Recently, attenuation measurements by optical time domain reflectometry (OTDR) have been investigated to better understand the theoretical limits of this technique1 and to obtain high spatial resolution, high dynamic range, and, when requested, limited measurement time.
Correlation techniques have been proposed2,3 to solve the problems involved in pulsed OTDR arrangements. The optical energy necessary to perform a measurement can be spread over a long time interval instead of being concentrated into a single short pulse. This overcomes the need for high peak power laser diodes, not yet commercially available at 1.3-1.5 μm. The practical feasibility of this approach has been shown in the experimental arrangements proposed so far. Nevertheless, the achieved spatial resolution is still poor. In this Letter, experimental results will be presented to show how this technique also allows high spatial resolution, comparable with that attainable by pulsed techniques, with low cost electronics.
The general properties of the so-called pseudorandom sequences (PRS) have been extensively analyzed4-6 and will not be discussed here. For our purposes, the main feature of a pseudorandom sequence f{t) concerns its autocorrelation function (ACF). It holds
where k is an integer, n is the number of bits of the PRS, T = nΤ is the PRS period, and 1/τ is the PRS transmission rate.
According to Eq. (1), CT will be considered as the spatial resolution of the measurement technique, being c the light velocity in the fiber. As a consequence, an optical fiber of length L can be thought of as a chain of M = L/cτ independent sections. Thus, if the period T is longer than twice the light propagation time in the fiber, any two different fiber sections will echo two practically uncorrelated PRSs. By correlating the whole backscattered signal with a 2kτ delayed
Fig. 1. Schematic of the attenuation measurement setup.
Fig. 2. Details of the correlator and the integrate and dump stages.
Fig. 3. Experimental results on a 1.1-km long fiber. Symbols represent measured values; lines are for visual aid only.
version of the transmitted PRS, one can recognize the contribution of the kth section.
To characterize the setup from the point of view of its sensitivity, one must investigate causes and effects of noise. A first noise source arises from electronics. However, owing to the coherent processing, the very narrow equivalent bandwidth of the receiver allows us to neglect it with respect to the interfering signal coming from all the remaining (M — 1) sections. If the fiber is very long, these contributions limit
the maximum measurable fiber length. A detailed analysis of this point is reported in Ref. 7 where some encouraging conclusions are drawn. In the hypothesis of an exponential decay of the backscattered signal, the signal-to-interference ratio S can be easily calculated. As an example, at the transmission rate of 10 Mbit/sec, with a PRS of (228 - 1) bits, S is >40 dB at the end of a 20-km long fiber with 1-dB/km losses. Moreover, one can show that, for a given fiber length, a further reduction of the effects of the interfering signal can be obtained simply by increasing the PRS length, i.e., the measurement time.
To test the experimental feasibility of this setup at high transmission rates, the apparatus, schematically illustrated in Fig. 1, was realized with the characteristics of the theoretical example. Note that a transmission rate of 10 Mbit/sec corresponds to a spatial resolution of CΤ = 20 m, comparable with that achievable using classical pulsed OTDR techniques. Measurements were performed on a multimode fiber ~ l km long with a break in the middle. The cw laser diode, operating at λ = 850 nm (NEC 3205P), is modulated by the PRS generated by the PRS generator 1. The backscattered signal is detected by a PIN photodiode (HP 5082-4203) ac coupled to a wideband amplifier. The PRS generator 2 drives the correlator and controls the integrate and dump filter. This filter was introduced to fully take advantage of the PRS property [Eq. (1)]. The integrating time equals the sequence period T = 27 sec. In Fig. 2, a more detailed schematic of the Correlator and Integrate and Dump sections is reported. The correlator has been realized with a commercially available integrated circuit (MCl455). It is a double-gated wideband amplifier. Its output is proportional to the input signal if channel A is selected. On the contrary, it is proportional to the inverted input signal if channel B is enabled. If G is the voltage gain, the gated amplifier multiplies in each time slot τ the input signal by + G or - G. At the end of the integrating period T, the output is proportional to the kth section back-scattered signal k, depending on the delay of the second PRS generator.
Results of two different measurements performed on the same fiber are reported in Fig. 3. In the two plots, the reflections due to the break and the fiber end are recognizable very precisely. No mode scrambler was used in the launching section. In this way it was possible to stress the effects of the fiber break under different launching conditions. Different groups of modes were excited changing the relative position of the laser diode with respect to the input section. As a complete mode mixing is not yet achieved at the break, it acts differentially in the reflection and transmission of the two sets of modes. This explains the differences between the two backscattered signals.
An OTDR setup based on correlation techniques has been proposed and realized. This technique allows high spatial resolution combined with high dynamic range. The apparatus was implemented with commercially available electronics, and a pulse repetition rate as high as 10 Mbit/sec was realized. An experimental example of high precision fault location was presented to confirm the viability of the approach.
Part of this work has been sponsored by CEAT Cavi SPA under a contract of cooperation with the University of Bologna. A complementary sponsorship provided by the Italian National Research Council is acknowledged as well.
References 1. M. Eriksrud, A. R. Mickelson, S. Lauritzen, and N. Ryen, in Pro
ceedings, Symposium of Optical Fiber Measurements, Boulder, Colo. (1982), pp. 63-66.
2. K. Okada, K. Hashimoto, T. Shibata, and Y. Nagaky, Electron. Lett. 16, 629 (1980).
3. P. Healey, in Technical Digest, Seventh ECOC, Copenhagen (1981), paper 5.2.
4. R. C. Dixon, Spread Spectrum Systems (Wiley, New York, 1976).
5. R. A. Scholtz, IEEE Trans. Commun. COM-25, 748 (1977). 6. G. A. Korn, Random-Process Simulation and Measurements
(McGraw-Hill, New York, 1966). 7. M. Zoboli, P. Bassi, P. Pongetti, and E. Reniero, in Proceedings,
Fourth National Conference on Applied ElectroMagnetism, Firenze (1982), paper F0-7.
