pith. sign in

arxiv: 1906.10427 · v1 · pith:3LU3JVWFnew · submitted 2019-06-25 · 💻 cs.IT · math.IT

On the Relationship Between Measures of Relative Efficiency for Random Signal Detection

Nagananda , K. G This is my paper

Pith reviewed 2026-05-25 16:18 UTC · model grok-4.3

classification 💻 cs.IT math.IT
keywords relative efficiencyasymptotic relative efficiencyrandom signal detectionPitman efficiencyTaylor expansiontest statisticsignal detection theory
0
0 comments X

The pith

A formula connects relative efficiency to its asymptotic limit for random-signal detectors via higher-order Taylor terms.

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

The paper derives an explicit relationship between the finite-sample relative efficiency (RE) and the Pitman asymptotic relative efficiency (ARE) when the signal to be detected is random rather than constant. Earlier results had shown how RE converges to ARE for constant signals in additive i.i.d. noise; the random-signal case had remained open. By keeping additional terms in the Taylor expansion of the expected value of the test statistic under the alternative, the authors obtain a correction formula that describes the approach of RE to ARE as signal strength vanishes. The derivation supplies a practical bridge between the hard-to-compute small-sample RE and the analytically tractable ARE. Preliminary comparisons indicate that the rate of convergence for random signals differs from the constant-signal setting.

Core claim

A relationship between RE and ARE for random signal detection is established by retaining the higher-order terms in the Taylor series expansion of the mean of the test statistic under the alternative hypothesis; the resulting expression describes the convergence of the finite-sample efficiency measure to its asymptotic value.

What carries the argument

Higher-order terms in the Taylor series expansion of the mean of the test statistic under the alternative hypothesis, which furnish the correction that links RE to ARE.

If this is right

  • The correction formula supplies an analytical handle on detector performance at finite but moderate sample sizes where full RE evaluation remains intractable.
  • Convergence rates of RE to ARE can now be compared directly between random-signal and constant-signal detection.
  • Designers can use the extra terms to adjust sample-size requirements when the signal is unknown rather than deterministic.
  • The same expansion technique may be applied to other detectors whose test statistics admit a Taylor expansion of their means.

Where Pith is reading between the lines

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

  • Detector comparisons that previously relied only on ARE may now be refined by the correction term when sample sizes are only moderately large.
  • The random-signal formula could be inserted into sequential detection procedures to tighten stopping boundaries at finite horizons.
  • Numerical verification of the formula would immediately yield practical tables of required sample sizes for common random-signal models.

Load-bearing premise

The higher-order Taylor terms of the test-statistic mean under the alternative are enough to capture how RE converges to ARE for random signals, just as they do for constant signals.

What would settle it

Direct Monte-Carlo computation of the exact RE for small sample sizes in a Gaussian-noise random-signal detection problem that fails to match the numerical value predicted by the derived correction formula.

read the original abstract

Relative efficiency (RE), the Pitman asymptotic relative efficiency (ARE) and efficacy are important relative performance measures of signal detection techniques. These measures allow comparing two detectors in terms of the relative sample sizes they require to achieve the same prescribed level of false alarm and detection probabilities. While the finite-sample-size measure RE is useful to analyze the small sample behavior of detectors, in practice it is difficult to compute. In the limit as the signal strength approaches zero at an appropriate rate, the RE converges to an asymptotic limit only for very large sample sizes. This limiting ratio of the number of samples is the ARE, which lends analytical tractability, but does not provide insights into the finite sample behavior of detectors which is important for practical applications. This led researchers to study the convergence of RE to ARE, and has been well-reported for the problem of constant signal detection in additive, independent, and identically distributed noise. When the signal to be detected is random ({\ie}, unknown), such a convergence analysis is lacking in the literature and is the focus of this paper. A relationship between RE and ARE for random signal detection is established. We use the higher-order terms in the Taylor series expansion of the mean of the test statistic under the alternative hypothesis to derive this formula. We present preliminary remarks on the convergence of RE to ARE for random signals in comparison to that for constant signal detection.

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 manuscript claims to establish a relationship between finite-sample relative efficiency (RE) and Pitman asymptotic relative efficiency (ARE) for detectors of random (unknown) signals. It derives this by retaining higher-order terms in the Taylor series expansion of the mean of the test statistic under the alternative hypothesis, and offers preliminary remarks comparing the rate of convergence of RE to ARE for random signals versus the constant-signal case in additive i.i.d. noise.

Significance. If the derivation is valid under appropriate conditions, the result would usefully extend existing convergence analyses (previously limited to deterministic constant signals) to the practically relevant random-signal setting. This could aid finite-sample performance assessment of detectors when signal strength vanishes at a controlled rate. The work addresses a documented gap in the literature and employs a standard analytic tool (higher-order Taylor expansion) in a manner that, if rigorously justified, would be a modest but concrete contribution to detection theory.

major comments (2)
  1. [Abstract] Abstract and derivation outline: the central claim that higher-order Taylor terms of the test-statistic mean (after averaging over the random-signal distribution) govern the RE-to-ARE convergence rate is asserted without stating the required regularity conditions (e.g., existence and uniformity of joint moments of signal and noise, differentiability of the statistic, or the precise rate at which signal strength vanishes). These conditions are load-bearing because the averaging step can alter the orders of the retained terms relative to the constant-signal case; without them the truncation argument does not automatically carry over.
  2. [Preliminary remarks] Preliminary remarks section: no explicit verification (analytic example, moment calculation, or numerical check) is supplied showing that the retained higher-order terms indeed produce a non-trivial correction to the ARE that matches the finite-sample RE for a random-signal detector. This absence makes it impossible to confirm that the claimed relationship is non-vacuous or that the orders remain controlling after the expectation over the signal distribution.
minor comments (2)
  1. Notation for the test statistic and its moments is introduced without an explicit equation reference or table summarizing the expansion orders; adding a displayed equation for the Taylor series (with remainder term) would improve clarity.
  2. [Abstract] The abstract states that the RE-ARE relationship 'is established' but supplies no sample-size or signal-strength scaling that would allow a reader to reproduce the leading correction term; a brief statement of the scaling regime would help.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive report. The comments correctly identify places where additional rigor and illustration would strengthen the manuscript. We address each point below and will incorporate revisions accordingly.

read point-by-point responses

  1. Referee: [Abstract] Abstract and derivation outline: the central claim that higher-order Taylor terms of the test-statistic mean (after averaging over the random-signal distribution) govern the RE-to-ARE convergence rate is asserted without stating the required regularity conditions (e.g., existence and uniformity of joint moments of signal and noise, differentiability of the statistic, or the precise rate at which signal strength vanishes). These conditions are load-bearing because the averaging step can alter the orders of the retained terms relative to the constant-signal case; without them the truncation argument does not automatically carry over.

    Authors: We agree that the required regularity conditions should be stated explicitly. The derivation in Sections 3 and 4 implicitly relies on the existence of sufficiently many joint moments of the signal and noise, twice differentiability of the test statistic, and the standard Pitman rate at which signal strength vanishes. Because the expectation over the random-signal distribution can change the orders of the Taylor terms, we will add a dedicated paragraph listing these assumptions in the revised abstract and at the beginning of Section 3. revision: yes

  2. Referee: [Preliminary remarks] Preliminary remarks section: no explicit verification (analytic example, moment calculation, or numerical check) is supplied showing that the retained higher-order terms indeed produce a non-trivial correction to the ARE that matches the finite-sample RE for a random-signal detector. This absence makes it impossible to confirm that the claimed relationship is non-vacuous or that the orders remain controlling after the expectation over the signal distribution.

    Authors: The preliminary remarks compare the resulting convergence rates for random versus constant signals, but we acknowledge that an explicit verification (analytic or numerical) demonstrating that the higher-order correction is non-vacuous after averaging over the signal distribution is missing. We will add a short analytic example in the revised preliminary-remarks section that computes the first two correction terms for a simple Gaussian random-signal model and shows that they produce a measurable deviation from the ARE that is consistent with direct finite-sample RE evaluation. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation uses standard Taylor expansion independent of result

full rationale

The paper claims to establish a relationship between RE and ARE for random signals by retaining higher-order terms in the Taylor expansion of the mean of the test statistic under the alternative hypothesis. This is a direct analytic step using a standard mathematical tool (Taylor series) applied to the detector statistic; the resulting formula is not shown to be equivalent to its inputs by construction, nor does it rely on fitted parameters renamed as predictions, self-citations that bear the central load, or imported uniqueness theorems. The abstract and reader's summary indicate the method is self-contained against external benchmarks and does not reduce the claimed relationship to a tautology or prior self-result. No load-bearing circular steps are identifiable from the provided text.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No free parameters, axioms, or invented entities are identifiable from the abstract.

pith-pipeline@v0.9.0 · 5771 in / 1030 out tokens · 40680 ms · 2026-05-25T16:18:29.197653+00:00 · methodology

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

30 extracted references · 30 canonical work pages

  1. [1]

    The relation between Pitman’s asymptotic relative efficiency of two tests and the correlation coefficient betwe en their test statistics,

    C. van Eeden, “The relation between Pitman’s asymptotic relative efficiency of two tests and the correlation coefficient betwe en their test statistics,” The Ann. Math. Stat. , vol. 34, no. 4, pp. 1442–1451, Dec. 1963

    work page 1963

  2. [2]

    H. V . Poor, An Introduction to Signal Detection and Estimation . Springer, 1988

    work page 1988

  3. [3]

    Robust techniques for signal processing: A survey,

    S. A. Kassam and H. V . Poor, “Robust techniques for signal processing: A survey,” Proc. IEEE , vol. 73, no. 3, pp. 433–481, Mar. 1985

    work page 1985

  4. [4]

    Optimum data quantization in signal detec tion,

    S. A. Kassam, “Optimum data quantization in signal detec tion,” in Communications and Networks: A Surrey of Recent Advances , I. F. Blake and H. V . Poor, Eds. Springer, New Y ork, 1985, pp. 72–110

    work page 1985

  5. [5]

    Nonparametric detection,

    J. B. Thomas, “Nonparametric detection,” Proc. IEEE , vol. 58, no. 5, pp. 623–631, May 1970

    work page 1970

  6. [6]

    Small-sample efficiencies of ra nk tests,

    P . Papantoni-Kazakos, “Small-sample efficiencies of ra nk tests,” IEEE Trans. Inf. Theory , vol. 21, no. 2, pp. 150–157, Mar. 1975

    work page 1975

  7. [7]

    The asymptotic relativ e efficiency of score tests in a generalized linear model with surrogate c ovariates,

    T. D. Tosteson and A. A. Tsiatis, “The asymptotic relativ e efficiency of score tests in a generalized linear model with surrogate c ovariates,” Biometrika, vol. 75, no. 3, pp. 507–514, Sep. 1988

    work page 1988

  8. [8]

    A. W. van der V aart, Asymptotic Statistics. Cambridge University Press, 1998

    work page 1998

  9. [9]

    V . E. Bening, Asymptotic Theory of Testing Statistical Hypotheses . De Gruyter, 2000

    work page 2000

  10. [10]

    Nikitin, Asymptotic Efficiency of Nonparametric Tests

    Y . Nikitin, Asymptotic Efficiency of Nonparametric Tests . Cambridge University Press, 2009

    work page 2009

  11. [11]

    Numerical results on the c onvergence of relative efficiencies,

    J. H. Miller and J. B. Thomas, “Numerical results on the c onvergence of relative efficiencies,” IEEE Trans. Aerosp. Electron. Syst. , vol. AES-11, no. 2, pp. 204–209, Mar. 1975

    work page 1975

  12. [12]

    Detection in Laplace noise,

    R. J. Marks, G. L. Wise, D. G. Haldeman, and J. L. Whited, “ Detection in Laplace noise,” IEEE Trans. Aerosp. Electron. Syst. , vol. AES-14, no. 6, pp. 866–872, Nov. 1978

    work page 1978

  13. [13]

    A relative ef ficiency study of some popular detectors,

    D. L. Michalsky, G. L. Wise, and H. V . Poor, “A relative ef ficiency study of some popular detectors,” J. Franklin Inst. , vol. 313, no. 3, pp. 135–148, Mar. 1982

    work page 1982

  14. [14]

    Some relative efficiency resul ts for detectors for time varying signals,

    Y .-C. Liu and G. L. Wise, “Some relative efficiency resul ts for detectors for time varying signals,” in Proc. IEEE Int. Conf. Commun. , 1983, pp. 1592–1596

    work page 1983

  15. [15]

    Detector relative efficienci es in the presence of Laplace noise,

    M. I. Dadi and R. J. Marks, “Detector relative efficienci es in the presence of Laplace noise,” IEEE Trans. Aerosp. Electron. Syst. , vol. AES-23, no. 4, pp. 568–582, Jul. 1987

    work page 1987

  16. [16]

    Detectability of signals in Laplace no ise,

    C. W. Helstrom, “Detectability of signals in Laplace no ise,” IEEE Trans. Aerosp. Electron. Syst. , vol. 25, no. 2, pp. 190–196, Mar. 1989

    work page 1989

  17. [17]

    Approximate analysis of the convergence of relative efficiency to ARE for known signal detection,

    R. S. Blum and S. A. Kassam, “Approximate analysis of the convergence of relative efficiency to ARE for known signal detection,” IEEE Trans. Inf. Theory , vol. 37, no. 1, pp. 199–206, Jan. 1991

    work page 1991

  18. [18]

    Asymptotic expansions a nd rate of convergence in signal detection,

    M. V . Burnashev and H. V . Poor, “Asymptotic expansions a nd rate of convergence in signal detection,” IEEE Trans. Inf. Theory, vol. 41, no. 6, pp. 1773–1787, Nov. 1995

    work page 1995

  19. [19]

    A note on the limiting relative efficienc y of the Wald sequential probability ratio test,

    R. Bechhofer, “A note on the limiting relative efficienc y of the Wald sequential probability ratio test,” J. Amer . Stat. Assoc., vol. 55, no. 292, pp. 660–663, 1960

    work page 1960

  20. [20]

    Relative efficiency of the Wald SPRT and t he Chernoff information number,

    M. Sakaguchi, “Relative efficiency of the Wald SPRT and t he Chernoff information number,” Kodai Math. Seminar Reports , vol. 19, no. 2, pp. 138–146, 1967

    work page 1967

  21. [21]

    Relative efficiency of the sequential probability ratio test in signal detection,

    S. Tantaratana and J. B. Thomas, “Relative efficiency of the sequential probability ratio test in signal detection,” IEEE Trans. Inf. Theory , vol. 24, no. 1, pp. 22–31, Jan. 1978

    work page 1978

  22. [22]

    Compressive sensin g-based de- tection with multimodal dependent data,

    T. Wimalajeewa and P . K. V arshney, “Compressive sensin g-based de- tection with multimodal dependent data,” IEEE Trans. Signal Process. , vol. 66, no. 3, pp. 627–640, Feb. 2018

    work page 2018

  23. [23]

    On the optimality of likel ihood ratio test for prospect theory-based binary hypothesis testing,

    S. Gezici and P . K. V arshney, “On the optimality of likel ihood ratio test for prospect theory-based binary hypothesis testing, ” IEEE Signal Process. Letters, vol. 25, no. 12, pp. 1845–1849, Dec. 2018

    work page 2018

  24. [24]

    New rob ust statistics for change detection in time series of multivariate SAR imag es,

    A. Mian, G. Ginolhac, J. Ovarlez, and A. M. Atto, “New rob ust statistics for change detection in time series of multivariate SAR imag es,” IEEE Trans. Signal Process. , vol. 67, no. 2, pp. 520–534, Jan. 2019

    work page 2019

  25. [25]

    On multisensor detection of improper si gnals,

    J. K. Tugnait, “On multisensor detection of improper si gnals,” IEEE Trans. Signal Process. , vol. 67, no. 4, pp. 870–884, Feb. 2019

    work page 2019

  26. [26]

    A binning appr oach to quickest change detection with unknown post-change distri bution,

    T. S. Lau, W. P . Tay, and V . V . V eeravalli, “A binning appr oach to quickest change detection with unknown post-change distri bution,” IEEE Trans. Signal Process. , vol. 67, no. 3, pp. 609–621, Feb. 2019

    work page 2019

  27. [27]

    Locally optimum detection of s ignals in a generalized observation model: the random signal case,

    I. Song and S. A. Kassam, “Locally optimum detection of s ignals in a generalized observation model: the random signal case,” IEEE Trans. Inf. Theory , vol. 36, no. 3, pp. 516–530, May 1990

    work page 1990

  28. [28]

    E. L. Lehmann and J. P . Romano, Testing Statistical Hypotheses, 3rd ed. Springer, 2005

    work page 2005

  29. [29]

    Rates of convergence of estimates and te st statistics,

    R. R. Bahadur, “Rates of convergence of estimates and te st statistics,” The Ann. Math. Stat. , vol. 38, no. 2, pp. 303–324, Apr. 1967

    work page 1967

  30. [30]

    Rate of convergence of estimators based o n sample mean,

    S.-S. Perng, “Rate of convergence of estimators based o n sample mean,” The Ann. Stat. , vol. 6, no. 5, pp. 1048–1056, 1978

    work page 1978