PyOFDR

OFDR simulator in Python. Generates synthetic interrogator traces, from the swept laser to the digitized beat signal, so you can develop algorithms, train ML models, or size experiments without an instrument on the bench.

1000 microstrain applied to a 1 m segment of a 10 m fiber, recovered from a single PyOFDR run by sub-spectrum cross-correlation against an unstrained baseline.

Why

Optical Frequency Domain Reflectometry is an extremely powerful tool for distributed fiber sensing, but commercial interrogators are often black boxes. PyOFDR allowes you to:

• validate demodulation / strain-extraction algorithms against ground truth
• generate labelled OFDR training data at any chosen SNR
• size sweep range, sample rate, fiber length, expected noise floor before buying hardware

Signal chain

Swept laser  -->  Main MZ + aux MZI  -->  Fiber (Rayleigh + features)  -->  Photodetector  -->  ADC

What's simulated

Stage	Modelled
Source	linear sweep + nonlinearity + ripple, phase noise (Wiener + flicker + RW), RIN, coherence roll-off
Interferometer	main Mach-Zehnder, auxiliary MZI for k-clock resampling, optical circulator, weak FBG arrays
Fiber	Rayleigh backscatter, attenuation, macrobend loss, discrete reflectors, chromatic dispersion (GVD), stochastic n(z), multiple-scattering ghosts, MCF core-to-core crosstalk
Perturbations	static + dynamic strain (harmonic, impulsive, thermal relaxation, random vibration), Cox shear-lag coating coupling, distributed temperature with strain-T cross-sensitivity
Detection	balanced photodetector, shot / thermal / dark-current noise, polynomial PD nonlinearity, anti-alias + Butterworth LPF
ADC	clip, quantize, jitter, DNL/INL, ENOB
Output	HDF5 streaming, multi-sweep campaigns

Theory and references are in docs/theory.md

Install

git clone https://github.com/ameoli/PyOFDR.git
cd PyOFDR
pip install -e .              # runtime
pip install -e .[dev]         # + pytest, matplotlib, jupyter

Quick start

from pyofdr.core.config import load_config
from pyofdr.core.campaign import run_campaign

cfg  = load_config("configs/ofdr_basic.yaml")
acqs = run_campaign(cfg)        # one Acquisition per sweep
acq  = acqs[-1]
# acq.digital_main : int16 array, shape (n_cores, n_samples)

The pyofdr CLI is installed as well:

pyofdr info     configs/ofdr_basic.yaml
pyofdr validate configs/ofdr_basic.yaml
pyofdr run -o out.h5 configs/ofdr_basic.yaml

Tests

pytest tests/ -v

Examples & notebooks

• examples/ -- minimal runnable scripts
• notebooks/ -- Jupyter tutorials

Status

TBD

Citing

If you use PyOFDR in published work, please cite it.

@software{pyofdr,
  author = {Meoli, Alessandro},
  title  = {PyOFDR: an end-to-end Python simulator for Optical Frequency Domain Reflectometry},
  year   = {2026},
  url    = {https://github.com/ameoli/PyOFDR},
}

References

• Froggatt and Moore, High-spatial-resolution distributed strain measurement in optical fiber with Rayleigh scatter, Appl Opt 1998.
• Soller et al., High resolution OFDR for characterization of components and assemblies, Opt Express 2005.
• Hartog, An Introduction to Distributed Optical Fibre Sensors, CRC 2017.

License

MIT -- see LICENSE.

Third-party

felixpatzelt/colorednoise (MIT), vendored at src/pyofdr/utils/colorednoise.py for laser phase-noise PSD shaping. Full text in LICENSES/colorednoise-MIT.txt.

Issues and contributions

Bugs and feature requests: https://github.com/ameoli/PyOFDR/issues Pull requests welcome.