Open Access
Add to BibSonomyAbstract
An approach for reducing chromatic dispersion (CD) induced Brillouin gain spectrum (BGS) distortion and measurement instabilities in coherent Brillouin optical time domain analysis (BOTDA) sensing systems is proposed and experimentally demonstrated. By utilizing intensity modulated probe (IMP) instead of phase modulated probe (PMP), sensing performance is obviously improved. Reduction of ~6-MHz decoding error caused by the CD induced BGS distortion is achieved in the measurement of Brillouin frequency shift (BFS) along the whole 40-km sensing distance. Enhanced system stabilities are demonstrated by testing the BGS under different conditions.
© 2015 Optical Society of America
1. Introduction
Brillouin optical time domain analysis (BOTDA) technology is a promising tool for distributed temperature and strain sensing, which has diversity of applications in structural health monitoring, geotechnical engineering and leakage detection along pipelines, etc [1]. The temperature/strain distribution along the sensing fiber turns out to be linearly related to the local Brillouin frequency shift (BFS), which can be obtained by utilizing a pulsed pump light and a frequency-sweeping counter-propagating continuous-wave (CW) probe light [2,3]. Recently, coherent BOTDA sensors have been proposed to increase signal-to-noise ratio (SNR), resulting in measurement accuracies improvement and measurement time reduction by utilizing sub-GHz carriers to carry the Brillouin gain spectrum (BGS) [4–9]. The probe light generated in coherent BOTDA sensors until now are usually based on phase modulation (PM), which converts into intensity modulation (IM) in the presence of Brillouin gain/loss, and then the BGS can be reconstructed [4–9]. However, our recent investigations show that the PM-IM conversion introduced by the chromatic dispersion (CD) of the sensing fiber leads to a crosstalk on the PM-IM conversion caused by Brillouin gain, which distorts the recovered BGS and introduces decoding errors in the signal processing [10]. On the other hand, the shape of the BGS is highly related to the local Brillouin gain due to CD effect, which may severely degrade measurement stabilities.
In order to reduce CD-induced BGS distortion and improve system stabilities, a novel coherent BOTDA sensor by utilizing intensity modulated probe (IMP) instead of phase modulated probe (PMP) is proposed and experimentally demonstrated in this paper. Firstly, the theoretical models for BGS calculation without/with the effect of CD are implemented, following by corresponding simulation results. Furthermore, comparative experiments have been implemented to demonstrate the theoretical models, in which the PMP is generated by cascaded intensity and phase electro-optic modulators (EOMs), and the IMP is generated by cascaded two intensity EOMs, besides the BGS carried by ~GHz carriers is decoded by a fast IQ demodulator. At last, about 6-MHz decoding error along the whole 40-km sensing fiber in PMP case is cancelled in IMP case due to the reduction of BGS distortion, meanwhile 3-m spatial resolution and < 1°C temperature resolution at the far end of the sensing fiber is realized. On the other hand, enhanced system stabilities are demonstrated by measuring the BGS under different conditions (e.g. RF power, bias voltage).
2. Operation principle
Figure 1(a) shows the schematic diagram of the coherent BOTDA sensors with PMP [4–9]. When the phase modulated CW probe meets the counter-propagating pulsed pump in the fiber, the fP-fLO-fS frequency component will be amplified by the Brillouin gain induced by the pulsed pump. Therefore, the perfect phase balance of the phase modulation is broken, which engenders the PM-IM conversion to reconstruct the BGS. However, the PM-IM conversion is also introduced by the CD after fiber transmission, which leads to the distortion of the measured BGS. Figure 1(b) shows the proposed scheme, in which the PMP is replaced by IMP. The original motivation of this replacement can be found in [11], which indicates that the CD induced IM-IM conversion is much more insensitive to radio frequency (RF) modulating frequency change compared with PM-IM conversion (when the modulating frequency is lower than 2-GHz typically). Therefore, the distortion of the measured BGS when utilizing IMP can be ignored. More detailed numerical analysis about the effect of CD on BGS will be shown below, and an infinite extinction ratio (ER) of the pulsed pump is assumed.
Fig. 1 Schematic diagram of coherent BOTDA sensors based on (a) phase modulated probe (PMP) and (b) intensity modulated probe (IMP). PM: phase modulation; IM: intensity modulation; FUT: fiber under test; CD: chromatic dispersion. fP is the frequency of laser source light, fP-fLO is the frequency of local light, generated by the frequency downshift of laser source light by fLO, fP-fLO ± fSare the frequencies of the two sidebands of the phase modulation of the local light in PMP case and of the two sidebands of the intensity modulation of the local light in IMP case, respectively.
2.1 Theoretical model without the effect of CD
When CW probe interacts with the counter-propagating pulsed pump, the fP-fLO-fS frequency component will be amplified by the Brillouin gain, then the optical fields before the PD in PMP and IMP cases are
where ELO and ES are the complex amplitudes of fP-fLO frequency component and fP-fLO ± fS frequency components, respectively. fD = fs + fLO-fB is the detune between probe light frequency and peak frequency of Brillouin gain. fB is BFS. gSBS(f) = g0vB2/(vB2 + 4fD2) is the amount of Brillouin gain centered at frequency fD = 0 and φg(f) = −2g0vBfD/(vB2 + 4fD2) is the Brillouin phase shift, in which g0 is the local Brillouin gain and vB is Brillouin linewidth. Then the PD output voltages as functions of fS in PMP and IMP cases can be expressed as [10]where RC and Rd are the sensitivity and the load resistance of the PD respectively, ϕp and ϕi are the phases of PD output voltages in PMP and IMP cases respectively. The approximation for the last terms in Eq. (2) have been obtained considering a small g0. Equation (2) indicates that the detected intensity at frequency fs is modulated by the BGS, which can be easily demodulated by many methods (e.g. envelope detection and IQ demodulation). On the other hand, it is also modulated by the amplitude of local light (i.e. ELO), which can be set to be much bigger than Es*sqrt(1 + gSBS)(fD = 0), so the SNR of the demodulated BGS signal can be dramatically improved in both PMP and IMP cases.2.2 Theoretical model with the effect of CD
In the theoretical analysis above, the effect of CD has been ignored. However, in long distance and big range distributed Brillouin sensing, the CD induced PM-IM conversion plays a non-ignorable role. When considering CD, output voltages of the PD at frequency fs in PMP and IMP cases can be written as
where D and L are the chromatic dispersion coefficient and the length of the sensing fiber, respectively. λc is optical wavelength of the local light, c is the speed of light in vacuum, ϕdp and ϕdi are the phases of PD output voltages in PMP and IMP cases, respectively. Equation (3) indicates that symmetric BGS reconstructed through IM-IM or PM-IM conversion induced by Brillouin gain can be distorted by CD induced additional phase term (i.e. πλc2DLfS2/c). The CD induced IM-IM conversion in IMP case is proportional to cos2(πλc2DLfS2/c), which is much more insensitive to fS compared to the CD induced PM-IM conversion that is proportional to sin2(πλc2DLfS2/c) in PMP case (when fS is less than 2GHz, πλc2DLfS2/c becomes relatively small). Consequently, the CD induced BGS distortion in IMP case is much less than it in PMP case and may be ignored. The PD output voltages without pulsed pump gain in PMP and IMP cases arewhere ϕwp and ϕwi are the phases of PD output voltages in PMP and IMP cases, respectively. After subtracting the DC components in Eq. (4) (i.e. amplitudes of the PD output voltage), the Brillouin gain induced voltages in these two cases as functions of fD can be expressed as3. Simulation results
Equation (5) is simulated to observe the BGS and evaluate the effect of CD. In the simulation, λc = 1550nm, L = 40km, fD changes from −100MHz to 100MHz, fB = 10.8GHz, vB = 40MHz, and fLO = 9.6GHz. Figure 2(a) shows the simulated BGS when PMP is applied and assuming that D = 0 (i.e. ignoring CD) at different local Brillouin gain values (i.e. g0 = 0.005, 0.05, and 0.5), they overlap with each other with the same Lorentz shape of a linewidth of 68MHz. On the other hand, when D = 17ps/nm/km (corresponding to standard single mode fiber (SSMF)), the simulated BGS at these same g0 values are shown in Fig. 2(b). They are obviously different from the corresponding results shown in Fig. 2(a), which indicates that the CD causes significant distortion on the BGS in PMP case.
Fig. 2 Simulated BGS (i.e. Vgp) at different local gain values when PMP is applied and assuming that (a)D = 0 and (b) D = 17ps/nm/km.
On the other hand, the simulated BGS at these same g0 values in IMP case without and with the effect of CD are shown in Figs. 3(a) and 3(b), respectively. The simulated BGS in both figures are with the same Lorentz shape of a linewidth of 40MHz, consequently the effect of CD on the BGS in IMP case can be ignored. The results in Fig. 2 and Fig. 3 also indicate that when the sensing fiber, i.e. CD, is fixed, the BGS in PMP case is very sensitive to g0, which is decided by many factors, such as pump power, probe power, ER of the pulsed pump, and amplified spontaneous emission (ASE) noise. However, the change of g0 value almost has no effect on the BGS in IMP case, thus the stabilities in IMP case is much better than it in PMP case.
Fig. 3 Simulated BGS (i.e. Vgi) at different local gain values when IMP is applied and assuming that (a) D = 0 and (b) D = 17ps/nm/km.
4. Experimental setup
The experimental setup is shown in Fig. 4. A CW light with a central wavelength of 1500.12nm from the tunable laser source (TLS) with a 3-dB linewidth of 100-KHz and a peak power of 10dBm is split into two branches by a 50:50 coupler. One branch is modulated by two intensity electro-optic modulators (EOM1 and EOM2) sequentially. EOM1 is operated in carrier-suppressed mode and driven by a microwave source with fixed 9.6-GHz output frequency (i.e. fLO) and 18-dBm output power. EOM2 is biased at quadrature point and driven by a RF signal generator with 1.1~1.3GHz tunable output frequencies (i.e. fS) and 10-dBm output power. The polarization controller (PC) before EOM2 is adjusted to achieve max modulation efficiency. The output light of the EOM2 is then amplified by an erbium-doped fiber amplifier (EDFA1) with adjustable gain and then launched into to the same sensing fiber by an isolator, the peak powers of local light and probe light are 0dBm and −20dBm, respectively. The other branch is modulated by an intensity modulator (EOM3) which is biased at carrier-suppressed mode and driven by a pulsed signal generator (PSG) in order to obtain 30-dB ER single pulse light with 1-ms period and 30-ns duty cycle, allowing for 3-meter spatial resolution. The output light of EOM3 is amplified by the gain-adjustable EDFA2 to be 18dBm and then sent to the sensing fiber as pulsed pump light through a circulator.
Fig. 4 Experimental setup of the proposed coherent BOTDA sensor. TLS: tunable laser source; PC: polarization controller; EOM: electro-optic modulator; EDFA: erbium-dropped fiber amplifier; CW: continue wave; PD: photo-detector; BPF: band pass filter; LNA: low noise amplifier; OSC: oscilloscope.
After interacting with the pulsed pump light and the filtering of a narrow-bandwidth FBG (FWHM 0.1nm), the probe light is detected by a PD with a 3-dB bandwidth of 10GHz. The recovered RF signal is selected by a band-pass filter (BPF) and amplified by a low noise amplifier (LNA) to increase SNR. An IQ demodulator is utilized to decode the amplitude of the recovered RF signal into baseband signal, which is further sampled and stored by an oscilloscope with 100-MHz sampling rate. High demodulation bandwidth (i.e. 3-dB 500MHz) of the IQ demodulator makes it fast enough to decode the amplitude modulation. Note that, when the intensity EOM2 is replaced by a phase EOM, PMP instead of IMP can be generated.
5. Results
About 40-km SSMF with 10.8-GHz BFS at room temperature is employed as sensing fiber in the experiment. About 25-m testing fiber near the far-fiber end is heated up to 45°C by the oven and the remaining 39.975km is at room temperature (i.e. 25°C), the temperature difference of 20°C results in ~20MHz frequency shift on the BFS. By tuning the RF signal over a frequency span from 1.1GHz to 1.3GHz with a 4-MHz step, the BGS has been measured after 1000-time average at each sweeping frequency point. BFS along the whole fiber is further decoded through Lorentz fitting of the measured BGS.
In PMP case, the measured PM-IM conversion without pulsed pump gain (i.e. Vwp) is shown in the green line of Fig. 5(a), which increases with sweeping frequency and is mainly caused by CD as indicated in [10]. The measured PM-IM conversion with pulsed pump gain (i.e. Vdp) is shown in the blue line of Fig. 5(a), and the calculated BGS (i.e. Vgp) through Eq. (5) is shown in the purple line of Fig. 5(b). Due to the asymmetry caused by CD, ~6-MHz fitting error of the BGS is made as shown in Fig. 5(b). In IMP case, the measured IM-IM conversion without pulsed pump gain (i.e. Vwi) is shown in the green line of Fig. 6(a), on the other hand, the dashed black line shows the measured voltage jitter of the RF sweeping signal without fiber link transmission, it’s obvious that the measured IM-IM conversion mainly exhibits the properties of the voltage jitter of the RF sweeping signal (i.e. ES) due to that the CD induced IM-IM conversion is very slight and can be ignored. The difference between the dashed black line and the green line in the frequency region from 10.78GHz to 10.82GHz is obviously bigger than other frequencies due to the Brillouin gain amplification induced by CW optical leakage caused by finite extinction ratio of the intensity EOM3. The blue line (i.e. Vdi) in Fig. 6(a) shows the measured IM-IM conversion with pulsed pump gain, and the calculated BGS (i.e. Vgi) through Eq. (5) and its Lorentz fitting curve are shown in the purple line and red line in Fig. 6(b), respectively. It implies that the CD induced BGS asymmetry by utilizing IMP can be ignored since no fitting error is made.
Fig. 5 (a) Measured PM-IM conversion versus fS + fLO without pulsed pump gain (i.e. Vwp shown in the green line) and with pulsed pump gain (i.e. Vdp shown in the blue line). (b) Calculated BGS (i.e. Vgp shown in the purple line) and its Lorentz fitting curve (shown in the red line).
Fig. 6 (a) Measured IM-IM conversion versus fS + fLO without pulsed pump gain (i.e. Vwi shown in the green line) and with pulsed pump gain (i.e. Vdi shown in the blue line), the dashed black line shows the voltage jitter of the RF sweeping signal (i.e. ES) without fiber link transmission. (b) Calculated BGS (i.e. Vgi shown in the purple line) and its Lorentz fitting curve (shown in the red line).
The decoded BFS along the 40-km sensing fiber in PMP and IMP cases are shown in Fig. 7(a), the oscillation and fluctuation of the decoded BFS is introduced by additional strain due to fiber coiling and system noises. The purple line with an average value (AV) of 6MHz in the insert of Fig. 7(a) shows the difference between the decoded BFS in IMP case and PMP case along the whole sensing fiber (~6-MHz fitting error reduction along the whole sensing fiber), which indicates that IMP based BOTDA sensor is more suitable for accuracy temperature measurement.
Fig. 7 Decoded BFS (a) along the whole sensing fiber and (b) from 39.95km to 40km in PMP case (blue line) and IMP (red line) case. The insert in (a) shows the difference between the decoded BFS in IMP case and PMP case, with an average value (AV) of 6-MHZ
Figure 7(b) shows the decoded BFS from 39.95km to 40km in these two cases, ~20MHz frequency transition between the heated and unheated segment is observed obviously and ~3m spatial resolution is measured, which is with a good agreement to the 30-ns pulse width. The achieved temperature resolution has been calculated through the standard deviation of the BFS trace, resulting in less than 1MHz at far fiber end, which corresponds to a temperature resolution of less than 1°C.
In order to further compare the system performance in IMP and PMP cases, the system stabilities are measured by meausing the BGS under different conditions. Firstly, we adjust the power of the sweeping RF signal to be 0dBm and 20dBm, and the ER is kept at 30dB. The blue and green lines in Figs. 8(a) and 8(b) show the measured BGS under 0-dBm and 20-dBm sweeping signal powers, respectively. On the other hand, the ER of the pulsed pump is adjusted to be 20dB by controlling the bias voltage of EOM3, the measured BGS of the PMP and IMP cases are shown in the red lines in Figs. 8(a) and 8(b), respectively. The measured BGS are quite different from each other as shown in Fig. 8(a), but almost the same as shown in Fig. 8(b). This is because that the BGS in the PMP case is very sensitive to g0 variation caused by the change of the probe power and the ER value. Consequently, the system stability of the IMP case is much better than that of the PMP case, which agrees well with the simulation results.
Fig. 8 Measured BGS at three different conditions: 1) ER = 30dB and fS = 0dBm (blue line); 2) ER = 30dB and fS = 20dBm (green line), and 3) ER = 20dB and fS = 10dBm (red line) in (a) PMP case and (b) IMP case.
6. Conclusion
Coherent BOTDA sensor with IMP has been proposed and experimentally demonstrated. Experimental results indicate its advantage in the immunity to CD induced BGS distortion and system instabilities, implying that IMP instead of PMP is more suitable for coherent BOTDA sensing. Reduction of ~6-MHz BFS decoding error is achieved along 40-km sensing distance, while 3-m spatial resolution with less than 1°C temperature resolution at the far fiber end is obtained. Besides, BGS with good repetitiveness is measured under different conditions, thus enhanced system stabilities is realized.
Acknowledgments
The research is supported in part by the International Science and Technology Cooperation Program of China (2014DFA11170), the National Natural Science Foundation of China (NSFC) (No. 61325023),the Key Grant Project of Chinese Ministry of Education (No.313049) and the key project of Sichuan Province of China (2011GZ0239).
References and links
1. X. Angulo-Vinuesa, S. Martin-Lopez, P. Corredera, and M. Gonzalez-Herraez, “Raman-assisted Brillouin optical time-domain analysis with sub-meter resolution over 100 km,” Opt. Express 20(11), 12147–12154 (2012). [CrossRef] [PubMed]
2. W. Li, X. Bao, Y. Li, and L. Chen, “Differential pulse-width pair BOTDA for high spatial resolution sensing,” Opt. Express 16(26), 21616–21625 (2008). [CrossRef] [PubMed]
3. M. A. Soto, G. Bolognini, and F. Di Pasquale, “Analysis of pulse modulation format in coded BOTDA sensors,” Opt. Express 18(14), 14878–14892 (2010). [CrossRef] [PubMed]
4. X. Lu, M. A. Soto, M. González-Herraez, and L. Thévenaz, “Brillouin distributed fibre sensing using phase modulated probe,” in Proc. SPIE, Fifth European Workshop on Optical Fibre Sensors8794, 87943P (2013). [CrossRef]
5. A. Zornoza, M. Saguesand, and A. Loayssa, “Self-heterodyne detection for SNR improvement and distributed phase-shift measurements in BOTDA,” J. Lightwave Technol. 30(8), 1066–1072 (2012). [CrossRef]
6. J. Urricelqui, M. Sagues, and A. Loayssa, “Synthesis of Brillouin frequency shift profiles to compensate non-local effects and Brillouin induced noise in BOTDA sensors,” Opt. Express 22(15), 18195–18202 (2014). [CrossRef] [PubMed]
7. J. Urricelqui, M. Sagues, and A. Loayssa, “Phasorial differential pulse-width pair technique for long-range Brillouin optical time-domain analysis sensors,” Opt. Express 22(14), 17403–17408 (2014). [CrossRef] [PubMed]
8. J. Urricelqui, M. Sagues, and A. Loayssa, “BOTDA measurements tolerant to non-local effects by using a phase-modulated probe wave and RF demodulation,” Opt. Express 21(14), 17186–17194 (2013). [CrossRef] [PubMed]
9. X. Angulo-Vinuesa, D. Bacquet, S. Martin-Lopez, P. Corredera, P. Szriftgiser, and M. Gonzalez-Herraezet, “Relative intensity noise transfer reduction in Raman-assisted BOTDA systems,” IEEE Photonics Technol. Lett. 26(3), 271–274 (2014). [CrossRef]
10. Z. Li, L. Yan, L. Shao, W. Pan, and B. Luo, “Coherent BOTDA sensor with intensity modulated local light and IQ demodulation,” Opt. Express 23(12), 16407–16415 (2015). [CrossRef] [PubMed]
11. F. Ramos and J. Martí, “Frequency transfer function of dispersive and nonlinear single-mode optical fibers in microwave optical systems,” IEEE Photonics Technol. Lett. 12(5), 549–551 (2000). [CrossRef]
