pith. sign in

arxiv: 2605.03418 · v1 · submitted 2026-05-05 · 📡 eess.SP

Identification of Clock Ensemble Noise Parameters Using Differential Measurement Analysis

Jindrich Dunik , Ladislav Kral , Ivo Puncochar , Oliver Kost , Ondrej Daniel , Simona Circiu , Bernardino Quaranta This is my paper

Pith reviewed 2026-05-07 14:59 UTC · model grok-4.3

classification 📡 eess.SP
keywords atomic clock ensembleparameter identificationdifferential measurementsstate-space modelsnoise parametershydrogen maserstochastic modeling
0
0 comments X

The pith

Atomic clock ensemble noise parameters can be identified from pairwise phase differences alone.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops procedures to recover the unknown parameters of each clock in an ensemble when only differential phase measurements relative to one designated pivot clock are provided. A reader would care because many practical systems supply only relative data yet still require accurate individual clock models for stable timekeeping and prediction. Each clock is represented by a second-order linear stochastic state-space model driven by white frequency noise, random walk frequency noise, a deterministic drift, and measurement noise. Two identification methods are constructed to estimate the corresponding variances, drift value, and noise covariance from these differences. The methods are shown to work on both simulated scenarios and real recordings from atomic hydrogen maser clocks.

Core claim

The central claim is that the unknown parameters of each clock's second-order stochastic model, namely the variances of white frequency noise and random walk frequency noise, the drift rate, and the measurement noise covariance, can be estimated from differential phase measurements between a pivot clock and the remaining clocks using either of two designed identification procedures, with accuracy confirmed on simulated and real hydrogen maser data.

What carries the argument

The differential measurement analysis that reconstructs each clock's individual noise and drift parameters from observed pairwise phase differences, exploiting the structure of the coupled second-order linear stochastic state-space models.

If this is right

  • Individual clock behaviors can be predicted accurately without access to absolute time references.
  • Ensemble time scale generation and steering become feasible using only relative observations.
  • Noise sources for each clock are separated even though only difference signals are recorded.
  • The procedures apply directly to existing atomic clock hardware such as hydrogen maser ensembles.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The technique may lower the instrumentation cost for monitoring distributed clock networks by removing the need for precise absolute references.
  • Analogous relative-measurement approaches could be tested on other sensor arrays that track correlated states, such as vehicle formations or distributed oscillators.
  • Direct comparison of identified parameters against independent full-data calibrations on the same clocks would provide a clear empirical check of uniqueness.

Load-bearing premise

That each clock follows a second-order linear stochastic state-space model with white frequency noise, random walk frequency noise, drift, and measurement noise, and that these differential measurements alone contain sufficient information to identify all parameters uniquely.

What would settle it

A mismatch between parameters recovered from differential measurements and those obtained when absolute phase data for all clocks in the same ensemble become available.

Figures

Figures reproduced from arXiv: 2605.03418 by Bernardino Quaranta, Ivo Puncochar, Jindrich Dunik, Ladislav Kral, Oliver Kost, Ondrej Daniel, Simona Circiu.

Figure 1
Figure 1. Figure 1: Clocks AVAR for true and estimated parameters (ACOV method - simulation). view at source ↗
Figure 2
Figure 2. Figure 2: Clocks AVAR for true and estimated parameters (MDM method - simulation). view at source ↗
Figure 3
Figure 3. Figure 3: AVAR of the iMaser3000 atomic clock adopted from a datasheet and view at source ↗
Figure 4
Figure 4. Figure 4: AVAR of the iMaser3000 atomic clock adopted from a datasheet and view at source ↗
read the original abstract

The paper addresses the critical problem of identifying unknown parameters of an atomic clock ensemble. The ensemble model is considered as a set of individual clock models, where each clock is described by a second-order linear stochastic state-space model. The paper presents identification procedure for model unknown parameters based solely on the availability of differential measurements - that is, the measured pairwise phase differences between a designated pivot clock and all other clocks within the ensemble. Specifically, each clock model is defined by the following set of unknown parameters: the variances characterizing the white frequency noise and random walk frequency noise, the drift, and the (co)variance of the measurement noise. Two distinct identification methods are designed to estimate the unknown clock model parameters. The accuracy of the identified sets of parameters are demonstrated on a simulation scenario/real data combining atomic H-maser (AHM) clocks.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper claims to provide two identification procedures for estimating the unknown parameters (white frequency noise variance, random walk frequency noise variance, drift, and measurement noise covariance) of each clock in an atomic clock ensemble. Each clock is modeled as a second-order linear stochastic state-space model, and the procedures rely exclusively on differential phase measurements between a designated pivot clock and the other clocks. Accuracy is demonstrated through simulation scenarios and real data from atomic hydrogen maser (AHM) clocks.

Significance. If the methods correctly and uniquely recover all claimed parameters from differentials alone, the work would enable practical noise characterization in clock ensembles without requiring absolute phase references, which has direct value for metrology, GNSS, and time-scale generation. The data-driven framing and use of both simulated and real AHM data are positive features.

major comments (2)
  1. [Abstract / model section] Abstract and model definition: the claim that all four parameters per clock (including individual drifts) are uniquely identifiable from pairwise phase differences to a single pivot is not supported. Drift enters as a deterministic linear term; adding any constant c to every clock's drift (including the pivot) leaves all differential trajectories and their statistics invariant, so the likelihood is flat along this direction and absolute drifts cannot be recovered—only relative drifts among the n-1 non-pivot clocks.
  2. [Identification procedures] Identification methods (both procedures): the uniqueness assumption stated for the full parameter vector is violated by the common-drift degeneracy. While noise variances may be separable via cross-covariances of the differential channels, the drift component requires either an absolute reference measurement or an explicit reparameterization to relative drifts; neither appears to be addressed.
minor comments (2)
  1. [Abstract / results] Clarify whether the reported 'drift' estimates are absolute or relative to the pivot; if relative, update the abstract and parameter list accordingly.
  2. [Results / validation] Provide explicit validation metrics (e.g., parameter error statistics, confidence intervals, or cross-validation scores) rather than qualitative statements about accuracy on simulation and real data.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the insightful comments regarding parameter identifiability. We agree that the original manuscript did not sufficiently address the common-drift degeneracy and will revise to clarify that only relative drifts (with respect to the pivot) are identifiable. Below we respond point by point.

read point-by-point responses

  1. Referee: [Abstract / model section] Abstract and model definition: the claim that all four parameters per clock (including individual drifts) are uniquely identifiable from pairwise phase differences to a single pivot is not supported. Drift enters as a deterministic linear term; adding any constant c to every clock's drift (including the pivot) leaves all differential trajectories and their statistics invariant, so the likelihood is flat along this direction and absolute drifts cannot be recovered—only relative drifts among the n-1 non-pivot clocks.

    Authors: We agree that absolute drifts cannot be recovered from differential measurements alone due to the invariance under a common additive constant. Our procedures in fact recover relative drifts (differences with respect to the pivot clock). We will revise the abstract and model section to remove the claim of unique identifiability of absolute drifts and explicitly state that drifts are estimated relative to the pivot, whose own drift is set to zero without loss of generality for the differential model. revision: yes

  2. Referee: [Identification procedures] Identification methods (both procedures): the uniqueness assumption stated for the full parameter vector is violated by the common-drift degeneracy. While noise variances may be separable via cross-covariances of the differential channels, the drift component requires either an absolute reference measurement or an explicit reparameterization to relative drifts; neither appears to be addressed.

    Authors: We concur that the uniqueness claim for the full parameter vector (including absolute drifts) is not valid. The noise variances and measurement covariances remain uniquely identifiable from the differential statistics, but the drift terms require reparameterization. We will add an explicit reparameterization of the drift vector as relative drifts in the identification procedures section and update the uniqueness statements accordingly. This revision will be incorporated in the next version of the manuscript. revision: yes

Circularity Check

0 steps flagged

No circularity; data-driven parameter estimation from measurements

full rationale

The paper frames its contribution as two identification methods that estimate clock model parameters (noise variances, drift, measurement noise) directly from observed differential phase measurements between a pivot clock and others. These procedures are presented as standard state-space estimation applied to the available data, with accuracy checked via simulation and real H-maser data. No step in the described chain reduces a claimed prediction or uniqueness result to a fitted input, self-definition, or self-citation by construction; the derivation remains independent of the target quantities and relies on external measurements rather than tautological equivalence.

Axiom & Free-Parameter Ledger

4 free parameters · 1 axioms · 0 invented entities

The claim rests on the state-space model structure and the information content of differential measurements. Review is limited to the abstract, so full parameter and assumption details are unavailable.

free parameters (4)
  • white frequency noise variance
    Unknown parameter to be estimated from differential data for each clock.
  • random walk frequency noise variance
    Unknown parameter to be estimated from differential data for each clock.
  • drift
    Unknown parameter to be estimated from differential data for each clock.
  • measurement noise (co)variance
    Unknown parameter to be estimated from differential data for each clock.
axioms (1)
  • domain assumption Each clock is described by a second-order linear stochastic state-space model
    Stated explicitly in the abstract as the basis for the ensemble model.

pith-pipeline@v0.9.0 · 5454 in / 1340 out tokens · 54427 ms · 2026-05-07T14:59:25.992795+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

21 extracted references · 21 canonical work pages

  1. [1]

    and Jones, Richard H

    Barnes, James A. and Jones, Richard H. and Tryon, Peter V. and Allan, David W. , title =. Proceedings of the 14th Annual Precise Time and Time Interval (PTTI) Applications Planning Meeting , year =

    work page

  2. [2]

    , title =

    Barnes, James A. , title =. Proceedings of the 15th Annual Precise Time and Time Interval Systems and Applications Meeting , year =

    work page

  3. [3]

    and Jones, Richard H

    Tryon, Peter V. and Jones, Richard H. , title =. JOURNAL OF RESEARCH of the National Bureau of Standards , year =

    work page

  4. [4]

    and Tryon, Peter V

    Jones, Richard H. and Tryon, Peter V. , title =. JOURNAL OF RESEARCH of the National Bureau of Standards , year =

    work page

  5. [5]

    , title =

    Greenhall, Charles A. , title =. Proceedings of the Nineteenth Annual Precise Time and Time Interval (PTTI) Applications and Planning Meeting , year =

    work page

  6. [6]

    Proceedings of the 28th Annual Symposium on Frequency Control , year =

    Gray, James and Allan , title =. Proceedings of the 28th Annual Symposium on Frequency Control , year =

    work page

  7. [7]

    IEEE Transactions on Instrumentation and Measurement , year =

    Fest, Denis and Groslambert, Jacques and Gagnepain, Jean-Jacques , title =. IEEE Transactions on Instrumentation and Measurement , year =

    work page

  8. [8]

    2016 IEEE International Frequency Control Symposium (IFCS) , year =

    Vernotte, Fran. 2016 IEEE International Frequency Control Symposium (IFCS) , year =

    work page 2016

  9. [9]

    Stein, R. S. , title =. Proceedings of the 43rd Annual Symposim on Frequency Control , year =

    work page

  10. [10]

    , title =

    Barnes, James A. , title =. Proceedings of the 20th Annual Precise Time and Time Interval Systems and Applications Meeting , year =

    work page

  11. [11]

    IEEE Transactions on Instrumentation and Measurement , year =

    Audoin, Claude and Dimarcq, No\". IEEE Transactions on Instrumentation and Measurement , year =

    work page

  12. [12]

    IEEE Transactions on Instrumentation and Measurement , year =

    Premoli, Amedeo and Tavella, Patrizia , title =. IEEE Transactions on Instrumentation and Measurement , year =

    work page

  13. [13]

    Metrologia , year =

    Tavella, Patrizia and Premoli, Amedeo , title =. Metrologia , year =

    work page

  14. [14]

    and Burt, Eric A

    Torcaso, Fred and Ekstrom, Christopher R. and Burt, Eric A. and Matsakis, Demetrios N. , title =. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control , year =

    work page

  15. [15]

    Tsang, W. L. and Glower, J. D. and Bach, R. E. , title =. IEEE Transactions on Automatic Control , year =

    work page

  16. [16]

    Ferre-Pikal, E. S. and Vig, J. R. , journal =. 2009 , pages =

    work page 2009

  17. [17]

    2010 , volume=

    Galleani, Lorenzo and Tavella, Patrizia , journal=. 2010 , volume=

    work page 2010

  18. [18]

    and Ashby, Neil and Hodge, Clifford C

    Allan, David W. and Ashby, Neil and Hodge, Clifford C. , title =

    work page

  19. [19]

    Unobservable Systems: No Problem for Noise Identification , year=

    Kost, Oliver and Duník, Jindřich and Punčochař, Ivo and Straka, Ondřej , journal=. Unobservable Systems: No Problem for Noise Identification , year=

    work page

  20. [20]

    Estimation of Noise Covariance Matrices for a Linear Time-Varying Stochastic Process , author =

    work page

  21. [21]

    Measurement Difference Method: A Universal Tool for Noise Identification , year=

    Kost, Oliver and Duník, Jindřich and Straka, Ondřej , journal=. Measurement Difference Method: A Universal Tool for Noise Identification , year=

    work page