Locating bad PMD sections with a Polarization-OTDR

Transcription

1 Locating bad PMD sections with a Polarization-OTDR Johan Visser*, Michel Leblanc, Redha Salmi, Ann Conibear and Andrew Leitch *Telkom SA, Private bag X74, Pretoria, 0001, South Africa. +27 (0) ; fax: +27 (0) visserfj@telkom.co.za. Abstract High Polarization Mode Dispersion (PMD) values pose an obstacle to the upgrading of links for higher transmission rates. Polarization Optical Time Domain Reflectometry (P-OTDR) measurements give an indication of the specific fibre sections that cause the high PMD values, allowing an informed decision to be taken regarding the upgrading of a link. In this work we combine POTDR measurements with PMD measurements to pinpoint the fibre sections giving rise to high PMD in an overhead optic fibre cable in the Eastern Cape. These are the first measurements of this nature made in South Africa. Index Terms P-OTDR, PMD. I. INTRODUCTION Polarization-mode dispersion (PMD) is a serious limitation in modern optical communication systems, particularly with the advent of 10Gb/s and 40Gb/s systems. This, coupled with the use of optical amplifiers to increase the distance between signal regenerators, has led to a significant decrease in the level of PMD that may be tolerated. The installation of new transmission systems on existing fibre links can cause major difficulties on account of the frequent occurrence of high PMD fibres. It has been observed that old fibres (for which the PMD was not even measured at the time of manufacturing) very often exhibit considerable variability in PMD even within a given cable. This variability between fibres suggests that the PMD may change significantly along the length of any given individual fibre. It is therefore of considerable interest to develop an instrument capable of identifying which fibre segment or segments are the main contributors to its total PMD. The concept of a Polarization-OTDR (P-OTDR) was introduced twenty years ago by Rogers [1], who described an OTDR sensitive to the state of polarization (SOP) of the backscattered signal. The simplest P-OTDR consists of an OTDR, in which a polarizer is introduced in the return path, just prior to its detector. Although initially developed as part of a fibre sensor system to monitor spatially varying external physical parameters (temperature, strain, etc.), there has recently been heightened interest in variants of this approach to measure the distributed PMD [2-4]. Each of these proposed techniques requires that the P-OTDR have sufficient spatial resolution to see the evolution of the states of polarization (SOP) as the light propagates down the fibre. The SOP of the light transmitted through a fibre is known to rotate about the birefringence axis at a rate that depends on the local birefringence of the fibre. The beat length, L b, represents the period of this rotation and is defined by λ L b = β c (1) where λ is the wavelength of light, β is the local birefringence of the fibre and c is the speed of light. For example, a birefringence of 1 ps/km corresponds to L b = 5 m. If such a fibre was measured with a P-OTDR and if the fibre birefringence were only linear (no circular birefringence), the backscattering signature would oscillate with a period of 2.5 m since the light travels forward and backward in the fibre. The beat length of the backscattered signal is therefore equal to L b /2. Accordingly, a P-OTDR would need a spatial resolution of roughly 1 m (or less) in order to discriminate the oscillations, requiring optical pulses of 10 nsec or less. Typical OTDRs do not have a large useful dynamic range with such a high spatial resolution, and the requirements on the OTDR performance become even more stringent for fibres having a higher birefringence. Consequently, these techniques are normally limited to low birefringence fibres. Alternatively, one could use a more complex P-OTDR system in order to improve the dynamic range at these resolutions. The use of EDFAs to boost the transmitted and received signal has already been proposed, and other means, such as coherent detection techniques or photon counting, could be explored. When upgrading an optical communication system to a higher bit rate, there is a large economic incentive not to upgrade the fibre itself. This generates considerable interest in being able to localize where the PMD originates within a given fibre link exhibiting an unacceptably large overall PMD value. A practical instrument must be able to measure when the birefringence is high ( for instance between 1 and 10 ps/km). Dynamic range must be sufficient to cover the range of the transmission system and ideally the hardware must be not overly complicated since it is a field application. Bearing these criteria in mind, initial work by Huttner [5] showed that a simple P-OTDR with limited spatial resolution could be used to detect high PMD sections. We will show in this paper more detailed analysis of this system as well as some successful field trials.

2 III THE PRINCIPLE OF P-OTDR The experimental set-up is shown in Fig.1. A rotating quarter-wave plate followed by a polarizer precede the detector. By taking four different P-OTDR traces, with an appropriate orientation for the quarter-wave plate and the polarizer for each trace, we can obtain a fully polarimetric measurement of the SOP evolution against distance. The rest of the P-OTDR consists of a standard OTDR except for the use of a different laser type. Standard OTDRs use broad linewidth laser sources (Fabry-Perot laser) to avoid coherence effects that cause ripples on the backscattering trace. This is not appropriate for a P-OTDR because as the accumulated PMD becomes important, the SOP of the different spectral components of the source evolve differently as they propagate down the fibre. This will cause an effective depolarisation of the transmitted and backscattered light. In order to avoid this problem a narrow-linewidth DFB laser has been used in the prototype. detector pulsed DFB laser polarizer λ/4 Fiber under test Fig.1: Optical configuration of the P-OTDR used for the experiments The degree of polarization (DOP) of the backscattered signal should remain close to unity when the P-OTDR pulses are shorter than the backscattering beat length. By observing the oscillation periods of the different Stokes components describing the SOP (S1,S2,S3), one can deduce the local birefringence of the fibre. However when the P- OTDR pulses are longer than the backscattering beat length there is some fading in the oscillations that lead to a decrease in measured DOP. We will call this phenomenon temporal depolarization and it is important to stress the fact that this depolarization is dependent on the local birefringence only and that it is not affected by the accumulated PMD. For a very long pulse, the backscattered DOP will tend toward 1/3 rather than zero. This result, which may appear surprising at first, arises from the fact that the backscattered signal maintains some memory of the input SOP: when the transmitted SOP is linear at the point of backscattering, the SOP of the scattered light will return to the same state as the input SOP. Therefore the backscattered SOP is not completely random. A complete explanation of this phenomenon is presented by van Deventer [6] along with arguments based on Mueller matrix calculations. We expect that the mean DOP measured with the P-OTDR will give some insight into the mean birefringence of a fibre. An estimate of the mean DOP can be calculated based on the fading in signal that is expected from the ratio between the P-OTDR resolution and the backscattering beat length by the following formulae DOP = Lp 1+ / 2 Lb where L p is the spatial resolution of the P-OTDR, as defined by the full width at half maximum (FWHM) of the impulse response of the P-OTDR, measured at a Fresnel reflection. This takes into account the optical pulse width and rise time as well as the receiver bandwidth of the electronic circuits. We will see that this simple estimate yields a relatively good estimate of the mean DOP but is not exact, since it does not take into account the impact of mode coupling In a standard telecom fibre, the birefringence vector β ( z) fluctuates randomly, both in amplitude and orientation, over the distance z in the fibre with the following statistics 1 L 0 L (3) β ( z) β ( z + ) 2 dz = β exp 2 h where β is the RMS average of the birefringence over the length L (calculated for = 0) and h is the coupling length of the fibre, defined as the length of fibre for which the spatial correlation of β ( z) decreases by 1/e 2. When the P-OTDR pulse is much longer than the backscattering beat length, we expect the DOP to decrease as already discussed. However for the case of fibre with very long coupling length i.e. h longer than the OTDR spatial resolution, the birefringence vector orientation changes very slowly with distance and the measured DOP(z) will depend on the angle between SOP(z) and β ( z) ; see Fig. 2 and 3. For example if the SOP and the birefringence axis are nearly aligned, the backscattered SOP will not cover a large part of the sphere and even if the P-OTDR pulse is large, the DOP will remains close to 1. In practice, for fibre with weak mode coupling (large h), the DOP is expected to vary slowly against distance with values between 0 and 1. At a given position z on the fibre, the DOP should depend on the cosine of the relative angle on the Poincare sphere between the SOP(z) and β ( z). It seems reasonable then that the rate of change of the DOP should be related to the rate of change of the orientation of β ( z), which is itself related to the coupling length h. We verify this with numerical simulations. 2 (2)

3 Fig. 2: Calculated backscattered SOP for a large angle between the SOP and the birefringence axis. Left: short P-OTDR pulse, the DOP is large. Right: long pulse, the oscillations collapse and the mean DOP is low. Fig. 3: Calculated backscattered SOP for a small angle between the SOP and the birefringence axis. Left: short P-OTDR pulse, the DOP is large. Right: long pulse, even if the oscillation collapse, the DOP remains high The PMD of a fibre depends on both the birefringence β and the coupling length h through the following approximation. PMD β L L h = (4) where L is the length of the fibre. Based on this equation one can evaluate what is the minimum β needed to generate a PMD of 1 ps in a 1-km fibre link for different h. For instance, a very large birefringence ( >6 ps/km) results when h < 25 m. It is therefore expected that most fibre segments having a PMD coefficient above 1 ps/sqrt(km) will correspond to cases where h is long ( > 50 m). Consequently, the detection of a long h value should be sufficient to identify the high-pmd sections of the fibre. Fig. 4. Birefringence needed to generate a PMD of 1ps in a 1km fiber for different coupling length

4 In summary then, a study of the DOP statistics gives an indication of the coupling length. When the coupling length is short (a situation where energy is frequently exchanged between fast and slow modes), PMD does not accumulate rapidly with distance. Conversely, long coupling lengths are indicative of high PMD sections. III TEST RESULTS A General route information This is a 66.43km overhead optic fibre cable. The cable was installed after The route in the Eastern Cape (Southern Region ) where it is subjected regularly to strong winds. B Cable design information ADSS (Aerial Dielectric Self Support) cable. Loose tube cable, with 8 tube elements each containing 6 fibres. Glass Reinforced Plastic central strength member and Poly-Aramid strength member. Fig. 6: OTDR trace for fibre 40 (low PMD) The measurements are sufficient to illustrate however, that there are no severe losses. Table 1 gives a summary of the detected splice points and the splice loss at that point for the two fibres under discussion. The event numbers refer to the events marked on the OTDR traces in fig. 5 and 6. TABLE I SPLICE LOSS AND DISTANCE S C Tests done OTDR test with EXFO OTDR module in the FTB 400 Main Frame. PMD test with an EXFO 5500 PMD analyzer (Interferometric method) and a polarized laser source. POTDR test using Exfo P-OTDR Measurements were performed on 11 fibres. On the basis of measured PMD values, two fibres, numbered 38 and 40 were selected as typical examples of high and low PMD respectively. The experimental measurements for these two fibres are presented in sections D, E and F in order to illustrate the analysis that may be carried out when considering the suitability of a cable for higher transmission rates. The PMD characteristics of all the measured fibres are summarized in section G. D OTDR Test and Splice distances The OTDR tests were performed with a 100ns pulse width, and the sampling time was set at 1minute. The tests were done at 1550nm wavelength. This is a uni-directional test; for more accurate splice losses, the OTDR test should be performed from both sides and the bi-directional splice loss for each splice should be determined E PMD testing Fibre No Event Dist. Event No. Km db db Start of Cable End of cable Fig.5: OTDR trace for fibre 38 (High PMD) To measure the PMD, a polarized source was used at the far end and the PMD was measured with an EXFO 5500 PMD analyzer. Figures 7 and 8 show the PMD inteferograms obtained. It should be pointed out that the model used by the 5500 PMD analyzer assumes a Gaussian shaped interferogram. In cases such as the one above, this is not completely satisfied, the PMD values reported have an associated error that could be of the order of 25%. For an accurate determination, measurements would have to be obtained over many input / output polarization states. It is nonetheless true, that for the purposes of this study the measurement indicates unacceptably high PMD (for 10

5 GBps transmission) in fibre 38, while the PMD of fibre 40 is such that it could safely transmit signals at 10 GBps without having adverse impact on the bit error rate. For fibre 38 illustrated in fig. 9, a bad section from between 40.6km and 49.7km is indicated as the main contributor, with another section further on the link flagged as being a further cause of high PMD. Comparison with table 1 shows that there are indeed splices at these locations, and so the fibre between these may be even be of a different type. Further work will include cutting out these fibre sections and checking the result on the PMD Fig. 7:PMD interferogram for fibre 38, with high PMD value of 19.4 ps. Fig. 10: P-OTDR trace for fibre 40 with indications of sections likely to have high PMD. Fig. 8: PMD interferogram for fibre 40, with low PMD value of 3.6 ps. F P-OTDR testing Fibres 37 to 47 were tested with the Exfo 1100 P- OTDR. The test results of fibres 38 and 40 are shown below. In the case of fibre 40, even though the total PMD is sufficiently low to transmit safely, a bad section between 40.6 and 42.8km could be changed to eventually allow 40GBps transmission. It can be noted that by replacing the cable between 40.6km and 49.7km with good fibre, will not only improve the PMD on fibre 38 which is already bad, but will also improve the PMD on fibre 40 even further. G Cable Analysis: Table 2 below summarizes the PMD results for all the fibres measured with the FTB The colour coding indicates fibres that are definitely unsuitable for 10 GBps transmission, while yellow indicates a warning. Generally a PMD value of 10 ps is taken as adequate for 10 GBps transmission. The reason for PMD values above 7 ps being flagged in this case, is the uncertainty in the PMD measurements. TABLE 2 Suitability of Fibres for 10 GBps Transmission Fig. 9: P-OTDR trace for fibre 38 with indication of regions having high PMD. The OTDR trace is accompanied by an indication of regions likely to have high PMD. These are inferred from the DOP statistics as described in the introduction. It should be stressed that independent PMD measurements need to be carried out to confirm the high PMD sections, as the P-OTDR gives an indication only. In all 11 fibers tested, the P-OTDR always flagged more or longer sections when the total PMD was high, which is of

6 course expected. The results for all 11 fibres are shown in fig 11. BIOGRAPHY Johan Visser is currently employed as a Technical Specialist: Optical Fibre Cable and Accessories. He is registered as a Professional Technician with the Engineering Council of South Africa visserfj@telkom.co.za. Michel Leblanc is Manager, Research and Development for Exfo in Quebec, Canada. michel.leblanc@exfo.com Fig. 11 Representation of regions flagged by P-OTDR Fibre color with reference to Table 2 It is quite obvious that changing the section between 40.6km and 49.5km would solve most PMD problems, and would allow safe 10GBps on most fibers. To be safe on even more fibers, section 40.6km all the way to the end of the link could be changed. It is speculated that since the high PMD regions all lie in a similar region in the link, the actual cable may be of a different type to the rest of the link. Further work is planned to investigate this by cutting out the problematic sections and making a PMD measurement of the resulting link sections. Redha Salmi is the Regional Sales Manager, Southern Europe, Africa and Middle East for Exfo. redha.salmi@exfo.com Ann Conibear is currently a contract lecturer in the Department of Physics at the University of Port Elizabeth.. phaabc@upe.ac.za Andrew Leitch is currently a full Professor in the Department of Physics at the University of Port Elizabeth. phaawl@upe.ac.za V ACKNOWLEDGEMENTS The Authors wish to thank Messrs Chris Nel, Tim Gibbon, Siphile Sibaya, and Ulricht van Antwerpen for assistance with making the measurements. The financial support of the UPE group by the National Research Foundation, THRIP, Telkom (Pty) Ltd., Aberdare Fibre Optic Cables, and Corning Optical Fiber is gratefully acknowledged. VI REFERENCES [1] A. J. Rogers, "Polarization-optical time domain reflectometry: A technique for the measurement of field distributions", Appl. Opt., vol.20, pp , 1981 [2] F. Corsi, A. Galtarossa, L. Palmieri, "Beat length characterization based on backscattering analysis in randomly perturbed single-mode fibers", J. Lightwave Tech., vol. 17, no. 7, pp ,1999 [3] H. Sunnerud, B.-E. Olsson, P. A. Andrekson, "Measurement of polarization mode dispersion accumulation along installed optical fibers", IEEE Photonic Tech. Letters, vol. 11, no. 7, pp , 1999 [4] M. Wuilpart, A. J. Rogers, P. Megret, M. Blondel, "Fully-distributed polarization properties of an optical fibre using the backscattering technique", Applications of Photonic Tech. (Photonics North 2000), SPIE 4087, pp , 2000 [5] B Huttner, B Gisin, N Gisin Distributed PMD measurement with a polarization-otdr in optical fibers,, [6] M.O. Van Deventer Polarization properties of Raleigh Backscattering in Single Mode Fibers, J. Lightwave Tech., vol 11, pp , 1993.