Raman distributed temperature sensors can now be calibrated with a single detector channel, according to a study that reports an attenuation correction method validated on a 450-meter sensing fiber. The technique, described by Ashok Kumar Sahoo, Jitendra Kumar and Om Prakash, achieved a temperature estimation accuracy of ±0.8 °C across a measurement range of 20–70 °C, a temperature resolution of 0.35 °C and a spatial resolution of 3.4 m. The work was published in the journal Discover Sensors on 30 October 2026.
Distributed fiber-optic sensors have attracted significant research attention because they can measure conditions continuously along an optical fiber, resist electromagnetic interference and withstand corrosive environments. Among these technologies, Raman distributed temperature sensing, or RDTS, has been studied extensively for tracking temperature profiles in power systems such as transformers and power cables, as well as in tunnels, oil pipelines and the coal industry. The operating principle relies on spontaneous Raman scattering combined with optical time domain reflectometry: short laser pulses are launched into the sensing fiber, and the backscattered optical signal is analyzed to determine the temperature distribution along its length. The temperature dependence of the Raman backscattered signal provides local temperature information, while the reflectometry technique enables spatial localization of these signals along the fiber.
The Raman backscattered light comprises two components, Anti-Stokes and Stokes. The intensity of the Anti-Stokes component exhibits a stronger dependence on temperature than the Stokes component: at room temperature of 293 K, the relative sensitivities are approximately 0.83% per kelvin for Anti-Stokes and 0.096% per kelvin for Stokes. Conventional RDTS systems use the ratio of Anti-Stokes to Stokes intensities in a single-ended configuration, in which one end of the fiber is interrogated while the opposite end remains free. Because both signals are measured, the ratio is independent of variations in laser power and detector gain. However, this dual-channel scheme suffers from two notable limitations. Wavelength dependent loss arises from differential attenuation between the Anti-Stokes and Stokes wavelengths, producing temperature errors particularly over long distances. The walk-off effect occurs because the two components, whose wavelengths typically differ by 100–200 nm, propagate at different velocities, resulting in positional errors along the fiber.
Several approaches have been proposed to address wavelength dependent loss. An attenuation correction technique using two light sources of different wavelengths was limited in practical applicability because it required strict wavelength matching between the sources. Double-ended dual-channel schemes interrogate both fiber ends through an optical switch to obtain forward and backward Raman traces, and dual-channel folded fiber configurations place different fiber segments in the same temperature environment. Although these configurations address differential attenuation, they require a sensing fiber twice the length of the measurement range. Signal processing methods based on wavelet decomposition and empirical mode decomposition have also been reported, but these yield inaccurate results when the hot zone width is large. Other approaches rely on curve fitting of measured attenuation differences or on stored Anti-Stokes and Stokes traces from a calibration stage.
Single-channel systems, which use only the Anti-Stokes signal, eliminate wavelength dependent loss and the walk-off effect and are more cost effective because they need only a single detector and amplifier. The challenge is that the attenuation experienced by the Anti-Stokes signal must be accurately compensated for reliable temperature estimation. Prior single-channel solutions have drawbacks: one method based on the Anti-Stokes signal and its reflection from a far-end mirror has a limited sensing range because the doubled optical path increases propagation loss; looped and folded fiber configurations require a sensing fiber twice as long as the measurement distance; and single-ended single-channel schemes reported previously either assume uniform attenuation along the entire fiber or rely on additional correction fibers and complex calibration procedures.
The new method consists of two main stages. In the calibration stage, the entire fiber is held at a uniform ambient temperature, and the backscattered Anti-Stokes signal is acquired with heavy averaging to capture the fiber’s attenuation profile, which is stored in memory. In the measurement stage, the unknown temperature profile is determined by dividing the measured Anti-Stokes trace by the stored attenuation profile. Because variations in laser output power, fiber coupling efficiency and detector gain appear as a single proportionality factor that does not vary with distance, they are eliminated by designating the initial few meters of fiber as a reference zone whose temperature is measured with a standard sensor such as an RTD or thermocouple. The temperature profile is then calculated from the attenuation-compensated signal using the known reference temperature. The resulting estimate is free from wavelength dependent loss, and because only the Anti-Stokes signal is used, spatial errors from the walk-off effect are inherently avoided.
The experimental setup used a pulsed laser generating 20 ns pulses at 1064 nm with 100 µJ pulse energy and a 1 kHz repetition rate, coupled into a 62.5/125 µm core-clad step-index multimode fiber covered by a stainless steel tube. A wavelength filtering module selectively transmitted the Raman Anti-Stokes signal, which was converted to an electrical signal by a photomultiplier tube, amplified with a gain of 25 and digitized. The initial 10 m of the fiber served as the reference zone, with its temperature measured by a Pt100 RTD sensor interfaced with an ATmega328P micro-controller. Controlled hot zones were created with a laboratory-developed thermal bath capable of temperatures from room temperature up to 70 °C with an accuracy of 0.1 °C, using a PID controller, a 2 kW heater and a Pt100 RTD in closed-loop control, with a stirrer maintaining uniformity. Three hot zones of 10 m, 10 m and 15 m were established at locations 115 m, 280 m and 418 m, lengths chosen to span multiple spatial resolution cells and minimize edge effects.
During calibration, the entire fiber was held at an ambient room temperature of 22 °C, and the data acquisition unit used an averaging factor of 2¹⁸, or 262,144 traces, corresponding to 300 s per acquisition. In the measurement stage, Anti-Stokes traces were collected for hot zone temperatures of 30, 40, 50, 60 and 70 °C, and attenuation compensation was applied by dividing the captured traces by the stored attenuation profile. Comparison between estimated and actual hot zone temperatures demonstrated accuracy within ±0.8 °C. To quantify noise, the entire fiber was held at constant room temperature and 430 consecutive traces were recorded; the temperature resolution, defined as the point-wise standard deviation of the estimated temperature, was approximately 0.35 °C at the far end. The resolution increased from 0.13 °C at the fiber input to 0.33 °C at 430 m, primarily due to degradation in signal-to-noise ratio with distance. An exponential fit yielded a coefficient of 2.16 km⁻¹, and extrapolating the fitted model beyond the investigated distance predicts an achievable sensing distance of approximately 1.26 km for applications tolerating a maximum resolution of 2 °C with the present hardware. Spatial resolution, measured as the distance between the 10% and 90% rise points of the temperature response, was 2.8 m at 108 m, 3.1 m at 274 m and 3.4 m at 410 m, degrading gradually with distance due to intrinsic fiber dispersion.
To test splice correction, an additional 85 m fiber was spliced to the sensing fiber, giving a total length of 520 m, and a 10 m hot zone was established at 510 m, beyond the splice point. The splice-induced loss at 435 m was compensated within a maximum error of ±1.3 °C beyond the splice location, with no abrupt change or discontinuity in the corrected traces. In a 20-day stability study, Anti-Stokes signals were recorded continuously with three measurements acquired daily, under an unchanged fiber layout and ambient room temperature that varied between 21.2 °C and 24.1 °C, using the same attenuation profile captured at the start. A maximum temperature error of 3.4 °C was observed, attributed primarily to slight variations in the effective attenuation profile over time. The authors note that the reported accuracy and resolution reflect both the compensation technique and the experimental conditions, particularly the 1064 nm laser and the high averaging factor, which enhance the signal-to-noise ratio.
The method’s practical advantages include simple calibration without auxiliary correction fibers or temperature-controlled thermal baths, a single photodetector and amplifier, and no recalibration when detectors or amplifiers are replaced, provided the fiber layout and environmental conditions remain unchanged. Unlike double-ended techniques, it does not require a looped sensing fiber, preserving the entire physical fiber length for sensing. The approach does assume that the attenuation profile remains unchanged between calibration and measurement; prolonged exposure to radiation, elevated temperatures, mechanical stress or changes in fiber layout can alter attenuation characteristics, and new splices or connectors change the profile, so recalibration is required in such cases and constitutes a limitation. Experiments with two spool diameters of 24 cm and 32 cm, both exceeding the 40 mm minimum bend radius of the stainless steel tube-encapsulated fiber, showed that the attenuation profile depends on the mechanical layout, arising from macrobending, microbending and stress-induced effects. The authors state that the algorithm is independent of sensing distance and applicable to longer fibers provided sufficient signal-to-noise ratio is maintained, potentially extended by increasing launched pulse energy within nonlinear limits, using a lower-noise receiver, increasing averaging or operating at 1550 nm, where lower fiber attenuation reduces signal degradation with distance.
Subject of Research: Attenuation-corrected temperature demodulation in single-ended, single-channel Raman distributed temperature sensing
Article Title: Studies on attenuation corrected temperature demodulation in single ended Raman distributed temperature sensors
Article References: Sahoo, A. K., Kumar, J., & Prakash, O. (2026). Studies on attenuation corrected temperature demodulation in single ended Raman distributed temperature sensors. Discover Sensors, 2(1), Article 60. https://doi.org/10.1007/s44397-026-00093-7
Image Credits: AI Generated
DOI: 10.1007/s44397-026-00093-7
Keywords: Raman distributed temperature sensing, distributed fiber-optic sensors, Anti-Stokes signal, attenuation compensation, wavelength dependent loss, walk-off effect, temperature resolution, spatial resolution, optical fiber sensing, calibration, Studies, attenuation
News Source: Denise Maddox. (November 6, 2026). Single-channel fiber sensor reads temperature along 450 m with ±0.8 °C accuracy. Scienmag.
