⚡ electrical engineering

Accurate Multi-perturbation Localization in Optical Fibers with Polarization-based Forward Sensing

This paper presents a polarization-based forward sensing method that accurately localizes multiple simultaneous perturbations in optical fibers by cross-correlating received polarizations at two wavelengths and interpolating peaks, achieving a median error of 3.5 meters for five perturbations in a 10 GSa/s numerical model.

Original authors: Lampros Lanaras, Rick M. Butler, Christian Häger, Alex Alvarado

Published 2026-08-05

📖 3 min read☕ Coffee break read

CC BY 4.0

Original authors: Lampros Lanaras, Rick M. Butler, Christian Häger, Alex Alvarado

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a giant, invisible spiderweb stretching for miles, made not of silk but of glass threads called optical fibers. These threads are the superhighways of the internet, carrying our videos, messages, and cat videos at the speed of light. But sometimes, something bumps into the web. Maybe a construction crew digs nearby, a tree branch falls, or a curious animal scratches the cable. When this happens, the light traveling inside the fiber gets "jostled," changing its direction or spin in a way that tells us something happened.

Scientists have long wanted to know exactly where along the fiber this bump occurred. The old way to do this is like shouting down a tunnel and listening for an echo; it works, but it requires massive, expensive equipment that acts like a giant, sensitive ear at the start of the line. A newer, cheaper idea is to look at the light coming out the other end. Think of the light as a spinning top. If you poke the fiber, the top's spin gets scrambled. By watching how the spin changes, you can tell a disturbance happened. But here's the tricky part: if you poke the fiber in two places at once, or if the fiber is very long, it's like trying to figure out which of two spinning tops was hit and exactly where, just by looking at the final wobble. It's a messy puzzle where the clues get blurry the further they travel.

This paper tackles that messy puzzle. The authors, working with computer models of fiber optics, developed a clever new trick to find multiple bumps at once with incredible precision. Instead of just looking at the light at one color (wavelength), they send two different colors of light down the fiber at the same time. Because different colors travel at slightly different speeds through glass, the "scramble" caused by a bump arrives at the receiver at slightly different times for each color. It's like two runners starting a race together but finishing a split second apart; that tiny gap tells you exactly how far they ran.

The team's big innovation is how they handle the data. Previous methods were like trying to guess the finish line by only looking at the track at specific, spaced-out markers (like every 10 meters). If the runner finished between markers, you had to guess, leading to errors. This new method uses a mathematical "zoom lens" called interpolation. It looks at the blurry spots between the markers and fits a smooth curve to guess the exact finish time, even if it doesn't land perfectly on a number. By combining this with a detector that can spot multiple "finish lines" at once, they simulated a 50-kilometer fiber and successfully pinpointed five different bumps happening simultaneously. In their computer simulations, they found the location of these bumps with a median error of just 3.5 meters. While this is a simulation and not a real-world test yet, it suggests a way to turn standard internet receivers into super-accurate sensors that can map out exactly where the fiber is being touched, all without needing expensive, specialized hardware.

Technical Summary: Accurate Multi-perturbation Localization in Optical Fibers with Polarization-based Forward Sensing

Problem Statement
Distributed Acoustic Sensing (DAS) effectively localizes fiber perturbations using backscatter phase but relies on expensive equipment. An alternative, forward polarization sensing, monitors the State of Polarization (SOP) of propagated waves at the receiver. While external perturbations alter the fiber cross-section geometry and scramble the SOP, standard coherent receivers lack the inherent localization capability of DAS. Previous distributed forward sensing schemes [3–5] utilized chromatic dispersion (CD) to localize perturbations by observing SOP variations at distinct wavelengths with a CD-induced delay. However, these existing methods face two primary limitations: they typically sense only one perturbation at a time, and their localization resolution is strictly limited by the receiver's sampling rate. For instance, prior work reported mean errors ranging from sub-25 meters to kilometers, heavily dependent on sampling frequencies (e.g., 32 GSa/s vs. 100 MSa/s).

Methodology
The authors propose a post-processing method to extend the single-perturbation approach of [3, 4] to handle multiple simultaneous perturbations with high resolution, without requiring increased sampling rates. The core methodology involves:

Propagation Modeling: The fiber is modeled using the coarse-step method, dividing a length- $L$ fiber into sections where random unitary matrices scramble the SOP. Perturbations are modeled as sudden changes in the scrambling matrix at specific sections. The model accounts for CD, attenuation, and amplifier noise but excludes Kerr nonlinearity and polarization-dependent loss.
SOP Angular Speed (SOPAS) Extraction: The system transmits two carrier waves at wavelengths $\lambda_1$ and $\lambda_2$ . Upon perturbation, the SOP changes propagate to the receiver at different group velocities ( $v_{g1} < v_{g2}$ ), creating a time delay $t_1 - t_2$ proportional to the distance from the perturbation to the receiver. The receiver calculates the SOP angular speed (SOPAS), $a_m[l]$ , for each wavelength to detect these fluctuations.
Cross-Correlation and Multi-Peak Detection: The SOPAS signals from the two wavelengths are cross-correlated ( $R[l] = a_1[l] \star a_2[l]$ ). While a single perturbation yields one peak, multiple perturbations generate multiple peaks. The authors employ a multi-peak detector that identifies all local maxima in $R[l]$ exceeding a threshold based on the median absolute deviation of the signal.
Interpolation for Sub-Sample Accuracy: To overcome the resolution limit imposed by discrete sampling ( $\Delta t_{wave}$ ), the authors apply a parabolic interpolation to the three samples surrounding each detected peak. By fitting a parabola and calculating its analytical maximum, the system estimates the delay $t_1 - t_2$ at non-integer multiples of the sampling period, significantly reducing quantization error.

Key Contributions

Multi-Perturbation Capability: This is the first forward sensing method to localize an arbitrary number of simultaneous perturbations using CD and SOP.
Resolution Enhancement via Interpolation: The paper introduces a parabolic interpolation technique applied to the cross-correlation peaks. This allows for localization delays at non-integer sampling intervals, decoupling localization accuracy from the sampling rate.
Robust Detection Algorithm: A thresholding scheme based on the median absolute deviation is used to filter false positives while accommodating the increased noise floor caused by multiple simultaneous perturbations.

Results
The method was validated using numerical simulations of a 50 km single-span fiber with two C-band carriers sampled at 10 GSa/s.

Single Perturbation: For a single perturbation, the proposed interpolation method reduced the median localization error by a factor of 8.1 compared to the non-interpolated method from [3, 4]. At a wavelength separation ( $\lambda_2 - \lambda_1$ ) of 35 nm, the median error was 2.33 m. The error performance remained consistent with theoretical expectations derived from quantization limits but showed a slight increase in sensitivity to amplifier noise compared to the discrete method.
Multiple Perturbations: The system successfully localized up to five simultaneous perturbations. With five perturbations, the median localization error increased to 3.5 m, with an interquartile range (IQR) of 7.5 m. The true positive (TP) rate remained above 90% for up to three perturbations; however, as the number of perturbations increased, noise in the cross-correlation signal grew, leading to a higher rate of false negatives due to the adaptive thresholding mechanism.

Significance and Claims
The authors claim that their work demonstrates the feasibility of accurately localizing multiple simultaneous fiber perturbations using standard coherent receivers and C-band wavelengths, achieving a median error of 3.5 m without requiring high sampling rates. The paper positions this as a significant step forward in forward sensing, moving beyond single-event detection to spatial profiling of multiple disturbances. The authors suggest the method could be implemented in coherent receivers to pinpoint spatial perturbation profiles. They modestly note that future work will involve more detailed fiber models and experimental validation, potentially considering different types of perturbations beyond the instantaneous random scramblers modeled here.

More like this