1. Introduction

Optical sensing techniques based on Brillouin scattering have been widely investigated over decades. The Brillouin optical time domain analysis (BOTDA) is one of the most attractive techniques, due to its capability of distributed temperature and strain measurement with meter or even sub-meter scale spatial resolution over tens of kilometers [1–3]. Generally the Brillouin frequency shift (BFS) of the single mode optical fiber is dependent on temperature and strain over very wide range [4]. For years, BOTDA has reached a typical temperature sensitivity of about 1.10 MHz/°C and a strain sensitivity of about 0.05 MHz/με, by employing standard single mode fiber (SSMF) [5, 6]. Assuming a tolerable but not high signal to noise ratio (SNR) of 6 dB at the far end of a sensing fiber in a long range Brillouin distributed sensor, besides considering the full width at half maximum (FWHM) of Brillouin gain spectrum (tens of MHz) and the sweep step of no less than 1 MHz, the measurement accuracy of BFS is estimated to be about 1 MHz [7], corresponding to a temperature resolution of about 1 °C or a strain resolution of 20 με [8, 9]. As can be seen, despite the remarkable performance of BOTDA sensors, the poor temperature/strain measurement resolution intrinsically hinders its successful implementation of sensing applications if high temperature/strain resolution is required. Although the measurement accuracy for a given sensing length has been increased through improving the SNR of system by ways of multi-wavelength coherent detection, pulse coding and distributed Raman amplification, the temperature resolution is still limited to sub degree centigrade [10–12]. Besides, it has been reported that better temperature resolution is achieved by higher temperature coefficient. A temperature coefficient of −4.09 MHz/°C is obtained by employing perfluorinated graded-index polymer optical fiber (PFGI-POF), which is about 3.5 times of that in comparison with the silica SMF [13]. However, the sensing range is limited by the low Stokes power because of the core diameter mismatch between the SSMF and PFGI-POF. An even larger temperature coefficient is acquired based on higher-order Stokes waves [14], reaching a temperature sensitivity of 7 MHz/°C. But it suffers from the lower power of high order Stokes wave and the complexity of setup.

On the other hand, a specific phase-sensitive optical time-domain reflectometry (Φ-OTDR) sensor has been demonstrated to offer ultrahigh temperature and strain resolutions, which are about 0.01 °C and 0.1 με respectively and even higher [8, 9, 15]. The technique retrieves the temperature and strain variations by scanning the wavelength of pump light and performing cross-correlation calculation between the achieved response and a reference. Despite its ultrahigh temperature and strain sensitivity, e.g. 0.1 °C temperature change leads to about 134 MHz laser frequency shift [8], the dynamic range is considerably restricted to a typical range from hundreds of megahertz to several gigahertz. Therefore, the measurement range of the temperature/strain is within the magnitude of 1 °C / 10 με, respectively. Apparently, this technique turns out to be particularly desired in occasions with high requirement of measurement resolution, but unsuitable for large dynamic range measurement.

Recently, the multicore fiber (MCF) has been proposed and demonstrated for distributed temperature and strain sensing [16, 17], showing great advantages in terms of spatial multiplexing capability that SMF cannot offer. Taking advantage of the redundancy in spatial channels offered by MCF, in this paper, we propose to combine the Φ-OTDR with BOTDA through space-division multiplexed (SDM) configuration to realize distributed sensing with high measurement resolution and large dynamic range simultaneously. Specifically, the BOTDA is dedicated to provide a large dynamic range, while the Φ-OTDR is deployed to offer the high measurement resolution. In this case, the two sensors complement each other with their own advantages. The hybrid system requires only one laser diode to generate pump pulse shared by the two sensors. However due to the distinct frequency scanning step demands, the frequency sweeping operations are performed on different modulators. Eventually, BOTDA and Φ-OTDR are implemented in different cores of the MCF. In order to validate the proposed concept, a distributed temperature sensing has been conducted with 2.5 m spatial resolution over the 1.565 km MCF. The experimental results confirmed that the large temperature change can be detected by the BOTDA sensor and at the same time the 0.1 °C small temperature change could be identified through the Φ-OTDR with about 0.001 °C uncertainty. The proposed system provides a feasible solution and will have great potential in a wide range applications.

2. Measurement principle

The BFS in optical fibers is determined by the effective refractive index of the guided mode $n_{eff}$, the acoustic velocity $V_a$ and the wavelength of pump light $\lambda$, given by:

Due to the fact that both the $n_{eff}$ and $V_a$ are sensitive to temperature and stain, the BFS change shows the dependence on temperature and strain over a very wide range. Any change of temperature and/or strain leads to the shift of BFS with the relationship given by:

While for the Φ-OTDR, the temperature/strain induced phase change can be compensated by varying the pump laser frequency, thus providing high sensitivity and high measurement resolution [8]. The sensor launches coherent optical pulse with a duration of $W$and an optical frequency of $\nu$ into the sensing fiber at a specific time slot. In the 1-dimensional backscattering impulse-response model, the measured backscattered optical power $P(t)$ actually consists of two components as presented by [8]:

The Φ-OTDR sensor manages to retrieve small temperature/strain variation by shifting the laser frequency to match the phase difference change that is caused by temperature/strain change on the sensing fiber. This is done by launching a set of pulses periodically with its optical frequency scanned step by step. The acquired traces are cross-correlated in frequency with a reference, and then the frequency shift of the correlation peak is located to reveal the change of temperature/strain. Specifically, at an initial temperature $T_a$, the back scattered power is recorded as $P_a({\nu }_i,z)$$(i=1,\mathrm{2...}N)$, which is a function of the scanned pump frequency ${\nu }_i$ and fiber distance $z$. Then at another moment, the backscattered power $P_b({\nu }_i,z)$ at temperature $T_b$ is also recorded. The cross-correlation $R_{T_a{T}_b}(f,z)$ between $P_a({\nu }_i,z)$ and $P_b({\nu }_i,z)$ will indicate the phase change caused by temperature variation within the time slot. The $R_{T_a{T}_b}(f,z)$ is calculated as follows:

3. Experimental setup

The experimental setup is schematically illustrated in Fig. 1. The optical source is a distributed feedback laser diode (DFB-LD) operating at 1550 nm with long-term stability and a linewidth below 100 KHz. The CW light from the LD is divided by a coupler (97:3) into two branches as pump and probe respectively, with the probe used only for BOTDA. The high extinction ratio pump pulse generated through a semiconductor optical amplifier (SOA) in the upper branch is amplified by the Erbium doped fiber amplifier (EDFA) and then split into two paths by a 3 dB coupler for the BOTDA and Φ-OTDR. In this case, only one laser diode is used in the system. Moreover, the optimized configuration by using only one set of pulse generation devices allows for the two sensors to employ pulses with the same width of 25 ns to ensure consistent 2.5 m spatial resolution. The upper one of the two paths is assigned to implement Φ-OTDR, in which, the amplified coherent pulsed light is modulated by a Mach-Zehnder modulator (MZM 1) driven by the microwave generator to precisely shift the optical frequency. In order to launch precisely a single frequency pulse into the fiber, a tunable narrow band filter (Filter 1) with about 7 GHz bandwidth is used to filter out the unwanted sidebands, the residual carrier component and the spontaneous emission noise (ASE) of EDFA. Finally, the obtained sideband pulse is launched into an outer core through a circulator and the fan-in coupler. The average insertion loss of the fan-in/out coupler is about 1 dB. At its receiver side, a photodetector (PD) with 125 MHz bandwidth is utilized for the backscattered Rayleigh signal detection.

 

Fig. 1 Experimental setup of the MCF based SDM hybrid BOTDA and Φ-OTDR system. LD: Laser diode; PC: polarization controller; SOA: semiconductor optical amplifier; MZM: Mach-Zehnder modulator; EDFA: erbium-doped fiber amplifier; PS: polarization switch; Att.: attenuator; PD: photodetector; Fan-in: fan-in coupler; Fan-out: fan-out coupler.

Download Full Size | PDF

The lower path is used to implement BOTDA with similar configuration to that of Φ-OTDR. The pulse is also firstly modulated by a Mach-Zehnder modulator (MZM 2) driven by a microwave generator with a frequency that is approximately equal to the BFS ${\nu }_B$ of the fiber under test (FUT). The lower modulation sideband (shorter wavelength) is then remained after a tunable narrow band filter (Filter 2) as the pump for BOTDA, which is injected into the central core of the MCF through a circulator and the fan-in coupler. It should be noted that the two similar paths cannot be simplified into one, because of the inconsistent frequency scanning steps and ranges of the two techniques. The temperature sensitivity of SSMF in traditional BOTDA sensors is approximately 1 MHz/ °C. To measure a temperature change of 1 °C, the frequency sweep range of Φ-OTDR system should be larger than 1 GHz, which is much greater than that of BOTDA sensor (typically in the range of several hundred megahertz). On the other hand, the frequency sweep step for BOTDA is normally no more than 4 MHz, e.g. typically 1 MHz or 2 MHz; while the step for Φ-OTDR is much larger, typically 10 MHz, 15 MHz or 20 MHz. Since the frequency scanning step and range of Φ-OTDR are larger than that of BOTDA, we have to perform the frequency modulation for the two sensors in different paths. In the probe branch of BOTDA, a polarization switch (PS) is employed before the fan-out coupler to alleviate polarization dependent fluctuations of Brillouin gain. Finally, the amplified probe signal is filtered by the FBG filter and detected by another photodetector with 125 MHz bandwidth. The peak pump powers for BOTDA and Φ-OTDR are about 20 dBm and 23 dBm, respectively. The probe power for the BOTDA is about −15 dBm.

A 1.565 km single mode 7-core fiber with ~0.25 dB/km attenuation coefficient has been employed in our experiments for proof of concept. The MCF has six side cores and a central core, with 150 μm cladding diameter and 42 μm pitch. The cross section of the MCF is shown in the inset of Fig. 1, where the six outer cores are symmetrically arranged as hexagon. All the cores are surrounded by deep trench. As a result, the crosstalk between adjacent cores has been suppressed to be as low as −45 dB/100km. In the system, the central core and one of the outer cores have been used for BOTDA and Φ-OTDR respectively, which could not only avoid cross-interaction that is imposed by probe light between the two sensors but also allows for simultaneous implementation of both measurements.

4. Experimental results

In order to determine the temperature sensitivity in BOTDA, about 40 m long fiber at the far end is heated in a water bath from 31 °C to 81 °C with a step of 10 °C. The frequency is scanned from 10.6 GHz to 10.8 GHz with a step of 2 MHz. The measured BFS distribution around the hot-spot with different temperatures is presented in Fig. 2(a). The temperature sensitivity of BFS is linearly fitted to be 1.073 MHz/ °C, as shown in Fig. 2(b).

 

Fig. 2 The calibration of temperature sensitivity. (a) The measured BFS distribution near the hot-spot at the far end of the sensing fiber with different temperatures. (b) The peak frequency shift of BGS as a function of temperature.

Download Full Size | PDF

Despite the large dynamic range of measurement provided by Brillouin distributed sensors, this kind of sensors are however suffering from the bad measurement resolution, especially for long range sensing (typically around 1 °C temperature uncertainty [9]). As a consequence, these sensors are not able to perceive small temperature/strain variation. The drawback has dramatically degraded the performance of Brillouin distributed sensors. On the other hand, the Φ-OTDR based on single mode fiber at 1550 nm has been demonstrated to offer about 1.339 GHz/°C temperature sensitivity. Thanks to the ultrahigh sensitivity, with 2 to 4 MHz frequency shift uncertainty of the cross-correlation peak, the temperature resolution of Φ-OTDR can reach milliKelvin magnitude [15]. This will be very helpful for identifying small temperature changes.

In order to evaluate the performance of the proposed system, two short segments, i.e. section A and section B (see Fig. 1) with length about 37 m at the far end are separately immersed in two water baths to apply different temperatures. Meanwhile a mercurial thermometer is used to monitor the temperature of water for real time calibration. The two heated sections are assigned in two separate water bathes with different initial temperatures. It should be noted that the residual fiber is immersed in the water to ensure stability of the system. In addition, the temperature of water during experiments is read out from the mercurial thermometer with 0.1 °C resolution. In the BOTDA-path, the frequency is scanned from 10.6 GHz to 10.85 GHz with a step of 2 MHz. Each trace has been averaged for 1024 times to improve the SNR. The frequency scanning process takes about 170 s. The measured Brillouin gain spectrum distribution along the fiber is shown in Fig. 3(a) when the fiber of section A and section B are heated. The enlarged view around the hot-spots is presented in Fig. 3(b), where temperature change (compared with the residual fiber) induced BFS shifts can be clearly observed for the two fiber sections. At the same time, the frequency is scanned from 10 GHz to 11 GHz with a step of 10 MHz in Φ-OTDR. Each OTDR curve has been averaged 512 times. The cross-correlation mapping of $R_{T_a{T}_b}(f,z)$ between $P_a({\nu }_i,z)$and $P_b({\nu }_i,z)$ with 0.1 °C and 25 °C temperature change for fiber section A and B respectively is presented in Fig. 3(c). The distribution of frequency shift around the hot-spots is shown in Fig. 3(d), where about 150 MHz frequency shift is distinguished clearly corresponding to 0.1 °C temperature change read by the mercurial thermometer. The slight difference between the calculated value (134 MHz) according to Eq. (9) and the measured value (150 MHz) may be caused by the reading error of thermometer. On the other hand, no regular cross-correlation peak can be found at fiber section B, noting that the messy peak fitted points are actually the fitting on noise, as shown in the inset of Fig. 3(d), therefore the fitted peak trace in this region is invalid and not reliable. This is because the frequency scanning range for Φ-OTDR (1 GHz) is not wide enough to compensate the refractive index variation of 25 °C temperature change.

 

Fig. 3 (a) The measured BGS along the whole fiber when the fiber section A and B are heated and (b) is the enlarged view around the hot-spots. (c) The calculated$R_{T_a{T}_b}(f,z)$ of temperature change for 0.1 °C of section A and 25 °C of section B along the fiber and (d) is the partial magnification of the $R_{T_a{T}_b}(f,z)$ around the hot spots.

Download Full Size | PDF

For better evaluation, the repeated simultaneous measurement results of both the BOTDA sensor and the Φ-OTDR sensor under different temperatures have been presented in Fig. 4, in which the fiber section A undergoes a natural cooling down process and fiber section B is heated. It should be mentioned that the natural cooling down process is quite slow since the water temperature is close to the ambient room temperature, so there is enough time to change and stabilize the temperature of fiber section B. As can be seen in Fig. 4(a), 30 °C temperature increment with a step of 10 °C at fiber section B from 1521 m to 1558 m has been successfully detected by BOTDA sensor. However, it is found that the BOTDA sensor is unable to distinguish the small temperature change, since the three traces obtained with 0.1 °C temperature difference between adjacent measurements are nearly overlapped at fiber section A from 1442m to 1479 m. On the contrary, the temperature decrement of about 0.3 °C with a step of 0.1 °C at fiber section A has been successfully identified by Φ-OTDR sensor as shown in Fig. 4(b). The slight difference between the Φ-OTDR retrieved temperature traces and the read values from the thermometer is supposed to be caused by the reading error of the mercurial thermometer. But for large temperature variation (10 °C in this case) at fiber section B, since the temperature induced phase change has been out of the range that the laser frequency scanning (1 GHz) can compensate for, the Φ-OTDR sensor is unable to measure these events. Note that in Fig. 4(b) the traces at the region of fiber section B are actually invalid fitting on the noise, as has been shown in Figs. 3(c) and 3(d). While the large temperature change has already been measured by the complementary BOTDA sensor, as shown in Fig. 4(a). The experiments demonstrate the good feasibility of carrying out SDM configuration that combines the BOTDA with Φ-OTDR sensor, in which large measurement dynamic range and high resolution are available simultaneously.

 

Fig. 4 (a) The measured absolute temperature distribution based on BOTDA. The inset shows an enlarged view around fiber section A; (b) The measured temperature change distribution based on Φ-OTDR.

Download Full Size | PDF

Additionally, in order to assess the ability of small temperature discrimination, also to monitor the tendency of the natural cooling down process, continuous frequency scanning is performed for Φ-OTDR sensor. The frequency scanning range is 1 GHz with 80 s period. The obtained temperature distribution around fiber section A as a function of time and distance has been shown in Fig. 5(a). The 80 s period yields about 0.02 °C temperature spacing at the monitored profile at fiber section A. The monitored temperature tendency of the cooling down process is shown in Fig. 5(b). The red dots are the average temperature of the fiber section A monitored at different time. The red lines are the calculated standard deviation of the measured temperature, which increases gradually with time due to the decreased stability of the system over longer period.

 

Fig. 5 (a) The continuous monitoring of temperature around fiber section A as a function of time and distance. (b) The monitored temperature tendency of the nature cooling down process of water.

Download Full Size | PDF

The temperature accuracy of BOTDA can be estimated by calculating the BFS uncertainty, which is obtained by computing the standard deviation of BFS [7]. The acquired temperature resolution along the whole fiber length is presented in Fig. 6(a) with the worst uncertainty of ~0.25 °C at the fiber end. The obtained temperature resolution here is much better that the 1 °C typical value in long range sensing, that’s because the used FUT is only 1.565 km, which ensures sufficient SNR for the measurement. As a matter of fact, extending the fiber length will considerably degrade the SNR and increase the uncertainty. On the other hand, by calculating the standard deviation of the cross-correlation peak frequency shift with a window of 1 m [15], about 0.001 °C temperature resolution has been obtained for the Φ-OTDR sensor as shown in Fig. 6(b), which is two orders of magnitude lower than that of the BOTDA sensor. The several unusual high peaks in Fig. 6(b) are caused by fitting error in cross-correlation peak. The results confirm again that the Φ-OTDR sensor can provide much higher measurement resolution than the BOTDA sensor. So, the introduction of Φ-OTDR will help to address the small temperature change identification issue in the BOTDA sensor. Thanks to the space-division multiplexed hybrid system, both large dynamic range and high measurement resolution are obtained.

 

Fig. 6 The estimated temperature uncertainty along the fiber of (a) BOTDA sensor and (b) Φ-OTDR sensor.

Download Full Size | PDF

It should be pointed out that both the BOTDA and Φ-OTDR are sensitive to temperature and strain. In order to address the cross-sensitive issue, potential solutions include employing the combination of the presented Φ-OTDR and Φ-OTDR based birefringence measurement to separate temperature and strain in the case of small temperature/strain variation [18]; While in the case of large temperature/strain change, one can extend the frequency scanning range of Φ-OTDR and then solve a coefficient matrix by using the measurement results of BOTDA and Φ-OTDR for strain and temperature discrimination. Additionally, the hybrid ROTDR and BOTDR configuration, or multiplexed BOTDA in heterogeneous MCF can also be used for discrimination between temperature and strain [16, 17].

5. Conclusions

We proposed and experimentally demonstrated a large dynamic range and ultrahigh measurement resolution sensing system by combining both BOTDA and Φ-OTDR through SDM configuration based on MCF. The hybrid system contains respective advantages, which is efficiently integrated by the unique SDM transmission link. Temperature sensing has been performed for validation with 2.5 m spatial resolution over 1.565 km MCF. Large temperature dynamic range and high measurement resolution are achieved by BOTDA sensor and Φ-OTDR sensor, respectively. Particularly, about 0.001 °C temperature resolution has been obtained for the Φ-OTDR sensor in the hybrid system. Thanks to the ultrahigh temperature resolution offered by Φ-OTDR, the natural cooling down process of water has been successfully recorded with around 0.02 °C spacing by using 80 s frequency scanning period. In comparison with the multiple SMFs packaged fiber cable, the benefit of multicore fiber is to ensure that every core at the same position is undergoing strictly identical longitudinal strain and temperature. The proposed system enables flexible access to see finer and/or farther upon requirement in distributed optical fiber sensing.