Distributed fiber sensor based on modulatedpulse base reflection and Brillouin

gain spectrum analysis

Qingsong Cui,1,2,* Sibel Pamukcu,2 Wen Xiao,1 Cyril Guintrand,3

Jean Toulouse,3 and Mesut Pervizpour4

1School of Instrument Science and Opto-Electronics Engineering, Beihang University, Beijing 100083, China2Department of Civil and Environmental Engineering, Lehigh University, Bethlehem, Pennsylvania 18015, USA

3Department of Physics, Lehigh University, Bethlehem, Pennsylvania 18015, USA4Department of Civil Engineering, Widener University, Chester, Pennsylvania 19013, USA

*Corresponding author: qingsong.research@gmail.com

Received 12 May 2009; revised 10 August 2009; accepted 24 September 2009;posted 2 October 2009 (Doc. ID 111275); published 16 October 2009

In recent years, several distributed sensor systems based on stimulated Brillouin scattering in opticalfibers have been proposed [ J. Intell. Mater. Syst. Struct. 10, 340 (1999); Proc. SPIE 5855, 555 (2005) ]. Wepropose a simpler scheme based on fiber-end reflection and Brillouin gain spectrum analysis. In thissetup, only one optical source is necessary to provide both the pump and the probe wave; the latteris provided by the modulated pulse base. First, the physical mechanisms for two different Brillouin scat-tering processes in our sensor system are analyzed and an approximate theory model is proposed. Inaddition, it is demonstrated that the simple system configuration allows simultaneous acquisition ofthe time-domain and the frequency-domain information. It is experimentally demonstrated thatthis configuration is effective for strain measurements and could as well be applied to temperaturemonitoring. © 2009 Optical Society of America

OCIS codes: 060.2370, 060.2630, 290.5900.

1. Introduction

Distributed sensor systems based on stimulated Bril-louin scattering (SBS) have attracted much interestin past years because of their increasingly variedapplications in the civil engineering area, such ashealth monitoring of civil infrastructure, vibrationsmeasurement, or environmental effects [1–4]. Oursensor configuration relies on the SBS effect in opti-cal fiber. When an optical pump wave is injected intoa fiber, incident light is partially backscattered dueto its interaction with acoustic phonons. The back-scattered light frequency is shifted toward a lowerfrequency compare to the incident light. It is also pos-

sible to amplify a counterpropagating signal when itsoptical frequency falls in the SBS gain spectrum. Thefrequency shift is directly related to the acousticproperties of the fiber in which the waves interact.The typical frequency shift value is∼10GHz in silicafibers. Change in temperature or strain conditions ofthe fiber induces variations in the acoustic propertiesof the medium and, by consequence, can be moni-tored by examination of the SBS frequency shiftobserved.

A number of research groups have engaged in thedesignof distributed sensor systembasedonBrillouinscattering. For instance,Horiguchi and his colleaguesreported the first distributed sensor, called Brillouinoptical time-domain analysis (BOTDA), using apulsed pump wave and a cw Stokes probe in acounterpropagation scheme [1]. Brillouin optical

0003-6935/09/305823-06$15.00/0© 2009 Optical Society of America

20 October 2009 / Vol. 48, No. 30 / APPLIED OPTICS 5823

time-domain reflectometry (BOTDR) was introducedby Kurashima et al. [5,6]. They analyzed the sponta-neous Brillouin backscattered light instead of theBrillouin amplification signal using coherent detec-tion. Niklès and coworkers presented a distributedBrillouin sensor consisting of a single laser and a sin-gle fiber based on a pump-and-probe technique [7,8].One important feature of the system was the genera-tion of pump-and-probe signals in a single LiNbO3intensity electro-optic modulator separately in atime-sharing method.All those studies have shown their advantages and

disadvantages. The disadvantage of the conventionalBOTDA is that it requires a laser to be placed at eachend of the fiber being tested and a fixed frequencyrelationship with each other [9]. Using only oneend of the sensing fiber is the advantage of theBOTDR system. The drawback is that the sponta-neous Brillouin scattering signal is much weakerthan that of BOTDA and the spatial resolution is lim-ited to 1m, which is related to the phonon lifetime.The single-laser BOTDA system [7,8] is simpler thanthe conventional BOTDA system. However, in thepresented design, the RF port is shared for genera-tion of both the pump and the probe waves, whichincreases the complexity of the system. This time-division-multiplexing method requires an electriccircuit to switch between the microwave and thepulse generator and requires a stable circuit design,especially for high resolution measurements.In this paper, we propose a distributed fiber sensor

system based on fiber-end reflection and Brillouingain spectrum analysis, and we demonstrate the effi-ciency of this system. Using theoretical simulationand experimental results, we attempt to show thatthe modulated pulse base reflection BOTDA is a sim-pler sensor system. In our configuration, the probe isgenerated by the modulated pump base reflection.Pulse andmicrowave signals are simultaneously con-nected to the electro-optic modulator (EOM) electro-des and no time multiplexing is required. Brillouininteractions are analyzed to detect the optimalelectrical power values necessary to drive the EOM.In order to demonstrate the effectiveness of our sys-tem and ensure the correctness of the interactionme-chanism, theBrillouin scatteringprocess is simulatednumerically to confirm the experiment results.

2. Principle and Experimental Setup

A schematic diagram of our proposed system is illu-strated in Fig. 1. In this technique, the light beamfrom a distributed feedback (DFB) laser is injectedinto the EOM. A polarization controller (PC) is posi-tioned before the EOM to insure correct polarizationof the input light. A sinusoidal RF signal with fre-quency f m is used to modulate the optical wave; thiscreates sidebands on each side of the laser frequencyf 0. Simultaneously, a pulse generator is also con-nected to the bias port of the EOM to create opticalpulses. No logical gate and electric circuit is appliedat this stage, which is a unique feature of our system.

A 30dBm output power erbium-doped fiber amplifier(EDFA) is used to amplify the input signal to a highpeak power. We use a circulator placed at the fiberunder test (FUT) input to collect the backscatteredlight. Reflection of the light at the fiber output endis provided by Fresnel reflection from the glass/airinterface (∼4% in our case). A fiber Bragg grating(FBG) is used to filter the Stokes component fromthe total backscattered light. Finally a PIN detectorand an oscilloscope are used to monitor the filteredbackscattered light power.

The process of interaction in the fiber is dividedinto two steps: (1) the pulse base (low state) preced-ing the pulse (high state) interacts with the reflectedportion of the base and (2) the pulse interacts withthe reflected base. When the modulation frequencyf m approaches the Brillouin frequency of the fiber,SBS interaction occurs and the Stokes componentin the reflected probe signal is amplified. Thetime-dependent output power detected by the oscillo-scope carries local information about the tempera-ture and strain change along the sensing fiber.

Figure 2 shows the EOM principle for our systemconfiguration. Figure 2(a) is the structure of thepush–pull intensity modulator. The phase differencebetween the two arms can be written as

ΔϕðtÞ ¼ ΔϕbiasðtÞ þΔϕRFðtÞ: ð1Þ

ΔϕbiasðtÞ and ΔϕRFðtÞ are the phase difference be-tween the optical waves from the two EOM branchesinduced by the pulse and microwave signal, respec-tively. This method is superior to the one presentedin [7], since it does not require a time-division multi-plexing of pulse and sinusoidal waves. The sinusoidalmodulation and the pulse modulation are superim-posed, as shown in Fig. 2(b). The pulse base approachhas already been used by other groups as prepumpwave to improve the spatial resolution [10]. In ourconfiguration, the pulse base is not only the prepumpwave but, more importantly, it is used as theprobe wave.

The energy transfer mechanism is schematized inFig. 3; on top, we have represented the probe wavebeam (reflected pulse base) spectrum, and on thebottom, the pump wave (pulse) spectrum. The pulsefinite extinction ratio is defined as ER ¼ Ppulse=Pcw,where Ppulse and Pcw are the power of the pulsetop and the pulse base, respectively.

Fig. 1. Diagram of the distributed fiber sensor system tested.

5824 APPLIED OPTICS / Vol. 48, No. 30 / 20 October 2009

When the modulation frequency f m is equal to theBrillouin frequency νB, the counterwaves interactwith each other and energy transfer occurs betweenthe waves. From Fig. 3, it is observed that there aremainly four Brillouin scattering energy transfer pro-cesses occurring in the fiber, Ep1 → Es0, Es0 → Ep2,Es1 → Ep0, and Ep0 → Es2. The four processes couldbe separated to two groups: (1) Ep1 → Es0 and Es0 →

Ep2 and (2) Es1 → Ep0 and Ep0 → Es2. The intensity ofgroup (2) is much higher than group (1); thus, theBrillouin interaction of group (1) is much less thangroup (2). In group (2), probe Es1 corresponds to lossspectrum and probe Es2 corresponds to gain spec-trum. Our sensor system is based on gain spectrumanalysis, thus Es2 is the probe component detected inour sensor system. By monitoring the power of Es2,we can obtain direct information on the fiber stressand temperature variations.

3. Mathematical Model and Numerical Simulations

In Section 2, we have shown that the main SBS in-teraction in the fiber involves the pump wave compo-nent Ep0 and the reflected signal component Es2(Stokes wave). A three coupled-wave equation sys-tem constitutes the basic theoretical model used tosimulate the Brillouin scattering in the sensing fiber.The equations, using the slowly varying amplitudeapproximation, can be written as [11]

�∂

∂z−nc∂

∂t−12α�Ep0 ¼ ig1QEs2; ð2Þ

�∂

∂zþ n

c∂

∂tþ 12α�Es2 ¼ −ig1Q�Ep0; ð3Þ

�∂

∂tþ Γ

�Q ¼ −ig2Ep0E�

s2: ð4Þ

Equations (2) and (3) are descriptions of the opticalfields and Eq. (4) is the acoustic wave equation, as-suming the acoustic amplitude varies slowly. WhereEp0, Es2, and Q are the amplitudes of the pulse beam(pump), the pulse base reflection (Stokes) beam, andthe acoustic wave fields, g1 and g2 are the photon–phonon coupling coefficients, α is the attenuationconstant of the fiber, and Γ ¼ Γ1 þ iΓ2, with

Γ1 ¼ 12τph

;

Γ2 ¼ 2πðν − νBÞ:

Γ1 is the damping rate with phonon lifetime τph ∼10ns and Γ2 represents the detuning frequency.The detected optical power could be defined asP ¼ Aeff · jEj2, where Aeff is the fiber effective corearea and E is the optical wave field. The numericalmethod of transient SBS is applied to solve theequations [12].

In the simulation, we assumed a 50m fiber length,an unstressed fiber Brillouin shift νB ¼ 10866MHz,and the effective fiber core area Aeff ¼ 54 μm2. Thepeak power, ER, and pulse width are 24dBm,

Fig. 2. (Color online) EOM structure and transfer function:(a) balanced push–pull EOM structure and (b) optical modulationtransfer function.

Fig. 3. Energy transfer mechanism between pump-and-probelight.

20 October 2009 / Vol. 48, No. 30 / APPLIED OPTICS 5825

27dB, and 10ns, respectively. The fiber-end reflec-tion is 4%. The length of the stressed section is1m, corresponding to the spatial resolution of the10ns pulse and positioned between the 20 and21m linear positioning of the fiber, and the Brillouinfrequency shift of the stressed fiber is 10; 950MHz.Figure 4 presents the simulation results of the Bril-

louin scattering interaction. The figure illustrates theBrillouin spectrum that carries the pulse base pumpwave and its reflected probewave interaction, and thepulse pump and the reflected pulse base probe inter-action, including the information on spatial resolu-tion. Figure 4(a) describes the stimulated gainBrillouin spectrum for the Stokes component (Es2)at the stressed section by scanning the frequencyf m. The lower peak is the pulse base component con-tribution and the higher peak is the pulse pump con-tribution to the Brillouin spectrum for the stressedspectrum. In Fig. 4(b), the base of the time-domainsignal corresponds to the steady interaction betweenthe pulse base and its reflection. The peak section isthe transient Brillouin interaction from the stressedsection (f m ¼ 10; 950MHz), which is much strongerthan other frequencies in the same stressed fiber.

4. Experimental Results

We carried out some experiments based on the prin-ciple presented in Section 2. We proved its efficiencyin terms of precise localization and the quantitativevalue of the stress applied in the fiber. The experi-mental results presented here were conducted usingthe experimental setup from Fig. 1. In this experi-ment, we use a laser wavelength of 1551nm andthe linewidth is less than 1MHz. The FUT is∼270m of SMF-28 (silica fiber); low cost and easyavailability dictate our choice. We measured a Bril-louin shift of 10; 866MHz for the SMF-28 when nostress was applied to the fiber (room temperature).The sinusoidal RF signal was provided by a signalgenerator with a 0–20GHz frequency range, and a2MHz sweeping step was applied. The pulse genera-tor was set at a 10ns pulse width with a 1KHz repe-tition rate. We used a three-point travelling pulleysystem to apply axial strain to a specific section ofthe fiber.Figure 5(a) shows the experimental spectrum at

the fiber input in the pulse high state and the pulselow state (around Vπ) [Fig. 5(b)]. The interaction ofgroup (1) can be highly suppressed by adjustingthe pulse voltage and the microwave signal power.Figure 5(a) shows that the intensity of pump Ep0is 16dB higher than the sidebands, Ep1 and Ep2. Fig-ure 5(b) shows that the intensity of probe Es2 is 18dBhigher than the carrier wave Es0.Figure 6 illustrates the correlation between the ef-

fective axial strain applied (in term of mass force)and the Brillouin frequency shift. Figure 6(a) was ob-tained by monitoring the power of the backscatteredlight for two different positions (z ¼ 45m andz ¼ 74m) in the same fiber, in which a stress was ap-plied at z ¼ 74m. The left curve (circles) corresponds

to the position 1 (at z ¼ 45m, where no stress wasapplied) and the right curve (squares) is related tothe position 2 (at z ¼ 74m, where stress is applied).The sinusoidal frequency modulation f m was tunedand the backscattered light powers correspondingto those two positions were monitored. For position1, the peak power is observed for a frequency modu-lation of 10; 866MHz, which corresponds to the nat-ural fiber Brillouin shift. When a stress was applied(540:4 g) to the fiber section (position 2), the peakpower shifted to 11; 016MHz, a result of the fiberproperties change. Looking closely at the curve cor-responding to position 2 (stressed section), we noticea second small peak (circled portion) locatedat 10; 866MHz.

The results displayed in Fig. 6(b) were obtainedwhen we applied a strain with a 540:4 g weight (on1m of fiber) at a distance of 74m from the fiber input.The sinusoidal modulation was set at 11; 016MHzand the time-dependent backscattered Stokes powerwasmonitored on the oscilloscope.Then the timevari-able was translated in fiber distance. Figure 6(a) ex-hibits a strong peak around 74m, which indicates astress at this point.

Fig. 4. (Color online) Energy transfer mechanism simulationbetween pump-and-probe light: (a) Brillouin gain spectrum and(b) time-domain signal output power.

5826 APPLIED OPTICS / Vol. 48, No. 30 / 20 October 2009

Figure 7 shows the measured Brillouin frequencyshift for different strain values induced by the appli-cation of the weight. The measurement error is about4MHz with a 2MHz sweeping step. The solid line isthe linear fit of experimental data (blue stars). Therelationship between Brillouin frequency shift andthe strain exhibits a linear variation given by

ΔvBðGHzÞ ¼ 2:615 × 10−4 ×WeightðgÞ:

These results show that the proposed system is aneffective sensor system for strain measurements.Under the conditions of the experiment and the

frequency scanning step (2MHz), the root-mean-square deviation (RMSD) between the linear fit lineand the experimental data is 23:6 g, and the straincoefficient of the Brillouin frequency shift is about0:26MHz=g. Hence the strain (weight) resolutionunder our certain experiment conditions could betreated as 30 g.

Another advantage of our system is that the pulseleakage (base) is used as a prepump, as in [10]. FromFig. 6(a), it is obvious that the spectrum has a nar-

Fig. 5. (Color online) Spectrum detection for pump-and-probesignal from optical spectrum analyzer: (a) pulse high level spec-trum, (b) pulse low level spectrum.

Fig. 6. (Color online) Brillouin frequency shift /strain dependencefor standard SMF-28 fiber: (a) Brillouin spectrum for two positions(with and without stress) and (b) time-domain signal output power(modulation frequency, 11;016MHz; weight, 540:4 g).

Fig. 7. (Color online) Weight strain and Brillouin frequency shiftrelation results.

20 October 2009 / Vol. 48, No. 30 / APPLIED OPTICS 5827

rower top peak than the normal BOTDA system (thebandwith for 10ns is about 80MHz [13]). Those prop-erties lead us to think that our system could offer bet-ter accuracy on stress localization of the order ofsubmeters.

5. Conclusion

We have presented a distributed fiber sensor using asingle optical source with simultaneous pulse andsinusoidal modulation. This is more convenient thanthe systems previously used and reported, which in-clude either two optical sources or more complex RFcircuits. We have derived and numerically solved thecoupled equations for the fields involved in the pro-cess. By using this configuration of pulse base reflec-tion as the probe wave, a Brillouin frequency shiftinduced by strain or temperature has been observed.This system allows accurate localization of stressand quantitative evaluation of the strain magnitude.Temperature monitoring could also be used with thesame experiment setup.

Q. Cui is supported by the China ScholarshipCouncil, China, and performs this research as a vis-iting Ph.D. student at Lehigh University. The work issupported financially by grant CMMI-0855603 fromthe National Science Foundation (NSF). We alsowant to thank Prof. Xiaoyi Bao, Dr. Yongkang Dong,Dr. Yun Li, and Dr. Feng Wang at the University ofOttawa for providing a visiting opportunity andvaluable suggestions.

References1. T. Kurashima, T. Horiguchi, and M. Tateda, “Distributed-

temperature sensing using stimulated Brillouin scatteringin optical silica fibers,” Opt. Lett. 15, 1038–1040 (1990).

2. A. W. Brown, J. P. Smith, and X. Bao, “Brillouin scatteringbased distributed sensors for structural applications,” J.Intell. Mater. Syst. Struct. 10, 340–349 (1999).

3. S. Texier, S. Pumukcu, and J. Toulouse, “Advances in subsur-face water-content measurement with a distributed Brillouinscattering fibre-optic sensor,”Proc. SPIE5855, 555–558 (2005).

4. S. Anastasio, S. Pamukcu, and M. Pervizpour, “Chemical Se-lective BOTDR Sensing for Corrosion Detection on StructuralSystems,” in Proceedings of the Seventh International Work-shop on Structural Health Monitoring (International Work-shop on Structural Health Monitoring, 2007), pp. 1701–1708.

5. T. Kurashima, T. Horiguchi, and M. Tateda, “Thermal effectsof Brillouin gain spectra in single-mode fibers,” IEEE Photo-nics Technol. Lett. 2, 718–720 (1990).

6. T. Kurashima, T. Horiguchi, H. Izumita, S. Furukawa, and Y.Koyamada, “Brillouin optical-fiber time domain reflectome-try,” IEICE Trans. Commun. E76-B, 382–390 (1993).

7. M. Niklès, L. Thevenaz, and P. Robert, “Simple distributedfiber sensor based on Brillouin gain spectrum analysis,”Opt. Lett. 21, 758–760 (1996).

8. A. Fellay, L. Thevenaz, M. Facchini, M. Niklès, and P. Robert,“Distributed sensing using stimulated Brillouin scattering: to-wards ultimate resolution,” inOptical Fiber Sensors, Vol. 16 of1997 OSA Technical Digest Series (Optical Society of America,1997), paper OWD3.

9. T. Horiguchi and M. Tateda, “BOTDA—nondestructive mea-surement of single-mode optical fiber attenuation characteris-tics using Brillouin interaction: theory,” J. Lightwave Technol.7, 1170–1176 (1989).

10. K. Kishida, CH. Li, and K. Nishiguchi, “Pulse pre-pumpmeth-od for cm-order spatial resolution of BOTDA,” Proc. SPIE5855, 559–562 (2005).

11. R. W. Boyd, Nonlinear Optics, 2nd ed. (Academic, 2003).12. R. Chu, M. Kanefsky, and J. Falk, “Numerical study of

transient stimulated Brillouin scattering,” J. Appl. Phys.71, 4653–4658 (1992).

13. X. Bao, A. Brown, M. DeMerchant, and J. Smith, “Character-ization of the Brillouin-loss spectrum of single-mode fiber byuse of very short (<10ns) pulses,” Opt. Lett. 24, 510–512(1999).

5828 APPLIED OPTICS / Vol. 48, No. 30 / 20 October 2009