The leakage light of optical pulses due to finite extinction ratio (ER) of an electro-optic modulator (EOM) leads to Rayleigh backscattered noises over the entire fiber length, and limits spatial resolution and sensing range in phase-sensitive optical time-domain reflectometry (Φ-OTDR). Two configurations are proposed to improve the ER of optical pulses for better spatial resolution over long sensing length. With ER of 55 dB using a nonlinear optical loop mirror, we achieved 2 m spatial resolution over 8.4 km sensing length; while with ER of 60 dB obtained by two cascaded EOMs, we can achieve a 1 m spatial resolution over the same range. Experimental results and analysis show that leakage of the optical pulses acts as a noise floor, which limits the highest spatial resolution over the same sensing range.
© 2016 Optical Society of America
Phase-sensitive optical time-domain reflectometry (Φ-OTDR) is a promising technique for vibration measurements because of its advantages of high sensitivity, large frequency response and simple configuration. Φ-OTDR was initially demonstrated by H. F. Taylor et. al. in 1993 as a distributed intrusion sensor . Then the sensor has been applied in field tests to detect human walking on the buried fiber [2,3]. In order to fulfill in most applications, high spatial resolution, long sensing range, and high signal-to-noise ratio (SNR) are desired for a distributed sensing system. Different schemes have been proposed to improve the performance of Φ-OTDR for dynamic measurements, such as using coherent detection method  and combing a Mach-Zehnder interferometer . Besides, signal processing methods such as wavelet denoising , and two dimensional edge detection  were adopted in Φ-OTDR to enhance SNR and spatial resolution.
Φ-OTDR utilizes a highly coherent laser source, and its Rayleigh backscattering signal from an optical fiber shows zig-zag pattern due to the interference of the scattered light within a pulse width. Since the interference signals are sensitive to external disturbances, intrusion or vibration can be detected by monitoring the intensity change of Rayleigh backscattering. Generally, optical pulses can be generated by using acousto-optic modulators (AOMs) or Electro-optic modulators (EOMs). AOMs provide higher ER and therefore are preferred in a distributed optical fiber sensor (DOFS) to achieve high sensitivity [7,8]. However, the generated pulse durations are usually greater than 20 ns which is limited by the acoustic wave decay time of 10 ns . For the applications that require high spatial resolution, EOMs are used to create the probe pulses with durations less than 20 ns [9–11]. Since ER of the standard telecom EOMs are usually limited to ~20 dB, much higher price has to be paid to select EOMs with higher ER (30-40 dB) in order to have decent SNR. In Φ-OTDR, lower ER can lead to higher power of continuous wave (CW) background light between optical pulses which result in higher Rayleigh backscattered noise that decreases the SNR and sensing accuracy. This unwanted Rayleigh backscattered noise varies with fiber length, and it adds considerable amount of trace-to-trace fluctuations due to different input state of polarization (SOP), which sets the noise limit for minimum pulse to be used for the spatial resolution limit at different fiber lengths. Although the influences of finite ER on performance of Brillouin based DOFS have been extensively studied in the literature [12–14], this effect has not been thoroughly discussed in Φ-OTDR.
In this paper, we demonstrate the influences of optical pulses with different ERs on performances of an Φ-OTDR system, especially on spatial resolutions with the same sensing range. Trace fluctuations with lower ER of EOM can be well observed, which limits the spatial resolution for long range measurement. Two methods of enhancing ER of optical pulses are then adopted in an Φ-OTDR system to improve the measurement performance. A 55 dB ER can be achieved by using a nonlinear optical loop mirror (NOLM) because of the switching effect [15, 16]. The vibration measurement can be achieved with a spatial resolution of 2 m over the range of 8.4 km. The ER can be further improved to 60 dB using two cascaded EOMs with well synchronized pulses, and spatial resolution can be further enhanced to 1 m with the same measurement length.
2. Operation principle
2.1. Signal-to-noise ratio
In Φ-OTDR, an EOM is used to convert CW light into optical pulses. Although the bias voltage is adjusted at the minimum point of the transfer function of an EOM, the CW leakage cannot be totally suppressed. Figure 1 shows typical optical pulses generated by an EOM with finite ER. Pp and Pcw are the peak and CW leakage powers of optical pulses. In an ideal case, Pcw = 0 where ER is infinite. So the power of Rayleigh backscattered light detected by a photo-detector at position z along the fiber can be defined as:Eq. (1) and behaves as an exponential decay along the fiber length. The CW leakage induced Rayleigh backscattered light is increased with the fiber length, and its power is expressed as:
Due to the highly coherent laser source used in Φ-OTDR systems, the backscattered light from CW portion will interfere with the pulse portion. This interference component is also added onto the received signal and the power of interference component, Pint (z) is estimated to beEq. (3) as the power Pcw is normally weak. The SNR can be approximated as:Eq. (4) represents almost the lowest SNR in the Φ-OTDR systems due to the finite ER of the optical pulses, because we consider fully constructive interference of the pulse and CW leakage in Eq. (3). Figures 2(a) and 2(b) show the SNR as a function of fiber length z with different ER values under the pulse width of 20 ns and 10 ns, respectively for 8.4 km sensing length to be consistent with the experimental results. The calculation started with ER of 50 dB and its SNR is below 3 dB for the pulse with of 10 ns, which is equivalent to 1m spatial resolution. Thus, the high ER optical pulse is essential for vibration sensing with high spatial resolution over long fiber length in the Φ-OTDR.
2.2. High ER pulse generation using NOLM
The high ER pulse generation is first achieved by combing a NOLM with an EOM. The configuration of proposed NOLM is shown in Fig. 3. The NOLM consists of a 50: 50 optical coupler, a 500 m single mode fiber (SMF) and a 1.2 m highly doped erbium doped fiber (EDF). The loop is asymmetric and L2 is assumed to be much longer than L1. The 980 nm pump laser is coupled to the EDF through a wavelength-division multiplexer (WDM). A polarization controller is used to bias the loop to the minimum transmission at low input power.
An ideal linear optical loop mirror with 50:50 optical coupler could provide full reflection and zero transmission. However with a gain medium in the NOLM, the nonlinear phase difference is induced through asymmetrical amplification of clockwise and counter-clockwise propagating light. The transmission function of an NOLM with 50:50 optical coupler is expressed as :Eq. (5), the transmission function of an NOLM depends on the input power P0, gain factor g and fiber length L used in the loop (L2 is assumed to be much longer than L1, so the loop length L is used in Eq. (5)). By optimizing g and L, the NOLM can operate as an optical switch which only allows the high input power to transmit. If the input light is an optical pulse, the CW light will be suppressed through an NOLM and therefore the ER of the optical pulse is improved. However, the transmission is also affected by the amplified Rayleigh backscattering in the fiber loop. This limits the highest ER obtained by using NOLM. In order to minimize the influence of Rayleigh backscattering, the fiber length and the gain in our experiment are both optimized to get the minimum CW transmission and the desired pulse peak power.
2.3. High ER pulse generation using two cascaded EOMs
High ER pulse generation can also be achieved by using two cascaded EOMs. In the two cascaded EOMs setup, the first EOM is used to modulate CW light into optical pulses with a finite ER determined by the property of this EOM. When the resultant pulse passes the second EOM, synchronization is achieved to allow the pulse transmitted to further improve ER. In this case, the second EOM can be considered as an additional gate, further block the unwanted background light between optical pulses. The ER value obtained by two cascaded EOMs is determined by the individual ER of EOM.
3. Experimental setup and signal processing method
The experimental setup of Φ-OTDR based on the conventional and the high ER generation configurations are shown in Fig. 4. A laser source used in all cases is a highly coherent laser operated at 1550 nm with narrow linewidth of 3 kHz. The CW light is injected into an EOM which is controlled by a pulse generator and a bias controller. Since the bias drift of the EOM can affect the ER of the optical pulse, the bias controller is locked at the minimum point of transfer function of the EOM. An electric pulse pattern is also applied on the RF port of the EOM to modulate the CW light. In the conventional Φ-OTDR setup, optical pulses after the EOM are amplified by an erbium doped fiber amplifier (EDFA) as shown in Fig. 4 as (a) configuration. In the case of high ER generation as (b) configuration, the proposed NOLM not only amplify the pulse portion but also suppress the CW light. An isolator is used to prevent the reflected light of NOLM from injecting into the EOM. In the third configuration (c), a second EOM is cascaded and synchronized with respect to the first EOM. Optical pulses are also amplified by an EDFA after emitting from the second EOM. For all the three cases, amplified optical pulses are sent into the fiber under test (FUT) via a circulator and the Rayleigh backscattered light is further amplified by an EDFA and passed through a filter. Then the filtered signal is detected by a photodetector and collected by a data acquisition (DAQ) card. Both filters are used to reduce the amplified spontaneous emission (ASE) noises generated by EDFAs.
The wavelet denoising method  is also adopted in the signal processing to remove the random Rayleigh noise and improve the vibration frequency response. The discrete-time wavelet transform (DWT) is applied to the raw Φ-OTDR traces to acquire wavelet coefficients and an appropriate threshold is chosen to obtain new wavelet coefficients. The denoised signal can thus be reconstructed by the inverse DWT.
4. Results and discussion
We firstly evaluate the performance of NOLM for suppressing the CW leakage light, and the CW portion is measured when the pulse is off. The CW light in conventional configuration and NOLM are shown in red and blue curves in Fig. 5(a), respectively. Both curves are measured after the filter before the circulator by a low noise and high gain photodetector with 125 MHz bandwidth. The signals are then acquired by an oscilloscope with 500M/s sampling rate. The photodetector intrinsic noise is also shown as black curve in Fig. 5(a). The photodetector is calibrated and the measured electric voltages are converted to the optical powers. The mean powers of CW leakage light and the standard deviations in three cases are provided in Table 1. The results clearly show that CW leakage power is effectively reduced from 1.1 to 0.09 µW by using NOLM corresponding to a ~11 dB improvement, indicating that NOLM is effective for suppressing CW light. Figure 5(b) shows the amplitude spectrum of three time domain signals in Fig. 5(a). In the low frequency range, there are more power noises for the conventional single EOM case which attributes to the laser intensity fluctuations and EOM bias control. The amplitude of DC component in NOLM case is ~11 dB lower than conventional EOM, and ~4 dB higher than the noise floor of photodetector. When the pulse is on, the pulse peak power in both cases have the same value of 56 mW from an input power of P0 = 10 mW. Thus the optical pulse is amplified by T = 5.6 times, which approximately corresponds to the gain g = 97, where γ = 2 W−1km−1, and L = 500 m are used in Eq. (5). Thus, ER of optical pulse is improved from 44 dB to 55 dB with NOLM configuration.
Figure 6 shows the 5 adjacent Φ-OTDR time domain traces obtained by using the conventional setup and NOLM under the pulse width of 20 ns and 10 ns. In our experiment, 8.4 km fiber is used and the traces are recorded by a high speed oscilloscope with sampling rate of 250 M/s. The pulse repetition rate is set as 5 kHz to ensure that only one pulse is travelling in the fiber at a time. The ripples between adjacent traces shown in Figs. 6(a) and 6(c) are caused by the interference of the CW leakage induced Rayleigh backscattering. In order to accurately detect the disturbances, the system requires the traces at non-disturbed locations exhibit the similar interference form. The CW backscattered noise could cause the fluctuation of time-domain traces and thus it decreases the stability and accuracy of sensor system. It is also noted that this effect is more obvious under the short pulse, which makes it difficult to achieve high spatial resolution. By suppressing the CW leakage light with NOLM, the ripples disappear in Figs. 6(b) and 6(d). It is verified that our proposed NOLM setup can successfully mitigate the effect of CW backscattered noise caused by the low ER pulse.
After realizing the high ER pulse generation, the vibration measurement is performed by using the proposed NOLM configuration. A section of fiber with the length of ~1 m is wrapped around piezoelectric (PZT) transducer which is inserted in between 7.2 km and 1.2 km fiber spools. The PZT cylinder is driven by a function generator to generate the vibrations. The fiber is fixed on the surface of PZT cylinder by instant glue and the vibration of PZT can be transmitted to the fiber. 500 normalized Rayleigh backscattering traces are collected in the experiment and processed with the wavelet denoising method, which is shown in Fig. 7(a). The zoomed in vibration position is shown in inset of Fig. 7(a). The superposition of these traces show the amplitude changes caused by vibration around 7.4 km location (circled in red dashed line). Figure 7(b) shows the normalized differential traces obtained by subtracting adjacent traces. The vibration location is evidently identified as a peak around 7.4 km in the superimposing differential signals. The detailed profile of vibration location is also presented in Fig. 7(c) and the full width at half maximum corresponds to 2 m spatial resolution of the sensor system and it is consistent with the pulse width of 20 ns. Figure 7(d) shows the power spectra obtained by fast Fourier transform (FFT) at vibration location, where vibration frequency of 1 kHz is indicated. The theoretical maximum frequency response is generally restricted by the repetition rate of the optical pulse according to the Nyquist sampling theorem and is 2.5 kHz in our case.
To further enhance the spatial resolution of Φ-OTDR, a configuration with two cascaded EOMs is proposed to generate ultra-high ER optical pulses. In this configuration, two EOMs with ER of 30 dB and 40 dB are cascaded and synchronized. The achieved ER is ~60 dB. Then the vibration measurement is performed based on this setup with the same experimental parameters except for the pulse width of 10 ns. Figure 8 shows the experimental results of the location and frequency information. Figure 8(a) shows the vibration location after traces subtraction. Figure 8(b) is zoomed in around the vibration position. 1 m spatial resolution is achieved with the vibration frequency of 2 kHz shown in Fig. 8(c), while using NOLM can only achieve 2 m spatial resolution since the ER is at least 5 dB lower.
It is observed that using the two cascaded EOMs setup achieves higher spatial resolution for the same sensing range compared with the NOLM case. It should be noted that the ER in two cascaded EOMs configuration strongly depends on the ER of individual EOM. The high ER generation is only achieved when both EOMs have the high ER values. The use of two EOMs with high ER is expensive, which limits its practical usage. The NOLM provides a low-cost way to obtain high ER pulse generation. Both proposed schemes effectively increase the SNR and sensitivity of the sensor and are suitable for long range and high spatial resolution sensing application. The sensing range could be further increased by applying the coherent detection based on our setup. The SNR and sensitivity of the sensor system would be also enhanced consequently.
We have investigated and verified the effect of the finite ER of an EOM on the performance of an Φ-OTDR system. Both theoretical and experimental results show that the CW leakage light between pulses induced by the finite ER will decrease the SNR and stability of the Φ-OTDR traces. Two approaches based on the NOLM and two cascaded EOMs have been demonstrated to improve the ER of an optical pulse. The distributed vibration measurements are achieved over 8.4 km fiber with a spatial resolution of 2 m and 1 m in NOLM and two cascaded EOMs configurations, respectively, as the latter case has higher ER of ~5 dB.
The research is supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grants and Canada Research Chairs (CRC) Program.
References and links
1. H. F. Taylor and C. E. Lee, “Apparatus and method for fiber optic intrusion sensing,” United States patent 5,194,847 (1993).
2. J. C. Juarez, E. W. Maier, K. N. Choi, and H. F. Taylor, “Distributed fiber-optic intrusion sensor system,” J. Lightwave Technol. 23(6), 2081–2087 (2005). [CrossRef]
4. Y. Lu, T. Zhu, L. Chen, and X. Bao, “Distributed vibration sensor based on coherent detection of phase-OTDR,” J. Lightwave Technol. 28(22), 3243–3249 (2010).
5. T. Zhu, Q. He, X. Xiao, and X. Bao, “Modulated pulses based distributed vibration sensing with high frequency response and spatial resolution,” Opt. Express 21(3), 2953–2963 (2013). [CrossRef] [PubMed]
6. Z. Qin, L. Chen, and X. Bao, “Wavelet denoising method for improving detection performance of distributed vibration sensor,” IEEE Photonics Technol. Lett. 24(7), 542–544 (2012). [CrossRef]
7. T. Zhu, X. Xiao, Q. He, and D. Diao, “Enhancement of SNR and spatial resolution of φ-OTDR system by using two-dimensional edge detection method,” J. Lightwave Technol. 31(17), 2851–2856 (2013). [CrossRef]
8. F. Peng, H. Wu, X. H. Jia, Y. J. Rao, Z. N. Wang, and Z. P. Peng, “Ultra-long high-sensitivity Φ-OTDR for high spatial resolution intrusion detection of pipelines,” Opt. Express 22(11), 13804–13810 (2014). [CrossRef] [PubMed]
10. A. Masoudi, M. Belal, and T. P. Newson, “A distributed optical fibre dynamic strain sensor based on phase-OTDR,” Meas. Sci. Technol. 24(8), 085204 (2013). [CrossRef]
11. X. Hui, S. Zheng, J. Zhou, C. Xu, H. Chi, X. Jin, and X. Zhang, “Electro-optic modulator feedback control in phase-sensitive optical time-domain reflectometer distributed sensor,” Appl. Opt. 52(35), 8581–8585 (2013). [CrossRef] [PubMed]
12. S. Afshar, G. A. Ferrier, X. Bao, and L. Chen, “Effect of the finite extinction ratio of an electro-optic modulator on the performance of distributed probe-pump Brillouin sensor systems,” Opt. Lett. 28(16), 1418–1420 (2003). [CrossRef] [PubMed]
13. Y. Lu, Y. Yao, X. Zhao, F. Wang, and X. Zhang, “Influence of non-perfect extinction ratio of electro-optic modulator on signal-to-noise ratio of BOTDR,” Opt. Commun. 297, 48–54 (2013). [CrossRef]
14. X. Wu, Z. Ying, Y. Zhang, and X. Zhang, “Performance improvement for long-range BOTDR sensing system based on high extinction ratio modulator,” Electron. Lett. 50(14), 1014–1016 (2014). [CrossRef]
17. B. E. Olsson and P. A. Andrekson, “Extinction ratio improvement using the nonlinear optical loop mirror,” IEEE Photonics Technol. Lett. 7(1), 120–122 (1995). [CrossRef]