Open Access
Add to BibSonomy
Abstract
Target detection is significant in many fields, including oceanic security, marine ecology, etc. In this paper, phase sensitive optical time domain reflectometry (Φ-OTDR) is introduced for the non-cooperative ship detection, with large-scale diversity technology and suspended sensitized optical cable. In outfield experiments, the ship’s voiceprint information is obtained in high fidelity, the ship’s power spectrum is analyzed, and the over-top detection is achieved. Moreover, an array orientation method based on adaptive phase difference correction (APDC) is proposed to track the ship, suppressing the delay jitter influence of acoustic transmission underwater. This is the first time that voiceprint information of the non-cooperative ship is high-fidelity acquired and deeply analyzed with Φ-OTDR and suspended sensitized optical cable, which is conducive to the detection and identification of marine targets, and proves the potential of Φ-OTDR in hydroacoustic detection applications.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Target detection is of great significance in the fields of oceanic security, maritime rescue, marine ecological protection, and fishery development [1]. The acquisition of the target's characteristic information is particularly important, including voiceprint information (unique signal characteristics of specific target), orientation information, etc. In recent years, Φ-OTDR has gained wide attention in hydroacoustic field for the advantages of the ability of large-scale networking, flexible array reconfiguration, and easy to automate production with uniform wet-side structure. Moreover, Φ-OTDR transforms optical fiber into transducer array, and quantitatively demodulating the acoustic wave information at each position in the axis of the optical fiber by injecting interrogating laser pulses at the single end of the optical fiber [2], which can provide a good solution for acquiring voiceprint information and tracking the orientation of targets.
At present, Φ-OTDR has been studied in the field of hydroacoustic detection due to its unique advantages. Its detection capability was verified in a laboratory tank for the first time in 2015 [3]. Subsequently, the detection capability was verified in open water in 2017 [4]. The orientation detection of a simulated target with strong sound source level was realized using a suspended horizontal array in the first lake test [5]. In addition, the first sea trial (sea state level 2-3) was reported, the localization of the artificial source target was achieved for distance of 5 km and 10 km, respectively [6]. The above technologies are studied with simulated sound sources, but the detection of real targets is still a challenge, with more complex spectrum features and lower sound source level. In recent years, some studies on the detection of real targets are reported. In 2023, our group realized the over-top detection of a small passenger ship using underwater communication optical cable [7]. In the same year, J. Chen et al. [8] reported the over-top detection and trajectory tracking of targets, with scatter-enhanced sensitized optical cable under the water. However, when the submerged optical cable is used to detect targets, the voiceprint information of targets is distorted and the target type is difficult to obtain, due to the coupling effect of the underwater medium. Therefore, due to the physical characteristics of the real targets and the coupling effect of the underwater medium with the submerged optical cable, it is still a great challenge to detect the high-fidelity characteristic information of real targets.
In this work, non-cooperative ship detection is achieved by using large-scale diversity Φ-OTDR and a suspended sensitized optical cable. A 50-level large-scale diversity Φ-OTDR system with frequency diversity technology [9], spatial diversity technology [10] and time diversity technology [7] is designed to improve the signal-to-noise ratio (SNR), and a sensitized optical cable is designed to improve the acoustic pressure sensitivity. In outfield test, the high-fidelity acquisition and voiceprint feature analysis of the ship are achieved, and the over-top detection of the moving ship is realized. Moreover, APDC is proposed to suppress the array orientation error from delay jitter of acoustic transmission in actual condition. In results, the Φ-OTDR system noise is reduced by 13 dB, the acoustic pressure sensitivity of the sensitized optical cable is about -137.12 dB rad/µPa, the azimuth fluctuation of array orientation is suppressed by nearly half. To our best knowledge, it is the first time that the voiceprint information of the non-cooperative ship is high-fidelity detected and analyzed with Φ-OTDR and the sensitized optical cable, which will promote the development of marine target classification and identification, and will greatly expand the application scenarios of Φ-OTDR.
2. Principle
2.1 Large-scale diversity Φ-OTDR system
In marine hydroacoustic detection, the real target signal is weak and requires high SNR and sensitivity of Φ-OTDR system, but the SNR is limited by the weak Rayleigh Backscattering coefficient, which makes it difficult to satisfy the detection of weak signal in this scenario. Here, a 50-level large-scale diversity Φ-OTDR system is designed to achieve high SNR of the system to detect weak hydroacoustic signals. Generally, larger diversity level brings higher SNR, but limited by various factors in implementation. In this work, the large-scale diversity strategy is deeply studied and the diversity level is carefully analyzed for each diversity technology. For frequency diversity, diversity level is limited by the energy of the sideband from the electro-optical modulator (EOM), a higher order sideband is difficult to achieve, and 5-level diversity is selected. For spatial diversity, diversity level is related to the acoustic frequency. The array element aperture must be less than the half of the acoustic wavelength, due to the array orientation principle. In array orientation, the half of the acoustic wavelength is 7.5 m for 97 Hz, and 5-level spatial diversity (the array element aperture 2.8 m) is chosen with 10 m spatial resolution, 5 m spacing for spatial diversity and 5.35 winding ratio. Finally, the 2-level time diversity is achieved by weighing the data amount against performance. In short, the 50-level large-scale diversity includes 5-level frequency diversity [9] achieved by EOM modulating the narrow linewidth laser light source, 5-level spatial diversity [10], and 2-level time diversity [7] applied in the data processing.
Φ-OTDR is the phase demodulation system based on heterodyne coherent detection which utilizes the spatial phase difference principle, shown in Fig. 1. The narrow linewidth laser light is divided into the probe light and the reference light by the optical fiber coupler (OC). The probe light part is modulated by EOM to ensure that the energy is concentrated in 5 levels of 0, ± 1, ± 2, and the modulating frequency of EOM is 20 MHz. The probe light is chopped by acousto-optic modulator (AOM), with 100 ns pulse width and 160 MHz shifting frequency, and then multi-frequency high coherent light pulses are injected into the sensing fiber. The Rayleigh Backscattering light from the fiber is mixed with the reference light and then detected by a balanced photodetector (BPD). Finally, the electric signal is sampled by a data acquisition (DAQ) unit. Clock synchronization [7] is realized by using Field Programmable Gate Array to provide the same clock to EOM, DAQ and AOM. Here, data recorded in the laboratory using a coil of sensing fiber is shown in Fig. 2. According to the power spectral density (PSD) of the signal, the noise floor of 50-level large-scale diversity result is about 13 dB lower than conventional system, which is of great importance for the detection of weak signals. The system noise reaches about -65 dB (re rad2/Hz) @500 Hz, which is at the current mainstream level.
2.2 Sensitized optical cable
Due to the axially sensitive feature of an optical fiber, the phase response of probe light to the acoustic wave is extremely low and the acoustic pressure sensitivity is about -212 dB rad/µPa [5]. To get higher acoustic pressure sensitivity, structure design is beneficial and the hollow winding [11] is introduced for the sensitized optical cable in this work. The supporting mandrel of the optical cable is made of acoustic sensitive material, on which the bending resistant fiber is tightly wound. The Young’s modulus of supporting mandrel is 660 MPa and Poisson’s ratio is 0.36. To protect the wound fiber and ensure the acoustic coupling efficiency, the outermost layer of the cable is extruded with a layer of polyurethane cable sheath. The cable diameter is about 16 mm, the density is 1.3 g/cm3, and the length of fiber uniformly wound on each meter of the cable is about 5.35 m.
To quantify the acoustic pressure sensitivity of the sensitized optical cable, a frequency response test of 20-1000 Hz is conducted. The diagrams of the schematic and experiment are shown in Fig. 3(a). A standard sound source is located at the bottom of the standing wave tube, and is driven by a signal generator. A 1m-long optical cable is wound into a loop and placed with the piezoelectric (PZT) hydrophone at the same wavefront position emitted by the standard sound source in the standing wave tube. The PZT hydrophone is the calibrated standard hydrophone as the reference. According to the phase change value demodulated by Φ-OTDR and the acoustic pressure detected by the PZT hydrophone, the sensitivity of the sensitized optical cable can be obtained, and the experimental results are shown in Fig. 3(b). In the frequency range of 20 Hz to1000 Hz, the sensitivity of the sensitized optical cable is about -137.12 dB rad/µPa with the standard deviation is 1.2 dB rad/µPa. Overall, the newly designed sensitized optical cable has a high-and-flat acoustic pressure sensitivity.
Fig. 3. (a) The sensitivity measurement schematic and experiment diagrams. (b) The sensitivity of the optical cable.
2.3 Array orientation method based on APDC
In actual environment, the detection signal of the array will certainly be interfered by cable position variation from water flow, which will lead to a large delay jitter, and the conventional array orientation algorithms have poor accuracy. To solve this problem, an array orientation method based on APDC is proposed. Firstly, the array model of Φ-OTDR is defined according to its sensing characteristics, and the response principle of the sensing array elements to the target source and the phase distribution of the detection signal of each array element are analyzed. Then, according to the phase distribution feature of the detection signal on the sensing array, an APDC method is proposed to construct the optimized signal that is not affected by delay jitter. Finally, the multiple signal classification (MUSIC) algorithm [12] is used to estimate the spatial spectrum of the optimized signal to obtain an estimate value of the spatial target source azimuth angle.
Φ-OTDR converts fiber into a distributed array of acoustic sensors, which is different from point sensors. The detected signal of Φ-OTDR is the integral of acoustic signals over a spatial aperture along the optical fiber, and is obtained from spatial phase difference between two positions separated by gauge length. Thus, the array signal model is necessary for Φ-OTDR. In the far field [13,14], the array signal model is built and the array parameters are defined, including array element aperture, array element spacing and array aperture, shown in Fig. 4. The corresponding spatial positions of the m-th array element beginning and the end are respectively assumed as ${Z_{m,a}}$ and ${Z_{m,b}}$, and the array element aperture is $\Delta Z = {Z_{m,b}} - {Z_{m,a}}$, corresponding to the spatial scale of gauge length. The array element spacing is defined as the distance between the initial positions of two neighboring array elements, $d = {Z_{m,a}} - {Z_{m - 1,a}}$. The array aperture is defined as the distance from the beginning of the first array element to the end of the last array element, $L = {Z_{M,b}} - {Z_{1,a}}$. M is the number of array elements.
According to the Φ-OTDR sensing principle, the detected signal is the integral of the fiber strain distribution generated by the external acoustic wave over the array element aperture [13]. In the far field, assumed a target at azimuth angle $\alpha $ to the cable normal, the target signal is expressed as, $\varepsilon (z,t) = \exp (j\omega t + jkz\sin \alpha )$. The detected signal for the m-th array element is denoted as,
According to Eq. (1), when the incident direction is fixed, the phase of detected signal is linearly related to the array element position and the linear relationship is insusceptible to the integration effect of Φ-OTDR in the far field. To further verify the phase distribution of the detected signal in Eq.1, the numerical simulation is carried out. Assuming that there is a far-field signal in space, take the underwater acoustic signal as an example, the target signal frequency is 100 Hz, the wave speed is 1450 m/s. The array configuration is one-dimensional uniform linear array, the array element spacing is 7.25 m that half of the wavelength of the signal source, and the number of the array elements is 20, and the length of the uniformly wound fiber on each meter of the cable is about 5.35 m. Figure 5(a) shows the phase variation curves of the detected signal of each equivalent array element obtained from the simulation. From the results, the phase variation curves show a linear trend when the angle is determined.
Fig. 5. (a) The phase distribution of the detected signal in simulation. (b) The phase distribution of the detected signal in experiment and prediction.
In the practical working environment, the detected signal is usually interfered by the delay jitter. The phase distribution results of field is shown in Fig. 5(b). The cable layout is shown in Fig. 6, as a linear array. The ship is located at about 200 m distance and -20° azimuth angle, visually detected with a telescope. The 97 Hz main frequency of ship is utilized for analysis. The spatial channel spacing is 5 m in the fiber axis (1∼40 channels for 1575m∼1775 m). The results without delay jitter are obtained by linear fitting and shown as the red line in Fig. 5(b), the delay jitter is about ±0.5 rad between the red and blue line. According to analysis, the delay jitter is derived from many factors, including cable position variation, element aperture, water velocity variation, etc. Firstly, cable position variation is assumed as 1 m and the phase change is $\omega \Delta y\cos \alpha /v = 0.4rad$, $\Delta y$ and v represent the cable position variation and the velocity in the water. Secondly, the non-negligible element aperture is 1.8 m in the cable axis, limited from gauge length, and the phase change is $\omega \Delta L\sin \alpha /v = 0.24rad$, $\Delta L$ represents element aperture limited from gauge length. Thirdly, water velocity variation is mainly influenced by temperature changes. Assumed the temperature fluctuation is ±5°C, the phase change is $\omega \Delta x\sin \alpha /v - \omega \Delta x\sin \alpha /v^{\prime} = 0.\textrm{002}rad$, $\Delta x$ and $v^{\prime}$ respectively represent the length of cable corresponding to spacing for spatial diversity and the velocity in the water affected by the temperature distribution. In addition, the influence of system SNR and reverberation is ignored, according to quantitative analysis. In short, the cable position variation and element aperture are the main factors for delay jitter.
To eliminate the influence of delay jitter, the APDC method is proposed, according to the linear relationship of the detected signal phase of each equivalent array element. Under the ideal condition without delay jitter, the detected signal of the m-th equivalent array element is ${S_m}(t)$ and its phase is ${\varphi _m}$. Due to the practical array working environment, the phase of the m-th equivalent array element detected signal is deteriorated as,
where $\varphi _m^{noise}$ is the phase error from delay jitter. The corresponding original detected signal is expressed as $S_m^{orig}(t)$, with delay jitter. By using the least square method to linear fit $\varphi _m^{orig}$ to obtain $\varphi _m^{fit}$, and the phase error introduced by delay jitter can be expressed as, $\delta {\varphi _m} = \varphi _m^{orig} - \varphi _m^{fit}$. And then, the optimized signal $S_m^{opt}(t)$ can further be constructed, where j is the imaginary unit. Therefore, the detected signal which is not affected by delay jitter can be obtained by this method.After obtaining the optimized signal $S_m^{opt}(t)$ by using the APDC method, the spatial spectrum is estimated by utilizing the MUSIC algorithm. The principle is to decompose the eigenvalues of the covariance matrix of the equivalent array element detected signal on the sensing array $[S_1^{opt}(t),S_2^{opt}(t), \cdots ,S_M^{opt}(t)]$. The eigenvectors corresponding to the different eigenvalues form the mutually orthogonal signal subspace ${U_s}$ and the noise subspace ${U_N}$. Then, the spatial spectrum function ${P_{MUSIC}} ={-} \lg ({A^H}(\alpha ){U_N}U_N^HA(\alpha ))$ is constructed by using the orthogonality between the signal subspace $A(\alpha ) = {[1,{e^{jkd\sin \alpha }}, \cdots ,{e^{jk(M - 1)d\sin \alpha }}]^T}$ and the noise subspace, and finally, the spectrum peak is searched within the solution space to obtain the estimation value of the spatial target source azimuth angle.
3. Testing and validation
The outfield test is conducted in a reservoir in Guangdong Province, and the deployment is shown in Fig. 6. The Φ-OTDR system is placed in the server room and a section of sensitized optical cable with a length of 300 m is placed in the water. The length of about 500 m of guide fiber is used to connect the Φ-OTDR system to the sensitized optical cable in the water. The sensitized optical cable is bound to a float ball and a weight block for every 10 m to keep horizontal state and suspended in a water depth of 5 m. From the top view, the optical cable is laid in a “U-shape”, which is divided into the long side and the short side of the optical cable, in which the spatial spacing distance between them is 2.8 m, and the cable interval is 10 m. The pulse repetition rate of the Φ-OTDR is 5kHz, and the total length of the sensing fiber is 2.1 km. In the test, a non-cooperative ship as the target source which the maximum sound level is about 150dB@97 Hz, moving towards the cable and approximately 200 m far from the cable.
3.1 Analysis of the ship’s voiceprint information
The ship signal is detected by Φ-OTDR and smoothing filtered is applied to obtain the low-frequency noise, from laser frequency drift and environmental noise. And then the result signal is extracted by subtracting the low-frequency noise from the detected signal. To quantitatively analyze the intensity distribution of the ship’s signal power spectrum, the detected signal is quantitatively converted from demodulation phase to sound pressure, and the PSD is shown in Fig. 7, at 1085 m along the optical fiber. Apparently, some important peaks (line spectra of the ship) are submerged by noise in the result without diversity (blue line), but clear in the result with diversity (red line), and thus the feasibility and necessity of the large-scale diversity are verified. Further, the power spectrum of the ship is composed of continuous spectrum and multiple line spectra, compared with the environmental noise (yellow line). On the one hand, the strongest peak of 97 Hz and the weak continuous spectrum around it are considered to be mechanical noise [16] of the ship. The spectrum on the left side of the peak is relatively flat, and on the right side shows a decreasing trend, like Fig. 7(b); the strongest line spectra of 97 Hz is higher than the nearby continuous spectrum by about 20 dB, which is consistent with the acknowledged feature of mechanical noise [16]. Moreover, the interval of line spectra is 12 Hz, shown in Fig. 7(b), which is closely related to the sailing state and mechanical working condition of the ship. On the other hand, the peak located at 530 Hz in Fig. 7(a), whose value is within the ten-octave range of 100-1000 Hz. Moreover, in the continuous spectrum at the left of the peak, the spectrum level increases with the frequency, and the slope is +8 dB/oct; at the higher frequency, the spectrum level decreases with the frequency, and the slope is -6 dB/oct. These features are consistent with that of cavitation noise from propeller [16]. Note that, the environmental noise (yellow line) is slightly higher than ship signal in frequency below 90 Hz, which is related to the inevitable environment changes. As reference, an electrical hydrophone (Soundtrap ST300) is placed in the same position as cable at about 1788 m in the fiber axis. The results are consistent with those of Φ-OTDR and the ship voiceprint information is further verified. In short, the high-fidelity acquisition of ship’s voiceprint information is realized, and the characteristics of ship noise spectrum are analyzed. It is believed that the voiceprint detection will promote the target identification and classification in the future.
Fig. 7. (a) Intensity distribution of the ship’s signal spectral lines and (b) its partial enlarged detail.
3.2 Over-top detection of the ship
Over-top detection of the target provides the target position as a function of the distance in the one-dimensional axial space along the cable, which is widely used in maritime security. A moving ship as surface intrusion target is detected, like Fig. 6. According to the maximum response of 97 Hz in Fig. 7(a), 4 Hz narrowband filtered is applied to the detected signal, the signal is converted into the square of sound pressure, and the waterfall diagram is obtained as shown in Fig. 8. It can be seen that when the ship starts moving, the signal response of the optical fiber sections 1000-1210 m and 1630-1840 m is large. Due to the cable is laid in a “U-shaped”, the maximum response position of the two sections of optical fiber to the opposite trend with time is in line with the theory. Therefore, the over-top detection of the ship signal with a lateral distance of 200 m is realized by utilizing the suspended optical cable.
Fig. 8. The waterfall diagram of the ship signal response. (a) The optical fiber section of 1000-1210 m. (b) The optical fiber section of 1630-1840 m.
3.3 Array orientation results
To further obtain the target's approach orientation and driving trajectory in designated area, the two-dimensional (2D) target orientation is particularly important. To obtain the azimuth angle of the moving ship in Fig. 6, the spatial spectrum of the signal acquired by the sensing array is estimated. The frequency of the ship signal is selected to be 97 Hz, which has the highest spectrum intensity. The sensing array is selected to be 1575-1775 m of the optical fiber, corresponding to the length of the optical cable is 37.38 m. The starting position of the sensing channel at 1575 m of the optical fiber is selected as the reference point, the array element number is 6, and the array element spacing is 7.47 m. Figure 9 shows the variation curve of the ship’s azimuth angle within 20 seconds, and each data point in the figure indicates the azimuth angle within 1 second. The blue data points are the azimuth angles obtained by utilizing the conventional MUSIC algorithm. The fluctuation boundary of the azimuth angle results is obtained to evaluate the angle fluctuation, by fitting the results and shifting to the upper and lower boundary, and the boundary spacing (angle fluctuation) is 5.6°. Similarly, the red data points are the results obtained by utilizing APDC method to construct the optimized signal and then carry out the orientation tracking, and the angle fluctuation is 3°. The comparison shows that the array orientation method based on APDC can reduce the azimuth angle fluctuation by nearly half, which is conducive to improving the orientation detection capability of Φ-OTDR.
4. Discussion
In this work, the voiceprint information of real target is obtained with Φ-OTDR, which will promote the target type identification in marine applications; the one-dimensional localization and 2D orientation are achieved, as a principal proof for non-cooperative targets. In the future, the target type identification will be achieved with deep learning, the offset distance between the target and the sensitized optical cable can be obtained by utilizing two far-apart arrays at different azimuth angle of targets [13], and thus the 2D localization will be achieved for area monitoring. Moreover, the target depth will be further obtained with either two parallel arrays or an additional vertical array.
5. Conclusion
This paper proposes the 50-level large-scale diversity Φ-OTDR system combined with a suspended sensitized optical cable for non-cooperative ship detection. The high-fidelity acquisition of the ship’s voiceprint information is realized for the first time and the signal features of non-cooperative ship are analyzed in the actual environment. Moreover, the over-top detection of the non-cooperative ship is realized. In addition, the array orientation method based on APDC is proposed to solve the problem of delay jitter in acoustic wave transmission. The results of the outfield test in the reservoir show that the array orientation method based on APDC can reduce the azimuth angle fluctuation nearly half. The proposed method will enhance the orientation detection capability of Φ-OTDR in practical applications. This work provides a marine target detection technology, and it will promote the development of hydroacoustic detection.
Funding
National Key Research and Development Program of China (2023YFB290530); National Natural Science Foundation of China (62175246, U23A20379); Science and Technology Commission of Shanghai Municipality (23DZ1203900, 23xtcx00500); Shanghai Rising-Star Program (22QB1406000); Youth Innovation Promotion Association of the Chinese Academy of Sciences (YIPA2023257); Shanghai Sheshan National Geophysical Observatory (SSOP202201).
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. S. Fang, S. Du, X. Luo, et al., “Development of Underwater Acoustic Target Feature Analysis and Recognition Technology,” Bulletin of the Chinese Academy of Sciences 34(3), 297–305 (2019).
2. Z. Pan, K. Liang, Q. Ye, et al., “Phase-sensitive OTDR system based on digital coherent detection,” in 2011 Asia Communications and Photonics Conference (ACP) IEEE, 83110S (2012).
3. C. Wang, Y. Shang, X. Liu, et al., “Distributed OTDR-interferometric sensing network with identical ultra-weak fiber Bragg gratings,” Opt. Express 23(22), 29038–29046 (2015). [CrossRef]
4. V. S. Lavrov, M. Y. Plotnikov, S. M. Aksarin, et al., “Experimental investigation of the thin fiber-optic hydrophone array based on fiber Bragg gratings,” Opt. Fiber Technol. 34, 47–51 (2017). [CrossRef]
5. B. Lu, B. Wu, J. Gu, et al., “Distributed optical fiber hydrophone based on Φ-OTDR and its field test,” Opt. Express 29(3), 3147–3162 (2021). [CrossRef]
6. G. Yan, D. Wang, J. Long, et al., “High-performance towing cable hydrophone array with an improved ultra-sensitive fiber-optic distributed acoustic sensing system,” Opt. Express 31(16), 25545–25556 (2023). [CrossRef]
7. Y. Liu, J. Yang, Z. Wang, et al., “High Performance Miniaturized DAS-based hydrophone (HyDAS) with Spatial Deviation Method for Ship Detection,” in 28th International Conference on Optical Fiber Sensors (OFS-28), Tu3.35 (2023).
8. J. Chen, H. Li, X. Xiao, et al., “Fully distributed hydroacoustic sensing based on ultra-highly sensitive and lightweight fiber-optic hydrophone cable,” Opt. Lasers Eng. 169, 107734 (2023). [CrossRef]
9. J. Gu, B. Lu, J. Yang, et al., “High SNR phi-OTDR Based on Frequency and Wavelength Diversity With Differential Vector Aggregation Method,” IEEE Photonics J. 12(6), 1–12 (2020). [CrossRef]
10. Z. Wang, J. Yang, J. Gu, et al., “Practical Performance Enhancement of DAS by Using Dense Multichannel Signal Integration,” J. Lightwave Technol. 39(19), 6348–6354 (2021). [CrossRef]
11. M. J. Murray, A. Davis, and B. Redding, “Fiber-wrapped mandrel micro-phone for low-noise acoustic measurements,” J. Lightwave Technol. 36(16), 3205–3210 (2018). [CrossRef]
12. R. Schmidt, “Multiple emitter location and signal parameter estimation,” IEEE Trans. Antennas Propag. 34(3), 276–280 (1986). [CrossRef]
13. J. Liang, Z. Wang, B. Lu, et al., “Distributed acoustic sensing for 2D and 3D acoustic source localization,” Opt. Lett. 44(7), 1690–1693 (2019). [CrossRef]
14. S. Doclo and M. Moonen, “Design of far-field and near-field broadband beamformers using eigenfilters,” Signal Process 83(12), 2641–2673 (2003). [CrossRef]
15. Y. Yang, T. Xu, S. Feng, et al., “Optical Fiber Hydrophone Based on Distributed Acoustic Sensing,” Fiber Optic Sensing and Optical Communication 10849, UNSP 108490B (2018).
16. B. Liu and J. Lei, Principles of hydroacoustics (2nd Ed.), (Harbin Engineering University Press),(2009), pp. 265–268. (in Chinese)
