Distributed transverse stress measurement alongan optic fiber using polarimetric OFDRCHANGJIANG WEI,1 HONGXIN CHEN,2 XIAOJUN CHEN,2 DAVID CHEN,3 ZHIHONG LI,4 AND X. STEVE YAO1,2,*1Photonics Information Innovation Center (PIIC), Hebei University, Baoding, China2General Photonics Corporation, 5228 Edison Ave., Chino, California 91710, USA3Verizon Communications, 2400 N. Glenville Dr., Richardson, Texas 75082, USA4Suzhou Optoring, Jiansu Province, China*Corresponding author: steveyao888@yahoo.com

Received 4 April 2016; revised 29 May 2016; accepted 30 May 2016; posted 31 May 2016 (Doc. ID 262457); published 15 June 2016

We report a novel polarimetric optical frequency domainreflectometer (P-OFDR) that can simultaneously measureboth space-resolved transverse stresses and light back-reflections along an optic fiber with sub-mm spatialresolution. By inducing transversal stresses and opticalback-reflections at multiple points along a length of opticfiber, we demonstrate that our system can unambiguouslydistinguish the stresses from the back-reflections of a fiberwith a fiber length longer than 800 m, a spatial resolution of0.5 mm, a maximum stress level of up to 200 kpsi(1379 Mpa), a minimum stress of about 10 kpsi (69 Mpa),and a stress measurement uncertainty of 10%. We showthat our P-OFDR can clearly identify the locations andmagnitudes of the stresses inside a fiber coil induced duringa fiber winding process. The P-OFDR can be used for fiberhealth monitoring for critical fiber links, fiber gyro coilcharacterization, and other distributed fiber sensingapplications. © 2016 Optical Society of America

OCIS codes: (060.2270) Fiber characterization; (060.2370) Fiber op-

tics sensors; (120.4640) Optical instruments; (260.5430) Polarization.

http://dx.doi.org/10.1364/OL.41.002819

Connectivity is arguably one of the most important aspects ofour society today. Fiber optic telecommunication systems makeup the backbone of such connectivity, and its reliability is be-coming increasingly more important. Therefore, health mon-itoring for the fiber infrastructure networks for applications,such as stock exchanges, financial institutions, and military in-stallations, are becoming increasingly critical. An ideal fiberhealth monitoring system is required not only to be able todetect a broken fiber, but also to detect and locate the sectionof an unhealthy fiber, such as a fiber section with tight bendingor with high internal stresses, before it breaks, to allow theservice providers to proactively prevent unexpected servicefailures.

Distributed stress sensing is another area of active researchwith many important applications [1–4], one of which is tomeasure the stresses along the fiber in a fiber gyro coil, the mostcritical component whose quality directly affects the accuracy of

the gyro [5,6]. One of the most common coil defects is thestress induced during the coil winding process. In a fiber coilmade with polarization maintaining (PM) fiber, the windingstresses at different locations can introduce polarization cross-talks at these locations along the fiber, which can be detected bya distributed polarization crosstalk analyzer [7]. However, afiber coil made with non-polarization maintaining single-mode(SM) fiber is commonly used for the depolarized fiber opticgyro [8]. In such SM fiber coils, the microbending can inducelarge stress on the fiber, especially at crossover locations wherethe fiber transitions from one layer to another. We speculatethat such bending, together with the action of potting adhesive,can introduce large polarization dependent loss (PDL) andpolarization mode dispersion (PMD) in the fiber coil, whichmay degrade the performance of the SM fiber coils.

Current commercially available optical time-domain reflec-tometer (OTDR) can monitor, detect, and locate a section ofbroken fiber or a fiber section with significant fiber lossescaused by severe fiber bending [9]. However, in such situations,the network failure has already occurred and the damage isalready done. Although a Brillouin OTDR is able to detectpotential optical cable fault due to excessive strain, such a tech-nique may give fault alarms because it cannot distinguish thetemperature effect from that of the strain [10,11]. Polarizationsensitive OTDRs were reported for measuring the distributedpolarization mode dispersion (PMD) in telecommunication fi-ber links [12,13]. However, the spatial resolution of the pro-posed method were not sufficient for detecting the extremelyhigh stress-induced birefringence for the applications intendedin this Letter, limited by the pulse width and the speed of theelectronics. In 2004, Froggatt et al. used a technique of cross-correlating a current measured optical frequency domain reflec-tometry (OFDR) signal with a reference OFDR trace to obtainthe space-resolved strain changes along the optical fiber [14].However, such a method also suffers from the deficiency thatit cannot distinguish the strain from the temperature. In addi-tion, no absolute strain or stress information can be obtainedbecause such a measurement is only relative to the reference.The polarimeteric OFDRs reported in [15–18] are capableof detecting distributed PMD and different birefringencesalong the fiber; however, they required the fiber under test

Letter Vol. 41, No. 12 / June 15 2016 / Optics Letters 2819

0146-9592/16/122819-04 Journal © 2016 Optical Society of America

(FUT) to be stable to avoid polarization fluctuation during themeasurement, which is not practical for field applications.Therefore, there is a need for a fiber stress and bend monitoringsystem to unambiguously detect and locate unhealthy fiber sec-tions with excessive stress caused by tight bends or transversepressure so as to prevent unexpected failures or breakage in fiberoptic networks. Similarly, there is also a need to use a length offiber to measure the space-resolved stresses with absolute values(not relative) along the fiber for distributed sensingapplications.

In this Letter, we report a novel polarimetric optical fre-quency domain reflectometer (P-OFDR) for space-resolved transverse stress or bend measurements along an opticfiber, which does not require the FUT to be physically stable.The method is based on measuring the local birefringence in-duced by transverse stress via the photoelastic effect in the fiber.By inducing stresses via fiber bending at multiple points alongthe optical fiber, we demonstrate that our system can simulta-neously obtain both the distributed stress and the reflectioncurves as a function of distance along the fiber, unambiguouslydistinguishing the stresses from the optical back-reflections.Our preliminary experiments show that our prototype systemhas a spatial resolution of 0.5 mm, a measurable distance ofover 800 m, a maximum stress level of up to 200 kpsi�1 kpsi � 6.9 MPa�, a minimum stress (or sensitivity) ofabout 10 kpsi, and a stress measurement uncertainty around10%. Note that unlike the cross-correlation method in [14],in our system, the temperature does not affect the stress or bi-refringence measurement, unless the temperature change indu-ces a real stress on the fiber via other effects, such as the thermalexpansion of a bobbin beneath the fiber or the thermal crimp-ing of adhesives around the fiber. Therefore, our reported tech-nique can also be implemented for the distributed fiber sensingapplications for detecting the value of stress along the fiber,although it cannot differentiate whether the stress is appliedor thermally induced. As an example for the distributed sensingapplications, we show that our prototype can clearly identifytransversal stress points in a fiber coil induced during the wind-ing process when the fiber transitioning from one layer to theother. The reported method is not sensitive to the state ofpolarization (SOP) variations from fiber cable movement sothat potentially it is well suitable for the field fiber networkhealth monitoring and fiber sensing applications.

In the sections below, we describe the hardware configura-tion and the theory behind our P-OFDR for measuring thespace-resolved transverse stress along an optical fiber withthe advantages of single-end operation, high spatial resolution,high stress range, and insensitivity to temperature and SOPvariations in comparison to other measurement techniques[10–15]. Figure 1 shows the setup for our P-OFDR measure-ment system, where a tunable laser with a linewidth around100 kHz is continuously tuned up to 4 nm around 1550 nm.A small fraction of light ∼5% from coupler 1 is split and thensent to a Faraday rotation mirror (FRM) based Michelson inter-ferometer with a relative time delay of 10 μs for generatingclock signals at each “zero-crossing” position of the optical in-terference fringe signals. The generated real-time clock signalswere then used for the P-OFDR raw data sampling. The lightwas then further split by a 90/10 coupler (coupler 3), in which10% of the light was used as a local oscillator light to coherentlyprobe Rayleigh back scattering (RBS) light, and 90% of the

light was then sent to the fiber under test (FUT) via a circulator.The SOP before entering and after exiting the FUT is ran-domly varied by an all-fiber polarization controller to uniformlycover the Poincaré sphere. The Rayleigh backscattered lightfrom different locations of the fiber is brought to interfere withthe portion of light directly from the tunable laser via a polari-zation diversity interferometer (the dotted-line box in Fig. 1).The interference signals of the two polarization components areanalyzed after detection by PD2 and PD3, and the local bire-fringence information is then calculated. The transverse stressrelated to the birefringence is then obtained using the signalanalysis procedures described below.

The optical fibers with non-perfect circular symmetry orwith bend, twist, or other transverse stress often exhibit localbirefringence. Under assumption of no polarization-dependentloss (PDL), the light passing through a section of optical fiberwith birefringence will experience a variation of SOP. When apolarized light is propagating in a birefringence fiber, the evo-lution of the light polarization state can be described using thewell-known equation of motion of the Stokes vector, S⃗�z;ω�,which can be expressed as [17,19]

ΔS⃗�z�Δz

� W⃗ rt�z� × S⃗�z�; (1)

where z is the distance of light propagated within the fiber, ω isthe optical frequency, W⃗ rt�z� is the local round-trip birefrin-gence vector. If the evolution of light’s SOP can be analyzedalong the fiber length, then the local birefringence can bededuced. In particular, we found that the W⃗ rt�z� can be deter-mined from a mean-square value of differences of local trans-mission between two closely spaced fiber sections, similar to[13] in which the local birefringence was obtained byanalyzing the SOP evolution between two closely spacedfrequencies of light. Assuming the local birefringence W �z� �W rt�z�∕2 and following the same procedure as in [13], exceptsubstituting the changes in optical frequency ω in [13] withΔz, we obtain

W �z� � 1

2Δz·

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi15hΔT 2�z�iI∕O-SOP

q; (2)

Fig. 1. Experimental setup of distributed transversal stress measure-ment using a P-OFDR. PBS, polarization beam splitter; PD 1–3,photo detectors. Note that the pigtails of couplers 1 and 3 are of polari-zation maintaining (PM) to avoid polarization fluctuations when lightinterferes. PBS 1 and 2 are free-space polarization beam splitters; BS 1and 2 are free-space polarization insensitive beam splitters. The PMfiber inputting to PBS2 is oriented such that the power is equally splitby PBS2.

2820 Vol. 41, No. 12 / June 15 2016 / Optics Letters Letter

where ΔT �z� � T �z � Δz∕2� − T �z − Δz∕2� is the normal-ized light power difference between two positions at z � Δz∕2and z − Δz∕2, and T �z� � Pi�z�∕�P2�z� � P3�z�� is the nor-malized light power at the location z, where Pi�z� is the lightpower of one polarization component from PBS1 at z measuredwith one of the detectors PDi �i � 2; 3�. In Eq. (2),hΔT 2�z�iI∕O-SOP stands for averaging over random inputand output SOP variations. The fiber transverse stress as a func-tion of the distance can be extracted from measured distributedfiber birefringence [20,21].

A key aspect of our P-OFDR approach is that the back-reflected light from the FUT at the two neighboring locationsis subsequently detected, and their SOPs are analyzed for manyrandom and uniformly distributed input and output SOPs thatare uncorrelated with respect to each other. It is worthy to notethat because the input SOPs incident into and the output SOPexited from FUT are random and without specific requirement,the system is independent of the SOP variations from fiber cablemovement and, therefore, is well suited for field applications,where the SOP along the fiber is constantly changing [13].

An example of the measured reflection trace for an FUTlength up to 860 m is shown in Fig. 2, where the OFDRback-reflection curve was extracted from the sum of twosquared light intensities measured by two detectors (PD2and PD3), and its reflectivity was calibrated by an internalreference reflector (not shown in Fig. 1).

To demonstrate the measurement principle described in thesections above, we made an experimental test by using bending-induced transverse stresses along a FUT made with a bendinginsensitive SM fiber (Corning ClearCurve ZB) where eightbending loops at eight different FUT locations were designed.The bending radius is different at different positions, and theyare typically between 5 to 15 mm. In addition to the eightbends, there are also six FC/APC connectors and a fiber splice.The results are illustrated in Fig. 3. In our experiment, thepolarization controller randomly adjusted its launched SOP,and the receiver analyzed the received signal correspondingto 200 different polarization controller settings randomly dis-tributed on the Poincaré sphere. The tunable laser is scanned ina range of 4 nm with a scanning speed of 10 nm/s. The results,as shown in the upper and lower plots of Fig. 3, were acquiredsimultaneously by the same data sampling of the raw data, butwith different signal processing algorithms, where the reflectiontrace in the upper plot was computed by the sum of the squaredmeasured results from two photodetectors (PD2 and PD3 inFig. 1), and the fiber birefringence trace W �z� was extractedusing Eq. (2) with an average of over 200 inputs and analyzed

SOPs. The displayed stress value was obtained by a calibrationprocedure with a calibration weight. The measurement andcomputation take about 5 min to complete.

As we can see, in the reflection trace (the upper plot ofFig. 3), there are seven light back-reflections associated withsix connectors (L1, 2, and 4–7) and a fiber splice (L3). Of par-ticular interest is the stress measurement as shown on the lowerplot of Fig. 3 with features associated with fiber bendings, con-nectors, and splice. With the cross-reference of reflection trace,eight fiber-bending-induced stresses features S1–8 that can beunambiguously distinguished from those of light back-reflec-tions. It is worth noting that the induced transverse stress ishigher for smaller bend radii than for larger bend radii, consis-tent with the bending-induced birefringence calculations [20].All eight bending-induced fiber birefringence events are seencleanly with a signal-to-noise ratio over 2. The spatial resolu-tion of our system for the stress measurement is of 0.5 mm,determined by the wavelength scanning range of the laser.The measurement uncertainty from multiple measurementsis ∼10% when 200 randomly distributed SOPs were used,and it can be reduced by averaging more SOPs, but with longermeasurement time. The relative stress levels of all eight fiberloops are consistent with [20].

As an example for distributed stress sensing, we use our P-OFDR to identify stress locations and magnitudes along theoptical fiber inside a fiber coil, as mentioned in the introduc-tion. The information obtained can be used for improving thewinding process and minimizing such stresses for making betterfiber coils, such as a crossover free coil. To demonstrate its capa-bilities, our prototype system is used to measure a fiber coilwith a diameter of 80 mm and clearly identifies periodic stresspeaks, as shown in Fig. 4. The periodicity of 23 peaks is thesame as the length of each layer of the coil, which is consistentwith the expectation that stresses are induced at fiber crossover

Fig. 2. Measured OFDR reflection trace with back-reflections for aFUT with a length up to 860 m.

Fig. 3. P-OFDR signal from a length of fiber with six connectorsand one splice for introducing reflections and eight fiber loops withdifferent diameters for inducing different amount of stresses. Uppercurve, measured reflections L1–7; lower curve, measured fiber-bending-induced stress S1–8.

Letter Vol. 41, No. 12 / June 15 2016 / Optics Letters 2821

points when the fiber transitioning from one layer to the otherduring the winding process. There are 23 peaks in total, cor-responding to the 23 layers of fiber in the fiber coil. To furtherprove that the periodic peaks are, indeed, from the coil wind-ing, we spliced a length of fiber around 130 m loosely woundon a large spool of about 50 cm onto the coil and perform thedistributed stress tests from both ends of the combined fibers Aand B, with the A side being the large spool and the B sidebeing the fiber coil. Indeed, we only observed the periodic stresspeaks in the fiber coil, not in the loosely wound fiber spool, nomatter which direction we launched the light. This preliminarytest result of transversal stresses on the fiber coil indicate thatour P-OFDR is sufficiently sensitive to detect such a smallwinding induced stresses and is promising for the productionprocess control to improve the fiber coil quality. All measure-ments are conducted at room temperature.

In conclusion, we demonstrated a novel polarimetric OFDRsystem that can measure space-resolved transversal stressesalong an optic fiber from a single-end of an optic fiber or fibercable, independent of the temperature of the FUT. By inducingtransversal stresses and reflections at multiple points along anoptic fiber, we demonstrated that our P-OFDR prototype sys-tem can simultaneously obtain both the distributed stress andreflection curves as a function of distance from the same acquis-ition, unambiguously distinguishing the stresses from any lightback-reflections, including the Rayleigh backscattering andFresnel back-reflection. Our P-OFDR system can measure fi-ber transversal stress or bend for over 800 m with a spatial res-olution of 0.5 mm, a maximum stress of up to 200 kpsi(1379 Mpa), a minimum stress of 10 kpsi (69 Mpa), and ameasurement uncertainty of 10%. The latter two can be re-duced with lower noise electronics and more SOP averages.Stress larger than 200 kpsi can be achieved by increasing thewavelength scanning range and, hence, enhancing the spatialresolution. We expect the measurement range can be increasedto over 20 km with a tunable laser with a linewidth less than5 kHz with appropriate frequency turning linearity [22]. We

further demonstrate that our system is sufficiently sensitive toidentify stress points in a fiber coil induced during the windingprocess when the fiber transitioning from one layer to the other,and can be used for fiber gyro coil inspection applications.Finally, our P-OFDR system is robust against SOP variationscaused by fiber cable movement and is well suited for the fieldmeasurement of stresses in the fiber and, therefore, is ideal forthe fiber infrastructure health monitoring and for distributedfiber optical sensing applications. Note that unlike the reportsin [16–18], we aim at detecting the high birefringence inducedby large stresses, which may cause the premature failures ofoptical fibers, especially the bending insensitive fibers withnegligible bending losses. Portion of this Letter were presentedat OFC2016 [23].

Funding. Comprehensive Capability EnhancementProgram of Regional Universities in Central Western China;Advanced Talents Program of Hebei Educational Committee(GCC20144020); International Science and TechnologyCooperation Program of China (2010DFB13180,2014DFA12930); Foundation Research Project of JiangsuProvince (BK20130373, BK20130374); InternalDevelopment Funding of General Photonics Corporation.

Acknowledgment. We thank the fruitful discussionswith Dr. Zhenyang Ding of Tianjin University.

REFERENCES

1. B. Soller, D. Gifford, M. Wolfe, and M. Froggatt, Opt. Express 13, 666(2005).

2. D. P. Zhou, Z. Qin, W. Li, L. Chen, and X. Bao, Opt. Express 20, 13138(2012).

3. X. Bao and L. Chen, Sensors 12, 8601 (2012).4. Z. Ding, X. S. Yao, T. Liu, Y. Du, K. Liu, Q. Han, Z. Meng, and H. Chen,

Opt. Express 20, 28319 (2012).5. Z. Li, Z. Meng, L. Wang, T. Liu, and X. S. Yao, IEEE Photon. Technol.

Lett. 27, 549 (2015).6. Z. Li, Z. Meng, T. Liu, and X. S. Yao, Opt. Express 21, 2521 (2013).7. Z. Li, X. S. Yao, X. Chen, H. Chen, Z. Meng, and T. Liu, J. Lightwave

Technol. 33, 372 (2015).8. B. Szafraniec andG. A. Sanders, J. Lightwave Technol. 17, 579 (1999).9. H. Chen and M. LeBlanc, IEEE Photon. Technol. Lett. 16, 2198 (2004).10. Y. G. Lu, X. P. Zhang, Y. M. Dong, F. Wang, and Y. H. Liu, J. Phys.

Conf. Ser. 48, 1387 (2006).11. Y. Dong, L. Chen, and X. Bao, Opt. Lett. 36, 277 (2011).12. A. Galtarossa, D. Grosso, and L. Schenato, IEEE Photon. Technol.

Lett. 20, 1944 (2008).13. N. Cyr, H. Chen, andG. Schinn, J. Lightwave Technol. 27, 4164 (2009)14. M. Froggatt, B. Soller, D. Gifford, and M. Wolfe, in Optical Fiber

Communication Conference (OFC) (Optical Society of America,2004), paper PDP17.

15. X. Fan, Y. Koshikiya, and F. Ito, Opt. Lett. 35, 25 (2010)16. L. Palmieri, A. Galtarossa, and T. Geisler, Opt. Lett. 35, 2481 (2010)17. L. Palmieri, Opt. Fiber Technol. 19, 720 (2013).18. A. Galtarossa, D. Grosso, and L. Palmieri, IEEE Photon. Technol.

Lett. 21, 1713 (2009).19. P. Wai and C. R. Menyuk, J. Lightwave Technol. 14, 148 (1996).20. R. Ulrich, S. C. Rashleigh, and W. Eickhoff, Opt. Lett. 5, 273 (1980).21. A. Galtarossa, L. Palmieri, A. Pizzinat, M. Schiano, and T. Tambosso,

J. Lightwave Technol. 18, 1389 (2000).22. Z. Ding, X. S. Yao, T. Liu, Y. Du, K. Liu, J. Jiang, Z. Meng, and H.

Chen, Opt. Express 21, 3826 (2013).23. D. Z. Chen, X. S. Yao, C. Wei, H. Chen, X. Chen, and Z. Li, in Optical

Fiber Communication Conference (OFC) (Optical Society of America,2016), paper W2A.29.

Fig. 4. Distributed stress measurement of a 250 m fiber coil connectedwith a loose fiber about 130m in length. Upper, light enters the fiber fromthe coil side B; Lower, light enters the fiber from the spool side A. Thefiber used is the bend insensitive fiber Corning ClearCurve ZBL single-mode optical fiber, and the tension for winding the coil is about 10 g.

2822 Vol. 41, No. 12 / June 15 2016 / Optics Letters Letter