Interferometric optical time-domain reflectometry for distributed optical-fiber sensing Sergey V. Shatalin, Vladimir N. Treschikov, and Alan J. Rogers The technique of optical time-domain reflectometry is analyzed to determine the effect of an optical phase modulation on light backscattered in an optical fiber. It is shown that the spatial distribution along the fiber of an external phase modulation can be measured with a spatial resolution close to that of optical time-domain reflectometry. A distributed interferometric sensor arrangement that employs this technique is investigated experimentally, and a satisfactory interrogation of more than 1000 resolution intervals is demonstrated. 1998 Optical Society of America OCIS codes: 060.0060, 030.1670, 060.2310, 060.2370, 060.2430, 060.5060, 120.3180. 1. Introduction Information concerning the distribution of attenuation along an optical fiber can be obtained by repetitive launching of an optical pulse into the fiber and subsequent analysis of the backscattered light. 1 The technique, known as optical time-domain reflectometry OTDR, has become established as a useful tool for investigation of optical-fiber communications systems. OTDR has been extended also to include measurements of the distribution along the fiber of a variety of external fields e.g., temperature, strain, magnetic, and electric fields that are capable of influencing one or more of the light s characterizing properties, such as power, spectrum, or polarization state, while the light is propagating in the fiber. 2 Here we study the interference effects that are associated with the OTDR backscatter signal, and then we use them to design a distributed sensor system. Distributed optical-fiber sensors have enormous potential for industrial application: they provide a new dimension of measurement for the monitoring and control of extended structures and process plant. Optical-fiber interferometric sensors offer extremely high sensitivity 3 5 but suffer from some complexity; it is this latter that has impeded their general applica- tion. The present idea provides the basis for a straightforward, distributed, interferometric, opticalfiber sensing system, and it could have wide application. 2. Theory Other published research 1 6 has studied the interference among various portions of the backscattered signal in OTDR in the context of the noise level imposed on the wanted OTDR signal. Our intention here is to show that these interferences allow a determination of the form and the location of external phase disturbances along the fiber. Consider a coherent optical pulse t that is launched into a single-mode optical fiber of length L in an OTDR arrangement Fig. 1. For simplicity, we assume in the first instance that an external source of phase modulation acts at one point along the fiber, producing an optical phase change of at a distance z 0 from the launch end. The backscattered optical electric field E, t at time t for the light reemerging from the launch end can be expressed as a superimposition of partial fields backscattered along the fiber. For a speed of light in the fiber and wave propagation constant, we utilize group and phase delays 2z and 2z, respectively, to express the emerging field in the form S. V. Shatalin and V. N. Treschikov are with the Institute of Radio Engineering and Electronics, Vvedensckogo Square 1, Fryazino 141120, Russia. A. J. Rogers is with the Department of Electronic Engineering, King s College London, Strand, London WC2R 2LS, UK. Received 2 December 1997; revised manuscript received 27 April 1998. 0003-693598245600-05$15.000 1998 Optical Society of America with E, t E 1 t E 2 texp2i, (1) E 1 z 0 t 2zrzexp2izdz, 0 5600 APPLIED OPTICS Vol. 37, No. 24 20 August 1998
Fig. 1. Arrangement for OTDR. E 2 L t 2zrzexp2izdz. z 0 The optical wave intensity at the OTDR output thus takes the form where I, t E, t 2 I 1 t I 2 t 2I 1 ti 2 t 12 cos2 0, (2) 0 arge 2 arge 1 I i E i 2, i 1, 2. The form of Eq. 2 clearly is that of an interferometer signal, and we must now evaluate the last interference term. We assume that the backscatter process in the fiber is due to Rayleigh scattering and that the scatter coefficient rz takes the form of a random Gaussian function with negligible spatial correlation, i.e., rzru z u, where the triangular brackets represent an ensemble average and is the Dirac delta function. For an incoherent source of light such as is normally used in OTDR systems 0 varies randomly so that in this case It I 1 t I 2 t. However, for a coherent source the interference term in Eq. 2 must be included. To derive the relationship between the interference and the position at which the external modulation is applied, we define an interference visibility as 2I 1 ti 2 t 12 V. It It is shown in Appendix A that this can be expressed in the form 2 12 t 2z 0 Vt 1, (3) z where z is the spatial resolution of the OTDR system, and we assume a rectangular input pulse of the form z z 1; 0 t, 0; t. Fig. 2. a OTDR trace. b Filtered trace when periodic phase perturbations are imposed. Equation 3 clearly indicates that Vt is real only when t 2z 0 z, in other words, for a time limited by z z z 0 t z 0. 2 2 Hence the interference becomes visible only for a time corresponding to the spatial resolution interval around the position of the applied phase perturbation. Thus the OTDR arrangement can, with a coherent light source, be used as a distributed sensor, with spatial resolution equal to that of OTDR. Clearly, for good visibility the coherence length of the laser should be greater than the spatial resolution interval of OTDR. This is not the case for commercial OTDR systems. For these systems the above results hold only for a narrow spectral range, and one can assess their performances as interferometers by averaging over the full spectral width of the source. Inevitably, this leads to reduced visibility of the interference pattern. 3. Experimental Results The experimental OTDR arrangement is shown in Fig. 1. The source was a Q-switched, diode-pumped, yttrium aluminum garnet laser with low levels of amplitude and phase noise. The laser used was a Lightwave Electronics 11 with a mean power of 10 mw at 1520 nm, and pulses of 7-ns width were generated at a 2-kHz repetition rate. The OTDR output was measured with a p-i-n photodiode and a boxcar integrator with a 30-ms time constant. Figure 2a shows measured OTDR traces for a fiber of 21-m length. The signal fluctuations evident in the figure were observed previously in coherent 20 August 1998 Vol. 37, No. 24 APPLIED OPTICS 5601
Fig. 3. Integrated effect of a periodic longitudinal extension. OTDR s and were shown to be due to the coherent addition of random waves from resolution-interval sections of fiber. They are clearly not due to shot noise or random variations in scatter conditions because they remain stable in form for several seconds. This demonstrates that the coherence length of the source is large enough to provide visible interference for the spatial resolution of OTDR. Effectively, the system consists of a series of Fabry Perot interferometers. We applied phase perturbations to the fiber, at points 7 and 16 m from the front end, by winding several turns of the fiber around piezoceramic cylinders at these positions. The electrical signals applied to the cylinders were at a frequency of 10 khz and had amplitudes of 10 and 1 v, respectively. The OTDR signal was passed through a 10-kHz filter to provide the output shown in Fig. 2b. The heights and positions of the two peaks correspond well with those of the applied perturbations. The two signals were demonstrably independent, and cross-talk suppression was measured as greater than 25 db. The spatial resolution of the sensor is seen to be 0.7 m, which is also that of the OTDR system. The relationship between the elongation of the fiber and the OTDR trace variation is shown in Fig. 3. The integration time for this measurement was equal to the time delay corresponding to the position of the first phase modulator. The evident variation of output intensity is similar to that of a two-beam interferometer and thus is in accord with the implications of Eq. 2. 4. Results from Commercial Optical Time-Domain Reflectometry To demonstrate the ready applicability of these ideas to conventional OTDR techniques, we applied them to a commercially available system, the AQ-714OD, operating at 1310 nm. This used a 100-ns 10-m spatial resolution light pulse that we launched into a single-mode fiber of 11-km length. The output was analyzed with a dedicated PC for a system response time of 2 s, allowing slow phase modulations such as Fig. 4. Effect of a localized rise in temperature: a temperature profile, b OTDR traces, c differential trace with the temperature perturbation. those caused by temperature variations, for example to be detected. Figure 4b shows a typical output trace from this OTDR. It has a form similar to that expected from a weakly coherent source. We checked this by substituting the 1310-nm source by one operating at 1520 nm, and by showing that a different fluctuation pattern resulted. We then imposed a slow phase disturbance by heating a 30-m coiled section of the fiber, 6 km from the fiber input end, at a rate of 2 Ks 1 Fig. 4a. The resulting OTDR trace Fig. 4c was calculated as a variance of the original trace Fig. 4b. The correspondence with the phase perturbation of Fig. 4a is impressive and confirms that the fluctuations are indeed due to interference. The measured sensitivity of the temperature change was 0.2 Ks 1. Clearly, also as seen in Fig. 4c, the spatial resolution of the measurement was close to that of OTDR at 10 m. This implies an effective 1000 separate temperature perturbation measurements over a 10-km fiber length. Such a performance compares well with the frequency-modulated continuous-wave FMCW approach to distributed interference sensing. 3 5 In the latter only 30 measurements were demonstrated, with cross-talk suppression of 17 db 3 compared with our results, which give 1000 measurements and demonstrate cross talk better than 25 db. Moreover, our results relate to a commercial system rather than to a laboratory demonstration. On the other hand, however, the FMCW approach has better sensitivity, owing to its use of coherent rather than direct detection, giving a better signal-to-noise ratio. A FMCW has been shown to be capable of detecting a temperature rise rate of 14 mk s 1. 5 This result 5602 APPLIED OPTICS Vol. 37, No. 24 20 August 1998
serves to demonstrate that our system has considerable room for improvement of performance by giving suitable attention to the coherence of the source and to improved detector design. 5. Applications The indeterministic amplitude of the quasiinstantaneous interference signal from the OTDR imposes a limitation on the value of this technique as an accurate indication of perturbation level. The indeterminism can be alleviated by signal averaging over a larger spatial interval, but this, of course, degrades the spatial resolution. The most promising applications are those that require primarily determination of only the location of a disturbance rather than its amplitude. One example of this is that of an intruder alarm; another is the security monitoring of a communications system. Clearly, as has been demonstrated with the rate-oftemperature-rise results, fire alarms also are possible. 4,5 6. Conclusions We have proposed and demonstrated a new method for sensing the location of a phase perturbation along the length of a single-mode optical fiber. The method is based on the well-known technique of OTDR. The technique can provide a spatial resolution of the order of 1 m and can give some indication of the level of perturbation. Using a commercially available OTDR, we have demonstrated the interrogation of more than 1000 sensing points, and with additional developments improvements in performance can be anticipated. The technique appears feasible and could form the basis for readily implementable intrusion and fire alarms and for distributed temperature-sensing systems. Appendix A To calculate I 1 I 2 we consider the statistical properties of E 1,2. These fields are the sums of a large number of independent random variables and hence become Gaussian random variables. In accordance with the complex Gaussian moment theorem see p. 44 of Ref. 7, we can write I 1 I 2 I 1 I 2 E 1 *E 2 E 2 *E 1. z 0 1 2z 2u t * t (A1) It is easy to see that the second term in Eq. A1 is negligible: E 1 E 2 * 0 z 0 rzruexp2izududz. (A2) rzru 2 z u 0 for z u only, but at that time 0 z z 0, z 0 u L. Summarizing I 1 ti 2 t 12 V 2, (A3) I 1 t I 2 t we can calculate this using the usual OTDR trace expressions. For this purpose we transform Eq. 1 into convolution forms: 2z 0 E 1 t t t rt, 2 2z 0 2L E 2 t t t rt, (A4) 2 where phase terms are omitted, is the Heaviside step function, and R is the convolution symbol, i.e., f 1 f 2 f 1 f 2 t d. Taking into account the correlation properties of rt 1 rt 2 2 t 1 t 2, we can obtain the intensity expression in the same way as for the optical incoherent image see p. 109 of Ref. 7: 2 2z 0 I 1 t t 2 t t, 2 2 2z 0 2L I 2 t t 2 t t. (5) 2 If we choose for the estimation the rectangular impulse form z z t 2 t t, then for the time domain in which I 1 I 2 0 we have I 1 t 2 2 z 1 t 2z 0 2 2z I 1 t 2 2 z 1 t 2z 0 2 2z,. (6) Substituting Eq. A6 into Eq. A3, we obtain Eq. 3, i.e., 2 12 t 2z 0 V 1. z The authors thank R. Juskaitis for his contribution in the early stages of this study and V. Handerek for his help in the experiment. S. Shatalin and V. Treschikov gratefully acknowledge D. Sedykh, Telecomservice, Ltd., for supplying the OTDR, and V. Potapov for valuable discussions. This study was supported in part by Russian Fundamental Research Foundation grant 96-02-1-9712. References 1. P. J. Healy, Review of long-wave singlemode optical-fibre reflectometry techniques, J. Lightwave Technol. 3, 876 890 1985. 2. A. J. Rogers, Polarization-optical time domain reflectometry: a technique for the measurement of field distributions, Appl. Opt. 20, 1060 1071 1981. 20 August 1998 Vol. 37, No. 24 APPLIED OPTICS 5603
3. R. Juskaitis, A. M. Mamedov, V. T. Potapov, and S. V. Shatalin, backscattering in an optical fiber, Opt. Lett. 19, 593 595 A distributed interferometric fiber sensor system, Opt. Lett. 1994. 17, 1623 1625 1992. 6. K. Shimizu, T. Horidichi, and Y. Koymoda, Characteristics and 4. R. Juskaitis, A. M. Mamedov, V. T. Potapov, and S. V. Shatalin, reduction of coherent fading noise in Rayleigh backscattering Interferometry with Rayleigh backscattering in a single-mode measurements for optical fibres and components, IEEE J. fiber, Opt. Lett. 19, 225 227 1994. Lightwave Technol. 10, 982 987 1992. 5. R. Rathod, R. D. Peschtedt, D. A. Jackson, and D. J. Webb, 7. J. W. Goodman, Statistical Optics Wiley, New York, 1985, Distributed temperature-change sensor based on Rayleigh Chap. 2. 5604 APPLIED OPTICS Vol. 37, No. 24 20 August 1998