REVIEW 3 major objections 5 minor 13 references
Long-Term Latency Measurement of Deployed Fiber
T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read This paper reports that buried optical fiber latency drifts by about 17 ns over a year, driven by slow soil temperature changes, and that the drift can be tracked with picosecond accuracy using a correlation OTDR.
desk verdict A solid two-week field measurement of deployed fiber latency, with an annual extrapolation that is a long way from the data. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing measurement technique is a correlation OTDR: a 10-Gbit/s 127-bit PRBS burst is launched into each fiber, the reflected traces are averaged and cross-correlated with the transmitted sequence, and a raised-cosine function is fitted to the correlation peaks to push position accuracy from the 20 ps oscilloscope sample spacing to about 2 ps. The load-bearing analysis tool is a first-order low-pass thermal model in which fiber temperature is the outside-air temperature convolved with a single exponential decay. Fitting its time constant (12.7 days) and the temperature delay coefficient (7.5 ppm/K) to the measured latency lets the model extrapolate a two-week summer record to seasonal and annual swings.
What would settle it
Measure soil or cable temperature at the burial depth in Meiningen through a full year, or run the C-OTDR for a full year, and compare with the 12.7-day filtered air temperature: the model predicts about a 28 K annual fiber-temperature swing and a ~17 ns round-trip latency swing, so a measurement showing a clearly different seasonal lag, a different amplitude, or a latency swing far from 17 ns would falsify the projection.
Extended reading notes
Core claim
The central discovery is that the latency of fibers in a deployed underground cable can be followed continuously at picosecond precision, that it changes by hundreds of picoseconds over days, and that this change behaves like a heavily damped, delayed copy of air temperature. Over the two-week record the maximum round-trip latency increase was 800 ps, the daily swing was about 200 ps, and skew between fibers in the same cable varied by up to 12 ps. Fitting the latency trace with a temperature delay coefficient of 7.5 ppm/K and a single-pole time constant of 12.7 days reproduces the slow multi-day trends and part of the day-night ripple. The same model applied to a year of air-temperature data yields an annual fiber temperature range of about 28 K, implying annual round-trip latency and skew variations of about 17 ns and 200 ps.
Load-bearing premise
The annual projection rests on treating the buried cable as a single-exponential thermal filter fitted to two summer weeks of air temperature and latency, with no direct ground-temperature check; if the soil's true seasonal response differs, the 28 K swing and the 17 ns figure change.
Editorial extensions
If this is right
- Network synchronization equipment should treat the round-trip latency of buried fiber as a slow variable: the model says an 8.5 km link drifts by roughly 17 ns over a year and about 200 ps per day.
- Fibers in one cable do not drift identically: skew variations up to 12 ps over two weeks and about 200 ps annually set a floor for differential-delay compensation in phase-array fronthaul.
- With a temperature delay coefficient near 7.5 ppm/K, a 1.5 K summer swing in cable temperature explains the observed 800 ps round-trip change; longer records should show similar proportionality.
- Because the filter time constant is much longer than a day, daily air-temperature cycles are strongly attenuated at cable depth, so short-term latency jitter is small even when surface temperature swings by 10 K.
- A monitored reference fiber could provide a real-time estimate of the common-mode temperature drift of all fibers in a cable, leaving only the smaller differential skew to be compensated.
Reading between the lines
- Beyond the paper: a full-year C-OTDR record on the same link would test whether the 12.7-day exponential lag is adequate; soil heat diffusion is not a single pole, so the 28 K annual estimate may be off if seasonal penetration differs.
- Editorial inference: if one fiber in a cable is monitored continuously, its latency could serve as a reference to compensate skew on the other fibers, provided the roughly 1% TDC differences observed here remain stable.
- Beyond the paper: the same correlation-OTDR technique could be applied to aerial or duct fiber, where the shorter thermal time constant would make daily latency swings much larger and more visible in timing protocols.
- The fitted 7.5 ppm/K temperature delay coefficient sits between the bare-fiber value near 6 ppm/K and the tight-buffered jumper value of 17 ppm/K, suggesting cable construction and strain contribute measurably to field latency drift.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper reports a 14-day field measurement of round-trip latency in four fibers inside a deployed 8.5-km underground cable, using a correlation OTDR with approximately 2-ps accuracy. The authors observe a maximum round-trip latency variation of 800 ps and skew variations up to 12 ps. They fit the measured latency to a first-order low-pass-filtered version of outside air temperature, obtaining a temperature delay coefficient of 7.5 ppm/K and a time constant of 12.7 days. Using this model with a year of outside air temperature data, they estimate the annual fiber temperature swing as about 28 K, which they translate into annual latency and skew variations of approximately 17 ns and 200 ps. The abstract states an annual temperature variation of 25 K.
Significance. The direct measurement portion is a valuable contribution: simultaneous four-fiber latency monitoring, 50 GS/s sampling, 2000 trace averaging, raised-cosine peak fitting, and a reference reflection provide a credible, high-accuracy characterization of buried fiber latency over two weeks. If the annual extrapolation were properly supported, this would be the first high-accuracy long-term latency characterization of deployed buried fiber, with direct relevance to 5G synchronization and differential-latency applications. The paper's strength is in the experimental methodology and the new quantitative data on the 12-ps scale of inter-fiber skew variation over two weeks. The weakness is the annual projection, which currently rests on a two-parameter model fitted to a record only slightly longer than the fitted time constant.
major comments (3)
- [Section 4] The annual extrapolation is not identified by the data. The single-pole low-pass filter parameters (7.5 ppm/K and 12.7 days) are fitted to the same 14-day July latency record that is then used to demonstrate agreement in Fig. 5a. Since 14 days is only about 1.1 times the fitted time constant, the record provides almost no constraint on the filter's low-frequency response, which determines the annual amplitude and phase. No out-of-sample data, ground-temperature measurements at cable depth, or soil heat-diffusion model are provided. The estimated 28-K annual swing and the resulting 17 ns and 200 ps projections therefore are not supported by the measurements. Please either add an independent validation (e.g., a different season or a second measurement campaign) or explicitly reframe the annual values as a model-based scenario with a sensitivity analysis.
- [Abstract vs. Section 4] The abstract states annual temperature variations of 25 K, while Section 4 estimates a peak-to-peak variation of approximately 28 K and Section 5 repeats 28 K. Since the annual temperature swing is the basis of the headline claims, this numerical inconsistency should be resolved; it also suggests that the fitting and projection procedure is sensitive to small choices in the data handling, which reinforces the need for an uncertainty quantification.
- [Section 4, skew estimate] The annual skew variation of approximately 200 ps is stated without derivation. It appears to be a linear scaling of the 12-ps measured skew variation by the ratio of the annual temperature swing to the two-week temperature swing (about 28 K / 1.5 K). This implicitly assumes that the skew variation scales linearly with temperature amplitude and that the low-pass filter affects all fibers identically. The paper should present the exact calculation and justify the linear scaling, especially because the 12-ps skew variation was observed over a period dominated by the 12.7-day time constant, not by annual fluctuations.
minor comments (5)
- [Section 2] The text says 'The cosine fit is show in Fig. 2c'; 'show' should be 'shown'.
- [Section 3] The paper states the measurement was interrupted 'around the 8th day' while the Fig. 3a caption mentions 'a half day interruption from 9th to 10th.' Please make the description consistent.
- [Section 3/4, weather data] The weather station providing the outside air temperature is not identified by location or distance from the cable route. Local microclimate, soil moisture, snow cover, and solar radiation can significantly affect soil temperature at the burial depth; please state the station details and discuss the potential systematic error.
- [Section 4] The fitted parameters (TDC and time constant) are reported without confidence intervals or a goodness-of-fit metric beyond visual inspection of Fig. 5a. A quantitative measure, such as residual standard deviation or Akaike information criterion, would allow the reader to judge the quality of the single-pole approximation.
- [References] Reference [6] is cited as evidence that ground temperature behaves similarly, but no quantitative comparison to the fitted 12.7-day time constant or to the measured 1.5-K temperature swing is provided. Including the relevant depth and soil type from that reference would strengthen the comparison.
Circularity Check
No significant circularity: the measured 14-day latency data are new observations, and the annual estimate is an explicit extrapolation using external weather-station temperature data, not a prediction equivalent to the fit inputs.
full rationale
The paper's central measurements are direct: a C-OTDR monitored round-trip latency of four deployed fibers over 14 days, yielding 800 ps maximum drift and 12 ps skew variation. These are observations, not outputs of a model. The thermal model in Section 4 is openly described as a fit: 'The low-pass filter parameters were derived by a fit of the filtered temperature to the measured latency variations of the fiber.' Thus the agreement shown in Fig. 5a is an in-sample fit, not a claimed independent prediction. The annual projection is obtained by applying this fitted first-order low-pass filter and the fitted 7.5 ppm/K TDC to external 10-minute air-temperature records from the German Meteorological Service: 'Using this model for the annual variations of the fiber temperature based on data of the air temperature obtained from the German Meteorological Service, we estimated the variation of the fiber temperature over the year.' The annual temperature record is independent of the 14-day latency fit, so the 28 K swing, 17 ns latency variation, and 200 ps skew projection are extrapolations of a fitted model rather than quantities that reduce by construction to the fitting inputs. The self-citations (Refs. [4] and [5]) support the linearity assumption and the C-OTDR accuracy, but the numerical TDC and time constant used in this paper are fitted here, so the self-citations are not load-bearing in a circular sense. The main weakness is statistical: a two-parameter fit to a 14-day summer record is a thin basis for an annual extrapolation, and no ground-temperature validation is provided. That is a robustness and validation concern, not circularity.
Assumptions & free parameters
free parameters (2)
- Temperature delay coefficient (TDC) =
7.5 ppm/K
- Low-pass time constant =
12.7 days
assumptions (3)
- domain assumption Fiber latency is linear in fiber temperature with a constant TDC
- domain assumption Buried fiber temperature follows a first-order low-pass filtered version of outside air temperature
- domain assumption Air temperature from the German Meteorological Service [7] represents the cable environment over the whole year
Cite this review
Pith. "Pith review of Long-Term Latency Measurement of Deployed Fiber." pith.science (2026). https://pith.science/paper/RPU4AJUG
@misc{pith2026190810670,
author = {Pith},
title = {Pith review of: Long-Term Latency Measurement of Deployed Fiber},
year = {2026},
howpublished = {\url{https://pith.science/paper/RPU4AJUG}},
note = {Machine review of arXiv:1908.10670}
}
read the original abstract
Using a Correlation-OTDR we measured the latency of fibers in a deployed cable and calculated the time coefficient of the fiber temperature changes. Annual temperature variations of 25K were estimated for the deployed fiber.
Reference graph
Works this paper leans on
-
[1]
Synchronization protocols require a stable and symmetric latency between master and slave clocks
Introduction Latency is becoming a critical parameter in future 5G networks. Synchronization protocols require a stable and symmetric latency between master and slave clocks. Other applications, like the transmission of radio phase array signals, as investigated in the European BlueSpace project [1], require a very low differential latency between differe...
-
[2]
Experimental setup We characterized four fibers in a deployed 8.5-km underground cable between the ADVA office and a Deutsche Telekom central office in Meiningen, in Central Germany. The setup is shown in Fig. 1. A laser at a wavelength of 1550 nm was modulated using a Mach-Zehnder modulator with a 10-Gbit/s 127-bit PRBS burst followed by zeros to fill a ...
work page 2000
-
[3]
As shown in the blue curve in Fig
Results Using the C-OTDR, the round-trip latency of the four fibers was monitored over approximately two weeks with a half day interrupted around the 8th day. As shown in the blue curve in Fig. 3a, the maximum latency increase over the time, referenced to the latency at the start of the measurement, was 800 ps. Fig. 3a also contains the outside air temper...
work page 2018
-
[4]
Modelling of the fiber temperature We modelled the evolution of the temperature of the fiber deployed in the ground as a 1 st order low-pass filtered function of the outside air temperature. The low-pass filter parameters were derived by a fit of the filtered temperature to the measured latency variations of the fiber. This assumes that the latency variat...
-
[5]
Summary Using a correlation OTDR, with an accuracy of approximately 2 ps, we measured the latency variations of four fibers in a deployed 8.5-km cable. Over a period of 14 days in the summer, a maximum round-trip latency variation of 800 ps was measured. The measured skew variations between the fibers were up to 12 ps. Based on these measurements, a time ...
-
[6]
Acknowledgment This project has received funding from the European Union´s Horizon 2020 research and innovation programme under grant agreement No 762055 (BlueSpace Project)
work page 2020
-
[7]
https://bluespace-5gppp.squarespace.com/
-
[8]
Phase Stability of ATA Fiber Optic Cables
J. W. Dreher, “Phase Stability of ATA Fiber Optic Cables”, SETI Institute, November 2000
work page 2000
Show all 13 references
-
[9]
Thermal coefficient of delay for various coaxial and fiber-optic cables,
G. Lutes and W. Diener, “Thermal coefficient of delay for various coaxial and fiber-optic cables,” TDA Progress Report 42-99, Nov. 1989, available online at https://ipnpr.jpl.nasa.gov/progress_report/42-99/99E.PDF
1989
-
[10]
Azendorf, A
F. Azendorf, A. Dochhan, and M. Eiselt, “Temperature Dependent Latency of Jumper Cable “, ITG Photonic Networks 2018, Leipzig, June 2018
2018
-
[11]
Eiselt and A
M. Eiselt and A. Dochhan, „Single-Ended Fiber Latency Measurement with Picosecond-Accuracy Using Correlation OTDR,” paper 5C1-3, Proc. 23st OptoElectronics and Communications Conference (OECC), Jeju, Korea, July 2018
2018
-
[12]
Annual Ground Temperature Measurement at Various Depths
G. Florides and S. Kalogirou, “Annual Ground Temperature Measurement at Various Depths”, 8th REHVA World Congr. 2005, available on https://www.researchgate.net/publication/30500353
2005
-
[13]
ftp://ftp-cdc.dwd.de/pub/CDC/observations_germany/climate/10_minutes/air_temperature/recent/ Fig. 5: Modelling of cabled fiber temperature a) Measured fiber latency(blue) and latency based on filtered cable temperature (red); b) Temperature evolution over 18 months: 10-min air...
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.