We propose a frequency splicing code-based Brillouin optical time domain collider (FSC-BOTDC) for fast dynamic sensing. By delicately designing the frequency splicing code (FSC), multiple collision modes with controllable characteristics are realized for probing multiple target areas with a high sampling rate. Moreover, the sensing system is simpler and more robust than the previous BOTDC. In the experiment, the FSC-BOTDC with 10-time enhanced sampling rate is implemented for single and multiple target areas measurements. Results demonstrate that tailorable measurements can be achieved by the tunable FSC. Furthermore, the FSC-BOTDC is executed to measure periodic mechanical vibrations over 7.9-km sensing range with the sampling rate of 625 Hz.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Brillouin-based distributed optical fiber sensors have gained significant attentions in the past few decades due to its capability to measure physical quantities, such as the strain and temperature along the optical fiber [1,2]. Among them, the Brillouin sensors based on time-domain detection scheme such as Brillouin optical time domain analysis (BOTDA)  and Brillouin optical time domain reflector (BOTDR)  are intensively investigated and widely applied in the dynamic sensing, since they can support long sensing distance with high spatial resolution, large dynamic strain measurement range and high precision. Generally, the dynamic sampling rate in the time-domain Brillouin sensors is limited by three main factors [3,5]: frequency sweeping process, trace averaging and pulse repetition rate limited by the fiber length.
So far, lots of research efforts have been made to enhance the sampling rate , and can be classified into two categories: 1) shortening or skipping the frequency sweeping process: Frequency-agile technique based on an arbitrary waveform generator (AWG) can largely shorten the frequency switching time [5,6]. Moreover, by incorporating with compressed sensing method, the number of sweeping frequencies can also be reduced, which enhances the sampling rate and reduces data volume . Furthermore, to entirely avoid the frequency sweeping process, three major techniques including slope-assisted [8–14], optical frequency comb [15–18] and optical chirp chain [19–22] are proposed; 2) decreasing the averaging number: system configuration optimization , optical pulse coding [24,25], signal denoising algorithms  and polarization diversity technology  are effective ways to enhance the signal-to-noise ratio (SNR) and thus reduce the averaging number. Fueled by the above methods, the sampling rate can be limited only by the fiber length [3,5].
Recently, we reported a Brillouin optical time domain collider (BOTDC) scheme to break through the limit of fiber length to the pulse repetition rate . In this BOTDC, by employing dual frequency hopping light waves (pulsed pump and CW probe), the sensing information crosstalk can be avoided even though the pulse period is shorter than the round-trip time in the fiber. Meanwhile, since the stimulated Brillouin scattering (SBS) interaction is periodically occurred at a specific fiber region, non-local effect caused by pump depletion can be alleviated due to reduced SBS interaction length . This allows the use of the probe wave with higher power to enhance the SNR and reduce the averaging number. Moreover, there is only one hopping frequency at each moment. Thus, the dynamic signal can be extracted by direct detection and low sample-rate data acquisition equipment. This results in low data volume and potentially high real-time capability. However, in the previous BOTDC, 1) the single collision mode can probe only one target area at one time. When multiple discretely distributed target areas are required to measure simultaneously, the sampling rate has to compromise with the spacing between the target areas; 2) The frequency hopping pump and probe waves are generated by two individual frequency hopping light sources (FHLS). This kind of dual FHLS-based scheme inevitably leads to a system with high complexity; 3) Although the probe bias tracking-based trace normalization method can effectively eliminate the bias fluctuation, two detection branches are required. This increases the system complexity and data volume accordingly. As such, it is highly desirable to investigate a new Brillouin collider that provides the capability to probe multiple target areas with a significantly simplified system.
In this paper, a frequency splicing code-based BOTDC (FSC-BOTDC) is proposed for high sampling rate dynamic sensing. In this FSC-BOTDC, counter-propagating pump and probe waves are sourced from a frequency splicing coded light (FSCL), and shaped by a synchronous light chopper (SLC) and frequency shifter (FS), respectively. The FSCL in each period is constituted by sequentially spliced sections with delicately selected hopping frequencies, lengths and delays. Each section has its specific feature and purpose. They harmonize and cooperate with each other to spark multiple collision modes with particular characteristics after pump-probe colliding in the fiber. By properly designing the frequency splicing code (FSC), the controllable collision modes can achieve the dynamic measurements with variable features including sampling rates, target areas numbers, locations and lengths. As a result, the key targets along multiple fiber areas can be measured with a significantly enhanced sampling rate, compared with the conventional dynamic BOTDA. Moreover, thanks to the FSC-based scheme, the system and operation are quite simple: 1) Only one frequency hopping light source is needed for generating the FSCL; 2) Once the power unevenness of the FSCL is pre-compensated, pump and probe power flattening can be automatically achieved in all hopping and sweeping frequencies. As a result, the sensing signals corresponding to different hopping frequencies can always have the same Brillouin gains and SNRs; 3) Instead of the probe bias tracking , the trace normalization can be directly realized with only one detection branch. This further decreases the system complexity and 50% of the data volume. In the experiment, the FSC-BOTDC with 10-times enhanced sampling rate is implemented for single and multiple (2 and 4) target areas measurements. The results demonstrate that the measurements with variable characteristics can be achieved by the tunable FSC. Moreover, the FSC-BOTDC is implemented to measure periodic mechanical vibration over 7.9 km sensing range with the sampling rate of ∼625 Hz.
The operation principle of the FSC-BOTDC for single target area measurement is illustrated in Fig. 1(a). The delicately designed FSCL [fi, fj│fi+1, fj+1│…│fi-2, fj-2│fi-1, fj-1] with N hopping frequencies is emitted from a single FHLS. Where i and j are positive integers (1 ≤ i, j ≤ N) and denote the start orders of the cyclic hopping frequency sequences. The f1<f2<…<fN-1<fN, and the frequency spacing between any two adjacent hopping frequencies is set to be much larger than the overall frequency range of the BGS. The FSCL in each pulse period (TP) contains two segments: FSC-P and FSC-CA1. The first short FSC-P segment [fi│fi+1│…│fi-2│fi-1] is employed for frequency hopping pump pulse generation. The second long FSC-CA1 segment [fj│fj+1│…│fj-2│fj-1] is executed to acquire sensing information along the collision area (CA1). The duration of FSC-P is determined by pump pulse width and, in general, tens of ns. After splitting into two branches, the FSCL on the left branch is periodically chopped by the SLC. Then, the FSC-P segment is only existed and becomes the frequency hopping pump pulses. Meanwhile, the FSCL on the right branch is frequency up/down-converted by the FS to make the pump-probe frequency offset Δf falls within the range of Brillouin loss/gain spectrums.
When the pump and probe counter-propagate in the fiber, two kinds of collision modes occur: 1) The pump interacts with the short probe FSC-P segment through the SBS: Since the frequency relation between the pump and probe FSC-P is constant, the collision area (CA0) is locked at the center of the fiber; 2) The pump interacts with the long probe FSC-CA1 segment through the SBS: Due to the large frequency spacing between any two adjacent hopping frequencies, each pump pulse with a specific frequency (e.g., f1) only interacts with the relevant frequency probe FSC-CA1 segment (f1±Δf). Thus, the sensing information crosstalk between adjacent pulses will not occur even though the pulse period is shortened to be one Nth of the round-trip time in the fiber. For different FSC, the pulse repetition rate (Rp) and collision area (CA1) are given by:
Based on Eq. (1) and (2), we can find that 1) the Rp and dynamic sampling rate are directly proportional to the number of hopping frequencies (N); 2) By changing pump-probe relative delay DR through adjusting the pump-probe frequency relation (i-j), pulse spacing (ΔZP) and pump delay (Dp), the CA1 (sensing area) can be flexibly varied. For instance, by employing the FSC [f1, f3│f2, f4│f3, f1│f4, f2] (N=4, Dp=0, i=1 and j=3), the target area from L/2-LT/2+LFSC-P to L/2-LT/4 can be continuously measured by 4 times within the total period, indicating 4-times enhanced sampling rate. By changing the FSC in similar ways, measurements with other characteristics can be achieved. From the above analyses, we can obtain that one collision mode can achieve high sampling rate dynamic measurements when the measurands of interest densely distribute in a specific fiber region. The sampling rate should be properly chosen according to the total length of the measurands.
On the other hand, when the measurands of interest discretely distribute in several fiber areas, more collision modes can be employed to keep the pulse repetition rate free from the spacing between the target areas. In detail, as illustrated in Fig. 1(b), the FSCL in each pulse period is constituted by more FSC-CA parts with delicately selected hopping frequencies, lengths and delays. This enables each probe FSC-CA part to have a particular relative delay DR with the pump. Finally, multiple collision modes occur and the CA can discretely distribute in several fiber areas (with the CA spacing larger than the pulse spacing ΔZp) to measure the multiple target areas simultaneously with a high sampling rate. For instance, by employing the FSC [f1, f4, f3│f2, f1, f4│f3, f2, f1│f4, f3, f2] (N=4, Dp=0, i=1, j=4, k=3, the durations of FSC-CA1 and FSC-CA2 are equal), even though two target areas from 1) L/2-3LT/8+LFSC-P/2 to L/2-LT/4 and 2) L/2+LT/4+LFSC-P to L/2 + 3LT/8+LFSC-P/2 are far away (3LT/4) from each other, they can be measured simultaneously with 4-times enhanced sampling rate. In this way, the pulse repetition rate can be mainly determined by the measurands’ total length (i.e., the sum of measurement lengths of the targets).
In addition, trace normalization and Brillouin gains flattening are also two important aspects for high-quality measurements. Generally, the trace normalization [24,27] is applied to eliminate detrimental probe bias fluctuations caused by system components (such as laser, modulator, amplifier and so on) or external disturbances. To implement the trace normalization in the FSC-BOTDC, two preliminary works need to be done: probe power flattening and DC component (without gain) acquisition. Moreover, to achieve the Brillouin gains flattening, pump power differences corresponding to different hopping frequencies need to be cancelled (i.e., pump power flattening). Fortunately, the FSC-based scheme enables the two aspects to be achieved with low system and operation complexities. Firstly, since the pump and probe waves are sourced from the same FSCL, once the FSCL power unevenness is eliminated via pre-compensation, the power flattening of the pump and probe can be automatically completed at the same time. Secondly, an additional short FSC-Z segment is introduced in the FSC for acquiring the DC component. In detail, since the pulse period is shorter than the round-trip time in the fiber, continuous SBS interaction occurs in most cases . The DC component can only exist when the collision area contains SBS interruption area (L-LT)/2+LFSC-P to 0 or L to L+(LT-L)/2. To acquire the DC component in other cases, the short FSC-Z is set behind the FSC-P. The relative delay between the FSC-P and FSC-Z is locked to ensure that their collision always locates in the SBS interruption area. After that, the DC component can always be acquired by the FSC-Z. After the probe power flattening and DC component acquisition, the trace normalization can be achieved with only one detection branch.
3. Experimental setup
The experimental setup of the FSC-BOTDC is shown in Fig. 2. A self-fabricated FHLS is employed to generate the FSCL, as shown in the blue dotted box. The CW light with a particular wavelength is emitted from a narrow linewidth (∼100 kHz) external cavity laser (ECL) and then converted into the FSCL through an electric-optic modulator (EOM1). The EOM1, worked in a carrier-suppressed double-sideband (CS-DSB) modulation mode, is driven by the frequency splicing coded electric signals from an AWG (hopping from 2 GHz with 250 MHz step). After amplified by an erbium-doped fiber amplifier (EDFA1), one of the sidebands is selected by a fiber Bragg grating (FBG) and then split into two branches by a 50:50 coupler. Since the responses of the AWG, LNA and EOM1 change with the hopping frequency, the FSCL amplitudes corresponding to different hopping frequencies are different. Thus, the pre-compensation is adopted to flatten the FSCL and obtain the flat pump and probe waves. In detail, the digital frequency splicing coded signal amplitudes are reversely adjusted according to the amplitude differences in the generated FSCL.
In the upper branch, the FSCL is frequency up-/down-converted by the EOM2, driven by radiofrequency (RF, sweeping from 10.76 GHz to 11.01 GHz with 2 MHz step) signals in the CS-DSB mode. After being amplified by the EDFA2, the probe is injected into the fiber-under-test (FUT). The probe input power is −7.3 dBm. The FUT is a 7.9-km standard single-mode fiber. A 6.95 m section at the end of the FUT is stretched and applied dynamic strain via an eccentric wheel driven by an electro-motor. In the lower branch, the FSCL is periodically chopped by a semiconductor optical amplifier (SOA) which is synchronized with the AWG. After that, the FSC-P section is only existed and becomes the pump pulse. The effective pulse width and extinction ratio are 60 ns and 60 dB, respectively. Then, to eliminate polarization fading, the state of polarization (SOP) of the pump is randomized by a polarization scrambler (PS). After being amplified by a pulse-EDFA (P-EDFA), the pump is injected into the FUT through a circular. The pump peak power is 26 dBm.
At the receiver, the lower sideband of the probe (Stokes component) is filtered by an optical band-pass filter (OBPF) and then detected by a photodetector (PD, Thorlabs, PDB425C). The bandwidth and saturation power of the PD are 75 MHz and −18 dBm, respectively. Finally, the detected electric signal is A/D converted by an oscilloscope with 100 MSa/s sample rate.
4. Experimental results
Figure 3(a) and 3(b) illustrate the Brillouin gain trace (BGT) at 10.86 GHz (around the BFS) and BGS distribution along the fiber in the BOTDA, respectively. In this and following static property analyses, averaging number of 500 is adopted to sufficiently eliminate the polarization noise and enhance the SNR. The pulse period is set as 82 μs which is slightly larger than the round-trip time in the fiber (∼80 μs). As a result, the DC component reaching the fiber near or far end without gain can be acquired and adopted to implement the trace normalization. The peak Brillouin gain is ∼11%. Significant BGS distribution variations from near and far end of the fiber caused by fiber pre-strain can be clearly observed. The frequency range of overall BGS distribution is less than 150 MHz. Thus, the 250 MHz hopping frequency spacing is enough to avoid the sensing information crosstalk and enable the FSC-BOTDC to work properly.
4.1 FSC-BOTDC for single-target area measurement
Before validating the FSC-BOTDC, the effectiveness of the FSCL pre-compensation, trace normalization and Brillouin gains flattening needs to be verified. Figure 4(a) shows the time-domain traces after the EOM1 (i.e., the FSCL). The corresponding FSC is the [f1, f3│f2, f4│f3, f1│f4, f2]. Clearly, without the pre-compensation (red lines), there are significant FSCL amplitude differences between different hopping frequencies (i.e., the FSCL unevenness). The unevenness ratio is 11.5%. After amplified by the EDFA1 and EDFA2, the probe unevenness is further enlarged to 14%, meanwhile the trace is distorted in various degree (3.3% in maximum) due to the transient response of the EDFA [29,30], as shown in Fig. 4(b) and 4(c) (the pump pulse is off). On the contrary, by employing the pre-compensation, the FSCL and probe become quite flat (probe unevenness less than 0.4%, trace distortion less than 0.2%), as depicted in Fig. 4(a)–4(c) (blue lines). This flatness results in 1) the feasibility of trace normalization and 2) the same Brillouin gains and SNRs between different hopping frequencies. Figure 4(d) to 4(f) illustrate the BGS distribution, BGT (at 10.86 GHz, CA: −120 to 1880 m) and BGS (at 100 m) in the FSC-BOTDC, respectively (the pump pulse is on). The trace section from −120 to 0 m corresponds to the DC component without gain and is adopted for the trace normalization. Obviously, the probe bias fluctuations are eliminated via the trace normalization. Meanwhile. the Brillouin gains corresponding to different hopping frequencies are almost the same (gain unevenness: less than 0.2%).
After validating the effectiveness of the FSCL pre-compensation, trace normalization and Brillouin gains flattening, the FSC-BOTDC is investigated. Here, two verification methods are employed in the following sections.: 1) BGT with polarization fading pattern: It is known that the efficiency of the SBS depends on the degree of parallelism of pump and probe SOPs. Meanwhile, the degree of parallelism is varied according to local fiber state (local strain, temperature, bend, twist, among other). As a result, different fading patterns occur at different fiber locations. Accordingly, the polarization fading pattern can be a perfect position marker for details observation. 2) BGS distribution: Besides the polarization fading pattern, the BGS distribution without polarization fading can also be a position marker for overall observation.
Figure 5(a) shows the FSC for measuring the target area from 2034 to 3980 m in the FSC-BOTDC. Since the SBS interruption area is not contained in the CA, the FSC-Z is adopted to acquire the DC component without gain for the trace normalization. Figure 5(b1) and 5(b2) illustrate the BGTs (at 10.86 GHz) with polarization fading pattern (i.e., without the PS) in the BOTDA and FSC-BOTDC, respectively. Figure 5(b3) shows the details from 22 to 41.4 µs. Clearly, the fading pattern from 2034 to 3980 m in the BOTDA totally coincides with that in each period in the FSC-BOTDC, and repeats by 4 times within 82 μs (total period). Moreover, Fig. 5(c) illustrates the BGS distribution in the FSC-BOTDC. Here, the polarization fading is eliminated by employing the PS. From Fig. 5(b) and 5(c), it can be seen that the target area is measured with 4-times enhanced sampling rate in the FSC-BOTDC. In this and following analyses, the durations of the FSC-P and FSC-Z are set as 500 ns and 1000 ns, respectively, for sake of visual clarity. Furthermore, the FSC with 10 hopping frequencies is studied. Figure 5(d) shows the FSC for measuring the target area from 7267 to 8037 m, as illustrated in Fig. 5(e) and 5(f). The 7891 to 8037 m corresponds to the SBS interruption area at far end of the fiber. Thus, the trace section from 7891 to 8037 m can be directly employed for trace normalization. Obviously, the sensing information along the target area is continuously measured by 10 times within 82 µs, indicating 10-times enhanced sampling rate. By changing the FSC in similar ways, measurements with other characteristics can be achieved.
4.2 FSC-BOTDC for multiple-target areas measurement
In this section, the FSC-BOTDC for multiple target areas measurement is studied. Figure 6(a) shows the FSC for measuring two target areas from 1) −108 to 411 m and 2) 4513 to 4757 m simultaneously in the FSC-BOTDC. The durations of CA1 and CA2 in each pulse period are 5.74 and 2.46 μs, respectively. Clearly, even though there is a large spacing (4.1 km) between the two target areas, the sensing information corresponding to the two target areas is probed simultaneously in each period, as illustrated in Fig. 6(b) and 6(c). This indicates the two target areas are measured with 10-times enhanced sampling rate. Moreover, by utilizing the FSC shown in Fig. 6(d), four target areas located from 1) −108 to 127 m, 2) 5047 to 5167 m, 3) 2708 to 2829 m and 4) 7751 to 8037 m are probed simultaneously with 10-times improved sampling rate, as illustrated in Fig. 6(e) and 6(f). The durations of CA1, CA2, CA3 and CA4 in each pulse period are 2.87, 1.23, 1.23 and 2.87 μs, respectively. From the experimental demonstrations above, the feasibilities of the FSC-BOTDC for single- and multiple-target areas measurements are validated, indicating that the tailorable measurements can be realized by the tunable FSC. The sampling rate enhancement should be properly chosen according the total length of the target areas. Based on the concepts in this paper, we can design more sophisticated FSC to execute the tailoring sensing with higher performance.
4.3 FSC-BOTDC for dynamic strain measurement
In this section, the FSC-BOTDC for strain measurement is studied. The FSC shown in Fig. 5(d) is adopted. Firstly, longitudinal strains are gradually applied on the stretched fiber section at the end of the fiber, with a step of 144 μɛ. Figure 7(a) and 7(b) show the measured BGSs and estimated BFSs (through Lorentz curve fitting) corresponding to different strains, respectively. Here, the 10 hopping frequencies-related BGSs and BFSs are plotted together. It can be seen that 1) the BGSs and BFSs related to the 10 hopping frequencies coincide with each other, which demonstrates that they have the same responses to the strain and 2) the BFS is linearly related to the strain. Then, after linearly fitting the strain-BFS relations, the strain sensitivities related to the 10 hopping frequencies are obtained and plotted in Fig. 7(c). Obviously, the 10 strain sensitivities are almost the same and equal to 0.0492 MHz/µɛ.
In the dynamic strain measurement, the slope-assisted method  is applied in the FSC-BOTDC. Figure 8(a) illustrates the BGSs corresponding to the stretched fiber section at the end of the fiber. The pre-strain is ∼1500 µɛ. Similarly, the 10 hopping frequencies-related BGS are plotted together. Clearly, the 10 BGSs coincide with each other. By employing the linear fitting, the linear region is found to be 10.941 to 10.958 GHz, which corresponds to a dynamic strain measurement range of 346 µɛ, as depicted in the blue region in Fig. 8(a). Meanwhile, the slopes related to the 10 hopping frequencies are estimated and shown in Fig. 8(b). Obviously, the 10 slopes are nearly the same and equal to −0.2107%/MHz.
Finally, the eccentric wheel-induced dynamic strain is measured by the FSC-BOTDC. Figure 9(a) illustrates the measured dynamic strain. Here, a 195-times moving average is employed to alleviate the polarization noise and enhance the SNR. It can be obtained that 1) the sampling rate is ∼625 Hz (1/8.2µs/195) and 2) the peak-to-peak strain variation is ∼120 µɛ. By employing the fast Fourier transform, the frequency spectrum of the measured dynamic strain signal is obtained and shown in Fig. 9(b). The insert figure plots the detail from 0 to 50 Hz. It can be observed that the frequency spectrum contains fundamental and harmonic frequencies of 14.83 and 29.7 Hz, respectively.
In summary, we have proposed and experimentally demonstrated a new FSC-BOTDC for high-speed dynamic sensing. In this FSC-BOTDC, controllable collision modes realized by the tunable FSC are employed for measuring multiple target areas with a high sampling rate. Meanwhile, the FSC-based scheme enables the system and operation to be significantly simplified. In the experimental demonstrations, firstly, the FSCL pre-compensation, trace normalization and Brillouin gains flattening are validated. Then, the FSC-BOTDC with 4- and 10- times enhanced sampling rates for single and multiple (2- and 4-) target areas measurements are demonstrated. After that, the strain response of the FSC-BOTDC is investigated and confirmed. Finally, the dynamic strain with a ∼14.83 Hz fundamental frequency and a ∼29.7 Hz harmonic frequency is measured by the FSC-BOTDC with the sampling rate of ∼625 Hz.
The FSC-BOTDC not only keeps all advantages of the previously demonstrated BOTDC , but also offers the following improvements:
- 1) being capable to probe multiple target areas simultaneously, thus breaking the restriction on the pulse repetition rate due to the spacing between target areas;
- 2) offering tunable FSC and tailorable measurements (including sampling rates, target areas numbers, locations and lengths), thus making the sensor adaptable for various applications;
- 3) requiring only one frequency hopping light source and detection branch, thus making the system complexity and data volume to be significantly decreased;
- 4) automatic power flattening of the probe and pump enables the probe bias fluctuation elimination and Brillouin gains flattening to be achieved easily.
Different from the conventional BOTDA that probe the whole sensing fiber with a fixed sampling rate, the proposed FSC-BOTDC aims to only measure the specific (key) fiber sections with a much higher sampling rate. Thus, it prefers to be applied in the fields that need to measure multiple key targets rapidly at one time, such as the analyses of track dynamic loads during the train moving in high-speed railway systems [31–34], monitoring of debris flow- or landslide-prone sections and so on. Moreover, more investigations about the improvements and the applications of the FSC-BOTDC are expected in the near future. For instance, by designing the FSC with more sophisticated structures such as heterogeneous or high-dimensional coding, the sensing performance and functionality may be further exploited. Furthermore, the design of the FSC may benefit from the artificial intelligent algorithms .
National Natural Science Foundation of China (61735015); Cultivation Program for the Excellent Doctoral Dissertation of Southwest Jiaotong University (2020YBPY05).
The authors declare no conflicts of interest.
1. X. Bao and L. Chen, “Recent progress in Brillouin scattering based fiber sensors,” Sensors 11(4), 4152–4187 (2011). [CrossRef]
2. A. Motil, A. Bergman, and M. Tur, “State of the art of Brillouin fiber-optic distributed sensing,” Opt. Laser Technol. 78, 81–103 (2016). [CrossRef]
3. H. Zhang, D. Zhou, B. Wang, C. Pang, P. Xu, T. Jiang, D. Ba, H. Li, and Y. Dong, “Recent progress in fast distributed Brillouin optical fiber sensing,” Appl. Sci. 8(10), 1820 (2018). [CrossRef]
4. Q. Bai, Q. Wang, D. Wang, Y. Wang, Y. Gao, H. Zhang, M. Zhang, and B. Jin, “Recent Advances in Brillouin Optical Time Domain Reflectometry,” Sensors 19(8), 1862 (2019). [CrossRef]
5. Y. Peled, A. Motil, and M. Tur, “Fast Brillouin optical time domain analysis for dynamic sensing,” Opt. Express 20(8), 8584–8591 (2012). [CrossRef]
6. B. Wang, Z. Hua, C. Pang, D. Zhou, D. Ba, D. Lin, and Y. Dong, “Fast Brillouin Optical Time-Domain Reflectometry Based on the Frequency-Agile Technique,” J. Lightwave Technol. 38(4), 946–952 (2020). [CrossRef]
7. Q. Chu, B. Wang, H. Wang, D. Ba, and Y. Dong, “Fast Brillouin optical time-domain analysis using frequency-agile and compressed sensing,” Opt. Lett. 45(15), 4365–4368 (2020). [CrossRef]
8. R. Bernini, A. Minardo, and L. Zeni, “Dynamic strain measurement in optical fibers by stimulated Brillouin scattering,” Opt. Lett. 34(17), 2613–2615 (2009). [CrossRef]
9. J. Urricelqui, A. Zornoza, M. Sagues, and A. Loayssa, “Dynamic BOTDA measurements based on Brillouin phase-shift and RF demodulation,” Opt. Express 20(24), 26942–26949 (2012). [CrossRef]
10. D. Zhou, Y. Dong, B. Wang, T. Jiang, D. Ba, P. Xu, H. Zhang, Z. Lu, and H. Li, “Slope-assisted BOTDA based on vector SBS and frequency-agile technique for wide-strain-range dynamic measurements,” Opt. Express 25(3), 1889–1902 (2017). [CrossRef]
11. A. Motil, R. Davidi, R. Hadar, and M. Tur, “Mitigating the effects of the gain-dependence of the Brillouin line-shape on dynamic BOTDA sensing methods,” Opt. Express 25(19), 22206–22218 (2017). [CrossRef]
12. G. Yang, X. Fan, B. Wang, and Z. He, “Enhancing strain dynamic range of slope-assisted BOTDA by manipulating Brillouin gain spectrum shape,” Opt. Express 26(25), 32599–32607 (2018). [CrossRef]
13. H. Zheng, D. Feng, J. Zhang, T. Zhu, Y. Bai, D. Qu, X. Huang, and F. Qiu, “Distributed vibration measurement based on a coherent multi-slope-assisted BOTDA with a large dynamic range,” Opt. Lett. 44(5), 1245–1248 (2019). [CrossRef]
14. C. Feng, H. D. Bhatta, J. Bohbot, R. Davidi, X. Lu, T. Schneider, and M. Tur, “Gain Spectrum Engineering in Slope-Assisted Dynamic Brillouin Optical Time-Domain Analysis,” J. Lightwave Technol. 38(24), 6967–6975 (2020). [CrossRef]
15. A. Voskoboinik, J. Wang, B. Shamee, S. Nuccio, L. Zhang, M. Chitgarha, A. Willner, and M. Tur, “SBS-based fiber optical sensing using frequency-domain simultaneous tone interrogation,” J. Lightwave Technol. 29(11), 1729–1735 (2011). [CrossRef]
16. X. Jia, H. Chang, K. Lin, C. Xu, and J. Wu, “Frequency-comb-based BOTDA sensors for high-spatial-resolution/long-distance sensing,” Opt. Express 25(6), 6997–7007 (2017). [CrossRef]
17. C. Jin, L. Wang, Y. Chen, N. Guo, W. Chung, H. Au, Z. Li, H. Tam, and C. Lu, “Single-measurement digital optical frequency comb-based phase-detection Brillouin optical time domain analyzer,” Opt. Express 25(8), 9213 (2017). [CrossRef]
18. Z. Liang, J. Pan, S. Gao, Q. Sui, Y. Feng, F. Li, J. Li, W. Liu, and Z. Li, “Spatial resolution improvement of single-shot digital optical frequency comb-based Brillouin optical time domain analysis utilizing multiple pump pulses,” Opt. Lett. 43(15), 3534–3537 (2018). [CrossRef]
19. D. Zhou, Y. Dong, B. Wang, C. Pang, D. Ba, H. Zhang, Z. Lu, H. Li, and X. Bao, “Single-shot BOTDA based on an optical chirp chain probe wave for distributed ultrafast measurement,” Light: Sci. Appl. 7(1), 32 (2018). [CrossRef]
20. Y. Dong, B. Wang, C. Pang, D. Zhou, D. Ba, H. Zhang, and X. Bao, “150 km fast BOTDA based on the optical chirp chain probe wave and Brillouin loss scheme,” Opt. Lett. 43(19), 4679–4682 (2018). [CrossRef]
21. B. Wang, B. Fan, D. Zhou, C. Pang, Y. Li, D. Ba, and Y. Dong, “High-performance optical chirp chain BOTDA by using a pattern recognition algorithm and the differential pulse-width pair technique,” Photonics Res. 7(6), 652–658 (2019). [CrossRef]
22. D. Ba, B. Wang, T. Li, Y. Li, D. Zhou, and Y. Dong, “Fast Brillouin optical time-domain reflectometry using the optical chirp chain reference wave,” Opt. Lett. 45(19), 5460–5463 (2020). [CrossRef]
23. S. Wang, Z. Yang, M. A. Soto, and L. Thévenaz, “Study on the signal-to-noise ratio of Brillouin optical-time domain analyzers,” Opt. Express 28(14), 19864–19876 (2020). [CrossRef]
24. Z. Yang, Z. Li, S. Zaslawski, L. Thévenaz, and M. A. Soto, “Design rules for optimizing unipolar coded Brillouin optical time-domain analyzers,” Opt. Express 26(13), 16505–16523 (2018). [CrossRef]
25. H. Zheng, J. Zhang, T. Zhu, G. Yin, Y. Bai, D. Qu, X. Huang, and F. Qiu, “Polarization independent fast BOTDA based on pump frequency modulation and cyclic coding,” Opt. Express 26(14), 18270–18278 (2018). [CrossRef]
26. M. Farahani, M. Wylie, E. Castillo-Guerra, and B. Colpitts, “Reduction in the number of averages required in BOTDA sensors using wavelet denoising techniques,” J. Lightwave Technol. 30(8), 1134–1142 (2012). [CrossRef]
27. Y. Zhou, L. Yan, H. He, Z. Li, P. Jiang, X. Zhang, W. Pan, and B. Luo, “Brillouin optical time domain collider for fast dynamic sensing,” Opt. Express 28(3), 3965–3974 (2020). [CrossRef]
28. Y. Dong, L. Chen, and X. Bao, “Time-division multiplexing-based BOTDA over 100 km sensing length,” Opt. Lett. 36(2), 277–279 (2011). [CrossRef]
29. K. Y. Ko, M. S. Demokan, and H. Y. Tam, “Transient Analysis of Erbium-Doped Fiber Amplifiers,” IEEE Photonics Technol. Lett. 6(12), 1436–1438 (1994). [CrossRef]
30. H. Iribas, J. Mariñelarena, C. Feng, J. Urricelqui, T. Schneider, and A. Loayssa, “Effects of pump pulse extinction ratio in Brillouin optical time-domain analysis sensors,” Opt. Express 25(22), 27896–27912 (2017). [CrossRef]
31. C. Du, S. Dutta, P. Kurup, T. Yu, and X. W. Wang, “A review of railway infrastructure monitoring using fiber optic sensors,” Sens. Actuators, A 303, 111728 (2020). [CrossRef]
32. L. Yan, Z. Zhang, P. Wang, W. Pan, L. Guo, B. Luo, K. Wen, S. Wang, and G. Zhao, “Fiber Sensors for Strain Measurements and Axle Counting in High-Speed Railway Applications,” IEEE Sens. J. 11(7), 1587–1594 (2011). [CrossRef]
33. X. Bian, H. Jiang, C. Chang, J. Hu, and Y. Chen, “Track and ground vibrations generated by high-speed train running on ballastless railway with excitation of vertical track irregularities,” Soil Dyn. Earthq. Eng. 76, 29–43 (2015). [CrossRef]
34. P. Hu, C. Zhang, S. Chen, Y. Wang, W. Wang, and W. Duan, “Dynamic responses of bridge–embankment transitions in high-speed railway: Field tests and data analyses,” Eng. Struct. 175, 565–576 (2018). [CrossRef]
35. M. I. Jordan and T. M. Mitchell, “Machine learning: Trends, perspectives, and prospects,” Science 349(6245), 255–260 (2015). [CrossRef]