A simplified implementation of coherent optical time domain reflectometry (C-OTDR) enables distributed sensing of temperature, strain, or vibration in km-scale length optical fibers.  Distributed sensing with optical fibers has grown dramatically in the past 15 years, with applications in monitoring the health and status of oil wells, aircraft components, wind turbines, roads, and bridges, and in security perimeter sensing. OTDR, the oldest, most well-established distributed sensing technique, is based on measuring backward Rayleigh scattering of an optical pulse; since the width and group velocity of the pulse are known, the measured OTDR time trace maps on to position within the fiber, with shorter pulses having higher spatial resolution.  When the pulsed source is coherent, one can recover phase information from the Rayleigh backscatter, and therefore sense path-length changes in the fiber, e.g. due to temperature or strain.  Since the phase also depends on the wavelength, if one takes C-OTDR traces at multiple wavelengths and calculates their mutual coherence, one can extract both path length changes and the sign of this change.

The first implementation of this technology required a microwave synthesizer to frequency sweep the optical source, coherent detection, and a 3-hour measurement time.  More recently, other groups have demonstrated the same functionality with direct-detection (no local oscillator), but still with microwave-synthesized frequency sweeps and a data analysis requiring a large number of C-OTDR traces (4 s measurement time), or with chirped pulses requiring high bandwidth detectors. In the demonstration by Liehr et al., only two wavelengths are used (generated by current-modulation of an inexpensive diode laser) in an alternating pattern, and the data is extracted from time correlation of adjacent C-OTDR traces measured with a low-bandwidth p-i-n photodiode.  This method allows for fast measurement times limited by the fiber length, though in the experimental results presented, the traces are averaged over milliseconds and the correlation is determined over a 2.5 minute time window.  Still, the experiment is simple to implement and shows promise for practical structural health monitoring.

---

---

 Add Comment

You must log in to add comments.