]. The bias caused by the unlocked phase will bring additional residual frequencies in the subsequent IQ demodulation, which will decrease the demodulation accuracy and NEP level of Φ-OTDR [18,20].
Figure 3 shows the schematic of our newly designed PL Φ-OTDR system [25]. A narrow linewidth laser of 1550.12nm and 3.7 kHz linewidth was used as the light source. A continuous-wave (CW) light of the laser was split into two parts by a 90:10 optical coupler (OC1), 90% as the probe light and the other 10% as an OLO light. The probe light was shaped by an AOM crystal into a pulse beam with a frequency shift fm of 200MHz, and amplified by an Erbium-doped fiber amplifier (EDFA). The amplified probe optical pulse injected into the fiber under test (FUT) through a circulator (CIR). The backscattered light of the FUT was output from the port3 of the CIR, then combined with OLO light through a 50:50 optical coupler (OC2) for beating. A balanced photoelectric detector (BPD) with a 350MHz bandwidth and 30×103 V/A trans-impedance gain converted the beat frequency optical signal into an electrical signal while eliminating DC and common components. The output of the BPD passed through a filter- amplification-filter (FAF) module with a bandwidth of 195MHz∼205MHz for signal conditioning. Then the signal was feed to a DAQ. To keep the initial phase bias of the optical pulse same, the new PL module was designed. As shown in Fig. 3, a 10MHz-sinusoidal reference clock and a modulated pulse signal are generated internally by DAQ and output synchronously, ensuring that the clocks of the phase-discrimination Φ-OTDR system are completely synchronized. The frequency of the reference clock will be doubled to 200MHz to form a radio frequency carrier signal, which is multiplied with the modulation pulse signal by a mixer. An AM RF signal will be generated and amplified to drive the AOM crystal. Then we instrumented the newly designed PL Φ-OTDR system by cooperating with Nanjing Fiber Photonics technology CO., LTD [26]. The appearance of the instrument was shown in Fig. 4, and whose major performance indexes were listed in Table 1.
Table 1. Major Performance Indexes of the New Φ-OTDR Instrument
In order to compare the proposed PL Φ-OTDR and the traditional Φ-OTDR, in addition to the new instrument with above configuration, the synchronization signal of another instrument of the same type was cut off to degenerate it into a traditional AOM driving mode. Noise performance experiments by using the two instruments were carried out respectively. The setting of experimental parameters was determined by the step interval that the system hardware can set. In addition, since we could not predict the frequency and amplitude of the vibration events to be measured, we first chose to use a relatively high repetition frequency parameter to cover more detectable frequencies, and also use the oversampling setting to enable it to detect large amplitude events. With a pulse width of 96ns and a repetition frequency of 19840Hz, the IF signals of conventional and PL Φ-OTDR were continuously collected for 10s in a quiet environment and the phases of each position along the fiber were extracted respectively. The inherent sampling rate of the DAQ was 250MHz. Sampling reduction at a ratio of 1:5 was carried out after the signal demodulation was completed. The sampling rate was reduced to 50MHz after data extraction. The original spatial resolution corresponding to the pulse width of 96ns was 9.6m, but for convenience in data processing, we adopted a phase discrimination width of 10m, so the final spatial resolution was 10m [27]. In the subsequent experiment in this paper, the same pulse width and phase-discrimination width were used, which will not be repeated in the following. The left side of Fig. 4 shows the root power spectral density of the two groups of data and their NEP are calculated respectively. It can be seen that, the noise data collected by the traditional AOM driver module brings a high-frequency residual signal of 4.7 kHz, while the PL noise data is very clean and NEP is reduced by 4.2dB. In addition, there is a more obvious improvement in the low-frequency part below 1 kHz, which proves the superior PL performance of our newly designed system.
3. On-line status monitoring of the existing underwater cable
In order to verify the ability of the new instrument to realize the on-line status monitoring and surrounding environment perception of existing underwater cables, we studied external vibration acting on the cable directly and indirectly with the new instrument in different scenarios. Field tests were carried out in Freshwater Lake, East China Sea and Yangtze River through existing communication cables laid under the lakebed, seabed, and riverbed respectively. Experiments of external vibration acting on the cable directly included noise acquisition from the lakebed, water impact, anchor damage towing and tide diagnosing. Experiments of indirect external vibration on the cable were personnel movement trajectory and ship flow monitoring.
3.1 Water impact and anchor damage towing monitoring in a natural freshwater lake
The natural freshwater lake test site is shown in Fig. 5. We placed the Φ-OTDR instrument on the shore, and used BeiDou Satellite (BDS) navigation system with telescope rangefinder to calibrate 10 points, and placed a buoy at each point before laying the cable. Then sunk a 400m common type of GYXTS single-mode dual-core armored underwater communication cable as straight as possible into the water. The average water depth along the cable was 8m, and the deepest depth was about 12m. Due to the influence of water flow, the cable finally settled naturally on the lakebed in an S-shape.
To verify the ability of the newly designed instrument to remove the residual frequency due to the unlocked phase in the field condition, we first continuously collected two sets of the background noise data from the lakebed for 80s by using the new and another traditional phase-unlocked (PUL) Φ-OTDR instrument, respectively. The pulse repetition frequency was 19840Hz. To observe the difference in the spectrum characteristics of the two sets of data, we selected a position of 200m, and performed a short time Fourier transform (STFT) on their output signals. Figure 6 (a) and (b) show the power spectrum analysis results of the two sets of data, respectively. The color maps the power spectral density (PSD) of the output signal. It can be seen that the noise data collected by the new PL Φ-OTDR instrument is very clean, while the traditional instrument shows a high-frequency residual signal of about 4.2 kHz, which proves that our new instrument still has superior phased-locked performance in the field condition. And for the PUL Φ-OTDR instrument, the deviation of the initial phase is random due to the influence of temperature, so the residual frequency brought by the PUL structure is deviated from the results measured in the laboratory above.
Fig. 6. Power spectrum analysis of noise data. (a) Result of the PL Φ-OTDR instrument; (b) Result of the traditional Φ-OTDR instrument.
3.1.1 Water impact monitoring
Drove the experimental ship above the cable and stayed about 295m from the start point of the cable. We lowered the water pump at a depth of 7m, then fixed it on the ship and turned it on for water spraying. And then we drove the ship back and forth perpendicular to the cable. The monitoring results are shown in Fig. 7. When the vibration generated by the water impact propagates to the cable, the waterfall map of time-distance distribution presents obvious bright stripes, as shown in the yellow rectangles in Fig. 7 (a). The color maps to the power of the signal. The center position of the vibration is 295m, and its coverage range is approximately ±40m. The time-domain waveform of the center position is shown in Fig. 7 (b). It can be observed that there are eleven consecutive impact signals in 60s. Corresponding to a spatial resolution of 10m, the dynamic axial strain of the cable caused by the continuous water impact reaches ±2.5µε, which exceeds the lower limit of the static strain that can be measured by traditional BOTDR or BOTDA instrument [28]. Furthermore, compared with the vibration events directly applied to the bare fiber, the measured dynamic strain is relatively small due to the protection of the fiber inside the cable. The detailed view of the sixth oscillation waveform is enlarged shown in the green rectangles of Fig7 (d). It can be seen that after a single impact, the damping oscillation lasts about 1.6 seconds. Figure 7 (c) gives the power spectrum of Fig. 7 (b), and the red rectangles in Fig7 (d) shows the detail view of frequency spectrum of the sixth impact signal. The results show that each impact signal had a wide-band spectrum characteristic at first, and then the spectrum range narrows rapidly into several low-frequency characteristics, which is consistent with the frequency features of the water impact signal. With an interval of 0.027s, the total energy distribution of all frequencies in time window is shown in Fig. 8 (a). It can be seen that the impact signal starts at 30.85s, and the energy rapidly attenuates to less than 10% within 0.23s then forms several oscillations until dissipation. Figure 8 (b) illustrates the energy distribution of the high-frequency components larger than 1 kHz and 2.5 kHz, which is consistent with the analysis results of frequency characteristics in Fig. 7 (c). The frequency higher than 2.5 kHz only maintains for 0.07s, then the energy rapidly drops to below ten percent.
3.1.2 Anchor damage towing
Put the anchor into the water at about 240m from the start point of the cable and let it sink to the bottom of the lake, then towed it perpendicularly to the cable. Mechanical vibration would be produced when the cable is towed by the anchor as shown in Fig. 9. It can be identified from the waterfall map of time-distance distribution in Fig. 9 (a) that towing occurs near 240m, and the entire towing process lasts for more than one minute. With the increasing of the dragging force, the range of the stressed cable gradually increases, up to ±100m. During that time, there are several drag and slacken processes. For a single drag process, a symmetrical V-shaped appears on the space-time two-dimensional map from the dragging point to both sides, which is consistent with the process of gradually expanding the forced area when a certain point of the cable is dragged. The time-domain waveform measured at 240m is shown in Fig. 9 (b), and the waveform of the single dragging from 68s to 73s is enlarged shown in the green rectangles of Fig9 (d). Compared with water flow impact, the waveform generated by the anchor dragging has no damping oscillation characteristic, and its dynamic strain is one order of magnitude higher, reaching ±29µε at the same spatial resolution of 10m. Figure 9 (c) gives the corresponding power spectrum and the red rectangle in Fig7 (d) illustrates the detailed view of the single dragging. Compared with the experiment result of water impact, the power spectrum intensity of anchor damage signal is much larger. And each dragging signal also has a wide-band frequency characteristic at first, then the frequency range narrows into the low-frequency characteristic. The difference is that the wide-band spectrum generated by anchor dragging covers almost the entire spectrum for about 1.19s, then the spectrum range narrows gradually into the low-frequency components. With an interval of 0.027s, the total energy distribution of all frequencies in time window is shown in Fig. 10 (a). It can be seen that the single dragging starts at 68.62s. Unlike the impact signal, since dragging is a process of “pulling” to “loosening”, the energy generated by the single dragging rapidly drops below 10% after 0.34s and then rises immediately. After 0.46s, it attenuates below 10% again and then remains at 10% for a period of time before dissipating. It can be seen that although the naturally sunk cable on the lakebed was in a free state and affected by the water flow, it is still possible to distinguish the water impact from the anchoring damage events directly acting on the cable by using our newly Φ-OTDR instrument.
3.2 Tide monitoring in the East China Sea
Affected by the complex underwater environment and the movement of ocean currents, part of the cable segments originally buried under the seabed may be exposed to seawater. The cable will be repeatedly washed by tides or ocean currents and then face the danger of being worn or broken. If we can know the vibration intensity database caused by the direct impact of seawater on the cable in advance, and form a long-term observation hydrological data, it may provide reference value for prevention of earthquake-like disasters in the future. For this research purpose, we laid a 2100m cable of the GYXTS type at a port in the East China Sea in June 2021. The test site was illustrated in Fig. 11. The top view of the test site shows that it was surrounded by islands, so the cable was influenced by multiple ocean currents among the islands. In the bay area, the average depth of water was about 60m, and there was a trench with a width of hundreds of meters under the barge fairway of the port. Two anchor points were selected on the island, then we connected two fixed ropes around the two points to form a triangular relationship with the experimental ship to prevent the ship drifting with ocean current. Additionally, in order to reduce the stress on the connection between the instrument and the cable and to avoid the cable break caused by seawater washing, we kept pulling on the cable during the experiment. After 22 minutes, the weather changed from cloudy to rainy, causing the sea level to rise and water movement to increase. As we felt the fluttering of the cable, we retracted the cable. In the experiment, the pulse repetition frequency was set to 4200Hz. For the large amount of the collected data, we performed data extraction by a ratio of 1:5 in distance and 1:10 in time respectively after signal demodulation and smoothing.
Fig. 7. Results of water impact. (a) Waterfall map of time-distance distribution; (b) Time-domain waveform at 295m; (c) Short-time Fourier transform spectrum; (d) Subplots of the strain and frequency over time.
Fig. 8. Normalized distribution of the sixth oscillation signal. (a) All frequencies; (b) High-frequency components.
Fig. 9. Results of anchor damage towing. (a) Waterfall map of time-distance distribution; (b) Time-domain waveform at 240m; (c) Short-time Fourier transform spectrum; (d) Subplots of the strain and frequency over time.
Fig. 10. Normalized distribution of the single dragging. (a) All frequencies; (b) High-frequency components.
During the high tide, we observed tidal-induced vibrations of ocean currents on the cable. A total of 44-minute observation was obtained, as shown in Fig. 12. The color maps the phase difference intensity of the demodulated signal. Due to the large amount of data, the color is visually superimposed. The 0.1-minute local enlarged figure on the right illustrated more clearly the regular light and dark changes of the vibration signals. In the red rectangle box, we can obviously see that after 22 minutes, as the cable was tightened, the dragged area gradually expands, which corresponds to the process of the cable from slack to tightening. And there is a large range of high intensity vibration in the section of 1000∼2000m for a long time, which can be speculated to be caused by the impact of undercurrent during the high tide. In the area of 1450∼1850m, it can be seen that the disturbed range caused by the high tide gradually expands with time, and the moment when the tide came at each positon was obtained, as shown in the red block arrows in Fig. 12.
To grasp the characteristics of these disturbances, we selected a typical signal at position 1610m for further analysis in the time and frequency domain, as shown in Fig. 13. Figure 13 (a) is the time-domain waveform, in which the red rectangles is the detailed view of the waveform from 600 to 610s. The whole signal presents a perfect sinusoidal shape, which is the periodic vibration of the cable caused by the high tide [29,30]. The strain intensity on the cable varied with time, and the maximum dynamic strain reaches ±1.8µε, which is of the same order of magnitude as the strain generated in the above experiment of water impact described in 3.1.1. Figure 13 (b) ∼ (d) illustrate the power spectrum of the time-domain waveform in different frequency ranges. From the spectrum in Fig. 13 (c), we can see that the vibration frequency generated by the tide is composed of the dominant low frequency and its high order harmonics. We speculate that the harmonics is caused by the topography of the test site surrounded by the multiple islands. In Fig. 13 (d), we note the dominant frequency changes from 2.3Hz to 2.7Hz with time, and we guess that such time-varying frequency characteristic is related to complex underwater hydrodynamics [30].
Fig. 13. Time-domain waveform and short-time Fourier transform spectrum at 1042 m. (a) time-domain waveform; (b) ∼ (d) Power spectrum of the time-domain waveform in different frequency ranges.
4. Surrounding environment perception of the existing underwater cable
In order to verify the ability of the newly designed Φ-OTDR system to perceive the surrounding environment, we used the instrument to carry out personnel movement trajectory monitoring and ship flow monitoring by a buried cable along the floodplain of the Yangtze River. The setup of the field tests is shown in Fig. 14. The GYXTS cable was artificially buried along the floodplain with a depth of about 10 centimeters during the drought period in February, 2021. The Φ-OTDR instrument was placed on the landing stage. The cable was laid along the direction of A-B-C-D-E-F-G, where the section A-B was naturally settled in water, and point D was the turn-back position. Points B, F and D were reinforced with wet mud blocks to increase the coupling of the cable and the floodplain geology. Additionally, the cable was integrally parallel to the waterway of the ship, and its start point was close to the waterway while the middle part was far away. The outmost waterway was about 300m perpendicular to the cable. To verify the Φ-OTDR instrument monitoring performance, we set up a closed circuit television (CCTV) system at the Landing Stage to observe the ship flow in the waterway, and its monitoring perspective is shown in the thumbnail. In the experiment, the pulse repetition frequency was set to 1000Hz.
Fig. 14. Top view of the artificial buried cable along the floodplain of the Yangtze River as well as the East China Sea. Thumbnail: Monitoring perspective of the CCTV system.
4.1 Personnel movement trajectory monitoring
In order to test the ability of the anti-amphibious landing monitoring, we carried out monitoring personnel movement trajectory in Feb 10, 2021. Two male adults Tom and Jerry ran back and forth along the cable between point C and E with a total distance of about 250m, and their movement trajectory is shown by the yellow dotted arrow in Fig. 15. The inset enlarged view displays the muddy soil of the floodplain around the cable. At the beginning, the two adults were separated by 150 meters. Then they started running at the same time, and Jerry was running from point D to C while Tom was running from point E to D, then they met several times. During the process, they changed speed and stopped to rest several times. About 1100s later, the two adults stopped moving and walked slowly along the cable. At the same time, they dug soil to compact the cable.
Figure 16 shows the monitoring result of personnel movement trajectory. Figure 16 (a) and (b) are the separate detail displays of the rectangular boxes of Fig. 16 (c), the waterfall map of time-distance distribution. It can be seen clearly that Tom and Jerry meet at moment t1, t2, t3, t4, and t5 respectively. About 630s later, at the moment t3, Tom catches up with Jerry, and they stop running at about 1040s and 1104s respectively. We can observed from Fig. 16 (a) that Tom performs an intermittent seem “stop-run” movement in the first half of the route, and then speed up in the second half. Figure 16 (b) illustrates the walking steps and the “dig-compact” action. The power of the bright block generates by the latter is much higher. Additionally, different from the continuous bright stripes presented by running in Fig. 16 (a), walking displays intermittent bright spots.
Fig. 16. Monitoring result of personnel movement trajectory. (a) ∼ (b) Separate detail displays; (c) Waterfall map of time-distance distribution.
We intercepts the data of 0∼1100s before they stop running for further analysis. Firstly, we find out the power peak along the time axis, and replace the peak value with the current position value. Then divide the data of Tom and Jerry to obtain their respective movement trajectory traces, as shown in Fig. 17 (a). By mirroring the reverse motion curves after the turning point, the distance-time (DT) traces of their movement can be obtained, as illustrated in Fig. 17 (b). Then, a universal third-order polynomial with the best fitting effect is used to fit the DT traces, and the fitting curve of Jerry is vertically displaced for clarity, as shown in Fig. 17 (c). From the slope, we can see that the overall running speed of adults Jerry is slower, and their total moving distance is 1147m and 746m respectively. A difference derivation operation of the DT fitting traces is further performed to estimate instantaneous velocity. Figure 17 (d) illustrates their instantaneous velocity based a difference interval of one second. It could be inferred that Jerry ran faster than Tom in 151s∼360s, and much slower than Tom in the rest of the time, which is consistent with the analysis results in Fig. 16. Moreover, the overall speed of them is slower than that normal walking speed of one person due to the muddy condition of the floodplain around the cable.
Fig. 17. The trajectory and instantaneous velocity analysis results of adults Tom and Jerry. (a) Original movement trajectory traces; (b) DT traces of their movement; (c) Fitting curves; (d) Instantaneous velocity.
4.2 Ship flow monitoring
Figure 18 shows the monitoring results of multiple ships passing through the waterway during 15:37∼15:44 pm on June, 2021. The color of the left figure maps the PSD of the demodulation signal. During this time period, the river is at a flood tide stage, and the cable has been covered by several meters of mud and water. In the yellow rectangles, it can be seen that two ships was detected while they passing through the waterway at about 200∼400m and 600m of the cable, with a time difference of about 40 seconds. The cable segment at 200∼400m was close to the waterway, so the detected signal was more obvious. Although the 400∼600m section was relatively farther, the cable was reinforced with wet mud blocks at about 600m, which increased the coupling of the cable and the floodplain geology, thus there was a strong response. On the right is the picture of the two ships captured by the remote surveillance device. At about 160m and its symmetrical position at 640m, there are two continuous red bright zones, this is because the cable at this position is horizontally suspended in the water, which is affected by water flow and continuously produces vibration to the cable.
Fig. 18. Results of multiple ships passing through the waterway during 15:37-15:44 pm on June 20, 2021. Left: waterfall map of the time-distance distribution; Right: picture of the two ships captured by the remote surveillance device.
By further analyzing the time-frequency domain characteristic with and without ships passing the waterway in Fig. 18, we find the typical vibration characteristics of ships in the waterway, as shown in Fig. 19. Figures 19 (a) and (b) are the waterfall map of time-distance distribution in 80∼165s and 270∼355s of Fig. 16, respectively. It is obviously that the noise base without ships is relatively clean. Figure 19 (c) and (d) show the time-domain waveforms with and without ships at 380m, and the peak-to-peak value of the dynamic axial strain of the cable with ships is about ±0.5µε. Figure 19 (e) and (f) give the corresponding power spectrums. It can be seen that whole duration of time domain signal is about 15s, and the frequency characteristic component of the ship event is basically between 2Hz and 3Hz. The frequency characteristic is caused by the ship’s vibration acoustic signal acting on the cable propagating through the water, rather than the wave directly disturbing the cable. According to the investigation, some large cargo or cruise ships generally use two-stroke low-speed engines, which usually directly drive the propellers. During sailing, their speed is around 80∼170 revolutions per minute, and noise signals with a frequency of 2Hz to 3Hz is generated, which is consistent with the experimental analysis results. On the contrary, no characteristic frequency can be observed in the corresponding spectrogram when no ship pass by in Fig. 19 (f).
Fig. 19. Analysis results of acoustic characteristics with and without ships passing the waterway. (a) and (b) Waterfall map of time-distance distribution in 80∼165s and 270∼355s of Fig. 6, respectively; (c) and (d) Time-domain waveforms at 380 m; (e) and (f) Corresponding power spectrums.
Time-delay estimation of vibration signal in 200∼600m is further performed. Firstly, generalized cross-correlation (GCC) [31] is used to get the relative delay of the phase signals at adjacent positions, then by integrating the relative delay, we obtain the time delay of received signal at each position, as shown in Fig. 20. Figure 20 (a) is the time-distance distribution map of phase delay, in which the color maps the time delay value, and the warmer the color, the earlier the signal propagated to that location. Figure 20 (b) gives the phase delay trace of the strongest vibration signal. The result presents that the signal firstly reached the turn-back point D of the cable at about 400m, and then gradually propagates to the near end, which because of the shielding of the Landing Stage. By calculating the slope of the fitting curve, it is found that the average velocity of the vibration signal is about 351m/s, which far exceeds the wave velocity caused by the approaching of the ship. It proves that the acoustic signal acts on the cable indirectly through its surrounding environment, rather than being caused by the direct beat of the waves. Moreover, 351m/s is between the propagation speed of acoustic signal in water and soil [32], which shows that the transmission of the signal is a complex process, not in a single medium. We can speculate that the signal propagated first in the water and then through the soil to the cable.
Fig. 20. Results of time delay analysis of vibration signal in 200∼600 m. (a) Time-distance distribution map of phase delay; (b) Time delay trace of the strongest vibration signal.
For ship flow monitoring, the rates of missing alarm and false alarm are both associated with the intensity threshold in Fig. 18. In order to achieve a balance, we recorded the monitoring results of ship flow with different thresholds for two weeks, a total of 7673 samples, and used the CCTV to check the actual number of ships passing the waterway. Suppose that the number of monitored ships and actual ships could be denoted by M and N, and the true alarm, the missing alarm, and the false alarm can be represented by t, m and f respectively. Then they will meet the following conditions.
The two-week statistics are shown in Table 2, and Fig. 21 gives the variation of the rates of missing and false alarm with different thresholds. The yellow solid points are the total of the two alarm rates, and the solid lines are the corresponding fitting curves by using the 5-order polynomial fitting method. It can be seen that with the increase of the intensity threshold, the false alarm rate decreases and the missing alarm rate increases. When the intensity threshold is 536rad, they are equal to 8.44%. And the total alarm rate reaches the minimum value of 17.9% at the threshold of 550rad. Therefore, we can consider 550rad as the optimal threshold.
Table 2. Results of Ship Flow Monitoring for Two Weeks
5. Discussion
Compared with the limitations of BOTDR or BOTDA in the measurement of dynamic strain events, the PL Φ-OTDR system proposed in this paper not only shows excellent performance in the monitoring of underwater optical cables, but also has early warning capabilities. An essential objective of the JTF SMART cable initiative is to develop a general, repeatable process to attach sensors to each new submarine telecommunications cable as a matter of course. SMART cables would have sensors at every repeater to provide ubiquitous coverage [9]. Φ-OTDR technology, based on the existing underwater optical cables, has a large number of sensing units to realize fully distributed monitoring, which not only provides a reliable technical path for the future SMART concept, but also greatly reduce its deployment costs.
As a vibration data analysis device, in order to verify the field monitoring performance of the proposed system, this paper carried out a series of field demonstration experiments in various scenarios such as lake, river and sea. The time-frequency characteristics of typical events were recorded and compared. Results show that it not only can monitor the status of the cable itself, but also sense the surrounding environment. On one hand, the direct and indirect effects of various vibration events on the underwater cable show different target characteristics, which results in different requirements for subsequent signal processing methods. On the other hand, parameter settings in application scenarios depend on the accumulation and analysis of long-term large-sample data in the field test, such as the optimal decision threshold. At the same time, it should be noted that in the experiment of ship flow monitoring, the false alarm and missing alarm rate based on the intensity threshold is still high, and similar problems should exist in other potential applications. This requires the combination of the technology in this paper and the specific application must closely connect and utilize the background knowledge of the industry, and testing activities should be carried out in the actual environment earlier. For the identification of target time, it is difficult to establish a clear and universal mathematical model in specific applications due to the differences in environment, background noise, vibration-cable coupling conditions. The technical route combining sample accumulation, feature database comparison, and artificial intelligence algorithms may be the only way for such applications to become be practical, which could be one of our future work.
In addition, the new Φ-OTDR system with PL structure proposed in the paper achieved the suppression of the residual frequencies brought by PUL from experimental results, but there was no relevant research on its theoretical derivation and the tolerance of demodulation algorithm to PUL structure, which will be the another focus of our next works.
6. Conclusion
This paper proposed a novel PL structure for Φ-OTDR sensing system, which significantly improved the detection sensitivity. In view of the different effects of cable states with different buried on monitoring results, we carried out field tests in various scenarios. Results showed that there had completely different characteristics of different vibration events in the waterfall map of time-distance distribution and the time-domain waveform at the vibration location. The events could be identified according to the differences. It was proved that the proposed phased-locked Φ-OTDR system can monitor the real time status and sense the surrounding environment of existing underwater communication cable, which provides an effective technical means for the status monitoring and the fault warning of the cable.
Funding
National Natural Science Foundation of China (61975076, 62175100, U2001601); The Key Technology R&D Program of Inner Mongolia Autonomous Region (2019GG374); Fundamental Research Funds for the Central Universities (0213-14380202); Science, Technology and Innovation Commission of Shenzhen Municipality (YFJGJS1.0); (NUPTSF) Nanjing University of Posts and Telecommunications (NY220054).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. Z. Zhao, “Past, present and future of optical fiber communication,” Guangxue Xuebao 31(9), 0900109 (2011). [CrossRef]
2. X. Tang, “The total investment in the maritime cable sector has exceeded $\$ 50$ billion,” (2019), http://www.c11-4.com.cn/news/22/c20324.html.
3. Q. Winston, “Next generation submarine network innovative repeater technology,” (2020), www.submarinenetworks.com/en/insights.
4. H. Chen, “Pingtan submarine cable was cut resulting power cut,” (2007), https://news.sina.com.cn/c/2007-09-25/000412626405s.shtml.
5. X. Chen, N. Zou, L. Liang, R. He, J. Liu, Y. Zheng, F. Wang, X. Zhang, and Y. Zhang, “A submarine cable monitoring system based on enhanced COTDR with simultaneous loss measurement and vibration monitoring ability,” Opt. Express 29(9), 13115–13128 (2021). [CrossRef]
6. M. Ren, “A powerful 7.5-magnitude earthquake has struck Alaska, triggering a locally destructive tsunami,” (2013), http://scitech.people.com.cn/n/2013/0106/c1007-20103790.html.
7. P. Phibbs, “Using submarine cables for climate monitoring and disaster warning: Engineering feasibility study,” Engineering feasibility study (Itu News, 2012).
8. C. Barnes, D. Meldrum, and H. Ota, “Emerging subsea networks: new market opportunities for, and societal contributions from, SMART cable systems,” in Proceedings of SubOptic 2016, (2016).
9. S. Lentz and B. Howe, “Scientific monitoring and reliable telecommunications (smart) cable systems: Integrati-on of sensors into telecommunications repeaters,” in Proceedings of OCEANS - MTS/IEEE Kobe Techno-Oceans, (2018).
10. A. Lv and J. Li, “On-line monitoring system of 35 kV 3-core submarine power cable based on Φ-OTDR,” Sens. Actuators, A 273, 134–139 (2018). [CrossRef]
11. X. Zhang, X. Chen, L. Liang, S. Zhao, R. He, S. Tong, F. Wang, N. Zou, and Y. Zhang, “Enhanced C-OTDR based online monitoring scheme for long-distance submarine cables,” Guangxue Xuebao 41(13), 1306001 (2021). [CrossRef]
12. T. Hara, K. Terashima, H. Takashima, H. Suzuki, Y. Nakura, Y. Makino, S. Yamamoto, and T. Nakamura, “Development of long range optical fiber sensors for composite submarine power cable maintenance,” IEEE Trans. Power Delivery 14(1), 23–30 (1999). [CrossRef]
13. A. Lv, “Research on condition monitoring of optical fiber composite submarine power cable based on distributed strain and temperature sensing technology using optical fiber,” Ph.D. dissertation (North China Electric Power University, 2015).
14. A. Masoudi, J. Pilgrim, T. Newson, and G. Brambilla, “Subsea cable condition monitoring with distributed optical fiber vibration sensor,” J. Lightwave Technol. 37(4), 1352–1358 (2019). [CrossRef]
15. Y. Shan, “The key technology research of distributed optical fiber vibration sensor based on Φ-OTDR,” Ph.D. dissertation (Nanjing University, 2019).
16. E. Williams, M. Fernández-Ruiz, R. Magalhaes, R. Vanthillo, Z. Zhan, M. González-Herráez, and H. Martins, “Distributed sensing of microseisms and teleseisms with submarine dark fibers,” Nat. Commun. 10(1), 5778–577811 (2019). [CrossRef]
17. N. Lindsey, T. Dawe, and J. Ajo-Franklin, “Illuminating seafloor faults and ocean dynamics with dark fiber distributed acoustic sensing,” Science 366(6469), 1103–1107 (2019). [CrossRef]
18. W. Xu, M. Liu, Y. Miao, T. Chang, and B. Yang, “Analysis and improvement for digital quadrature demodulation algorithmon phase-sensitive optical time-domain reflectometric system,” Chinese Journal of Lasers 44(12), 275–281 (2017).
19. B. Lu, B. Wu, J. Gu, J. Yang, and H. Cai, “Distributed optical fiber hydrophone based on Φ-OTDR and its field test,” Opt. Express 29(3), 3147–3162 (2021). [CrossRef]
20. X. Zhang, Y. Zheng, C. Zhang, Q. Dong, and F. Xiong, “A fading tolerant phase-sensitive optical time domain reflectometry based on phasing-locking structure,” Electronics 10(5), 535 (2021). [CrossRef]
21. M. Zabihi, Y. Chen, T. Zhou, J. Liu, and M. Chen, “Continuous fading suppression method for Φ-OTDR systems using optimum tracking over multiple probe frequencies,” J. Lightwave Technol. 37(14), 3602–3610 (2019). [CrossRef]
22. K. Shimizu and T. Horiguchi, “Characteristics and reduction of coherent fading noise in rayleigh backscattering measurement for optical fibers and components,” J. Lightwave Technol. 10(7), 982–987 (1992). [CrossRef]
23. Z. Wang, L. Zhang, S. Wang, N. Xue, F. Peng, M. Fan, W. Sun, X. Qian, J. Rao, and Y. Rao, “Coherent Φ-OTDR based on I/Q demodulation and homodyne detection,” Opt. Express 24(2), 853–858 (2016). [CrossRef]
24. G. Stegeman, “9. Acousto-Optic Phenomena,” Methods Exp. Phys. 19, 455–493 (1981). [CrossRef]
25. Y. Zhang, “A phase-locked Φ-OTDR system with high operation efficiency,” China patent, 202110652272.X (11 June 2021).
26. Fib-tech, “Product display,” (2021), http://www.fib-tech.com.
27. X. Zhang, Fully distributed optical fiber sensing technology (Science Press, 2012), Chap.1.
28. M. Xia, X. Tang, Y. Wang, S. Zhang, H. Jiang, and Y. Dong, “Comparative analysis of strain monitoring of the installed optical fiber composite overhead ground wire using BOTDA/R,” Acta Opt. Sin 40(15), 36–42 (2020).
29. J. Zhang, “Research on wave control mechanism in deep water channel,” Master dissertation (Guilin University of Electronic Technology, 2021).
30. V. Gordienko, V. Nekrasov, and N. Krasnopistcev, “The peculiarities of the effects of underwater currents on low-frequency hydroacoustic stationary vertically distributed receiving systems,” Moscow Univ. Phys. Bull. (Engl. Transl.) 69(4), 318–329 (2014). [CrossRef]
31. J. Liang, Z. Wang, B. Lu, X. Wang, L. Li, Q. Ye, R. Qu, and H. Cai, “Distributed acoustic sensing for 2D and 3D acoustic source localization,” Opt. Lett. 44(7), 1690–1693 (2019). [CrossRef]
32. Z. Ding, N. Zou, C. Zhang, Y. Chen, S. Tong, F. Wang, F. Xiong, Y. Zhang, and X. Zhang, “Self-optimized vibration localization based on distributed acoustic sensing and existing underground optical cable,” J. Lightwave Technol. 40(3), 844–854 (2022). [CrossRef]
