Modulated vs. Unmodulated Polarization Signatures for ML-Based Fiber Sensing
Bibliographic record
Abstract
Fast and accurate detection of various physical layer threats that target optical networks is key to secure and reliable global communications. Conventional monitoring methods often fail to detect subtle anomalies, which requires advanced sensing techniques. Machine Learning (ML) analysis of the State of Polarization (SOP) of unmodulated signals was shown to successfully detect such disturbances. However, real-world networks typically operate with high-speed modulated signals, which may alter SOP behavior and challenge the applicability of ML techniques developed for unmodulated signals. This paper investigates the implications of signal modulation supporting high-data rates on the interpretability of SOP signatures. We perform the first experimental comparison of anomaly detection approaches based on SOP for modulated and unmodulated optical signals subjected to identical physical perturbations caused by fiber tapping and vibrations. We analyze four representative events under both signal modalities and assess the impact of modulation on SOP dynamics using a 77 km fiber link in a real-world metro network. We design four datasets that isolate, merge, and jointly classify the different signal modalities, and compare the performance of ten best-performing supervised ML techniques in each case. Our findings indicate that modulated signals tend to exhibit smoother SOP trajectories, likely due to the temporal averaging effects introduced by high symbol rates, wherein rapid symbol transitions suppress highfrequency polarization noise. Importantly, this smoothening does not obscure the slower, event-induced polarization drifts observed during physical disturbances, allowing ML models to reliably differentiate between different physical events (e.g., bending, vibrations) and signal modalities (modulated vs. unmodulated), achieving accuracy values between 97.12% and 98.47%.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".