betaFeedback

WorksEngineering

Asymptotic Frequency-Domain FIM for multidimensional stationary and cyclostationary compound-Gaussian processes: Application to DOA estimation

Abstract

The classical Whittle formula provides a closed-form expression for the asymptotic Fisher information matrix (FIM) of stationary, zero-mean Gaussian processes (GPs) with finite-dimensional parameterizations, expressed in terms of their power spectra. While this result was recently extended to nonzero-mean, multidimensional, real-valued stationary compound-Gaussian processes (CGPs), its applicability remains limited. In this paper, we significantly broaden this theoretical framework with two primary contributions. First, we extend the Whittle formula to purely cyclostationary real-valued CGPs with parameterized cyclic spectra, and further generalize it to circular and noncircular complex-valued multidimensional stationary CGPs, encompassing both purely and almost cyclostationary CGPs. Second, we apply these novel Whittle-type formulas to evaluate the impact of source spectral characteristics on the Cramer-Rao bound (CRB) for direction-of-arrival (DOA) estimation, particularly in scenarios involving frequency offsets.

Submitted to Signal Processing

Introduction

Spherically invariant random processes (SIRPs), also known as compound-Gaussian processes (CGPs), provide a flexible framework for modeling a broad class of non-Gaussian phenomena encountered in signal processing. This class includes, among others, ϵ\epsilon-contaminated Gaussian, Student-tt, and KK-distributed processes, as well as certain symmetric α\alpha-stable processes, while also encompassing Gaussian processes as a particular case. The simplest models in this class are second-order (wide-sense) stationary parametric processes, for which the parameters of interest are estimated from a finite set of consecutive observations.

Evaluating the accuracy of parametric estimates in such models is fundamental to the analysis and design of signal processing systems. This accuracy is commonly characterized by the Cramér-Rao bound (CRB), which is determined by the inverse of the Fisher information matrix (FIM) and provides a lower bound on the covariance matrix of unbiased estimators. While the finite-sample FIM is given by the Slepian-Bangs formula [2, 3], evaluating it is computationally prohibitive for large data records. In contrast, Whittle’s formula [1] provides a closed-form expression for the asymptotic FIM rate of zero-mean, real-valued stationary Gaussian processes in terms of their power spectra. This drastically reduces computational complexity while offering a direct interpretation in terms of the spectral properties of the process. Recently, by exploiting the asymptotic properties of block Toeplitz matrices, this result was extended in [4] to real-valued, arbitrary-mean stationary SIRPs within a unified framework covering fully known, finite-dimensional parameterized, and completely unspecified density generators.

However, several important extensions are required to make such asymptotic FIM results applicable to a broader class of signal processing problems. First, many signals are frequency down-converted to baseband after filtering and are consequently represented by their complex-valued envelopes, whose distributions may be either circular or noncircular. Second, many man-made signals are not stationary but exhibit periodic statistical properties and are therefore naturally described by cyclostationary models. Such models arise in communication systems [5, 6], rotating machinery [7, 8], radar and sonar [9, 10], and power-line communications [11, 12]. Third, practical communication signals may exhibit almost cyclostationary behavior, for which the relevant second-order statistics vary periodically with time but do not necessarily admit the strict periodicity of conventional cyclostationary models. A general asymptotic FIM framework encompassing these cases is therefore of considerable interest.

In this paper, we develop a unified extension of Whittle’s formula to multidimensional stationary, cyclostationary, and almost cyclostationary CGPs. The main contributions are as follows.

  • We derive the asymptotic frequency-domain FIM rate for real-valued cyclostationary CGPs with parameterized cyclic spectra as a function of the poly-spectral matrix, thereby extending the stationary result of

[4] to processes with periodic second-order statistics.

  • We generalize the framework to multidimensional complex-valued stationary CGPs, covering both circular and noncircular processes. The latter case requires accounting for the complementary second-order statistics in addition to the conventional covariance structure.

  • We further extend the results to purely and specifically almost cyclostationary complex-valued CGPs. In particular, we establish results that enable the treatment of linearly modulated digital communication signals exhibiting cyclostationarity and their extension to specific multidimensional almost cyclostationary models.

  • We apply the resulting asymptotic FIM expressions to DOA estimation and characterize the impact of the source spectral characteristics on the corresponding CRB. Both stationary and cyclostationary source signals are considered, including scenarios with and without frequency offsets.

The proposed framework goes beyond the commonly used performance analysis based on independent and identically distributed Gaussian observations. In the considered setting, the observations are samples of continuous-time stochastic processes whose temporal and spectral characteristics can be arbitrary and whose distributions need not be Gaussian. Consequently, the resulting CRB explicitly incorporates information carried by the temporal spectrum, cyclic spectrum, and, for noncircular complex processes, complementary spectral characteristics. This provides a general analytical framework for studying how signal statistics influence parameter-estimation performance.

The remainder of this paper is organized as follows. Section II introduces the required preliminaries on real-valued multidimensional stationary CGPs and reviews the asymptotic frequency-domain FIM rate formula for these processes. Section III extends this formula to real-valued cyclostationary CGPs. Section IV further generalizes the results to circular and noncircular complex-valued multidimensional stationary CGPs, as well as to purely and specifically almost cyclostationary CGPs. Preliminary results for linearly modulated digital communication signals exhibiting cyclostationarity are also established. Section V applies the proposed asymptotic FIM rate formulas to investigate the influence of source spectral characteristics on the CRB for DOA estimation, with particular emphasis on measurements affected by frequency offsets. Finally, Section VI concludes the paper.

The following notations are used throughout this paper. Matrices and vectors are represented by bold upper-case and bold lower-case characters, respectively. Vectors are by default in column orientation, while the superscripts TT, HH and ∗* stand for transpose, conjugate transpose and conjugate, respectively. E(⋅)E(\cdot), ∣⋅∣|\cdot|, and Tr⁡(⋅)\operatorname{Tr}(\cdot) are the expectation, determinant, and trace operators, respectively. ∥⋅∥F\|\cdot\|_{F} denotes the Frobenius norm. I\mathbf{I} is the identity matrix with the appropriate dimension. ⊗\otimes is the Kronecker product of two matrices. Blkdiag⁡(B1,…,Bn\operatorname{Blkdiag}(\mathbf{B}_{1},\ldots,\mathbf{B}_{n} is the block diagonal matrix with B1,…,BnB_1,\ldots,B_n along its diagonal. 1C(f)=1\mathbb{1}_C(f)=1 if f∈Cf\in C and 1C(f)=0\mathbb{1}_C(f)=0 if f∉Cf\notin C and x∼Pxx\sim P_x means that the random variables xx has the distribution PxP_x. The acronyms r.v., p.d.f., c.d.f. and i.f.f. are used to represent random variable, probability density function, cumulative distribution function, and ”if and only if”, respectively.

Statistical Background

Preliminaries on multidimensional CGP

Let (xk)k∈Z(\mathbf{x}_k)_{k\in\mathbb{Z}} be a stochastic process with xk∈Rm\mathbf{x}_k\in\mathbb{R}^m. This process is defined as an mm-dimensional SIRP, or equivalently a CGP, if any finite collection of samples y:=(xk1T,…,xknT)T\mathbf{y}\mathrel{\mathop:}= (\mathbf{x}_{k_1}^{T},\ldots,\mathbf{x}_{k_n}^{T})^{T} with k1<…<knk_1<\ldots<k_n, follows a compound-Gaussian distribution. The p.d.f. is given by:

p(y)=∣Σ∣−1/2gnm[(y−μ)TΣ−1(y−μ)].(1)p(\mathbf{y})=|\boldsymbol{\Sigma}|^{-1/2}g_{nm}\left[(\mathbf{y}-\boldsymbol{\mu})^{T}\boldsymbol{\Sigma}^{-1}(\mathbf{y}-\boldsymbol{\mu})\right]. \tag*{(1)}

The first and second-order moments of y\mathbf{y} are assumed to exist, satisfying E(y)=μ\mathbb{E}(\mathbf{y})=\boldsymbol{\mu} and

Cov⁡(y)=Σ(2)\operatorname{Cov}(\mathbf{y})=\boldsymbol{\Sigma} \tag*{(2)}

respectively. This normalization in (2) is imposed to remove any scale ambiguity between Σ\boldsymbol{\Sigma} and gnm(⋅)g_{nm}(\cdot). The density generator gnm(⋅)g_{nm}(\cdot) is given by

gnm(t)=(2π)−nm/2∫0∞τ−nm/2exp⁡(−t/2τ) dFτ(τ),(3)g_{nm}(t)=(2\pi)^{-nm/2}\int_{0}^{\infty}\tau^{-nm/2}\exp(-t/2\tau)\,\mathrm{d}F_{\tau}(\tau), \tag*{(3)}

where FτF_{\tau} denotes the c.d.f. of the r.v. τ\tau, commonly referred to as the texture variable. According to [13 Th. 2.2], the p.d.f. in (1), with the density generator given by (3), is equivalently induced by the stochastic representation

y=μ+τΣ1/2n0,(4)\mathbf{y}=\boldsymbol{\mu}+\sqrt{\tau}\boldsymbol{\Sigma}^{1/2}\mathbf{n}_0, \tag*{(4)}

where τ\tau and n0\mathbf{n}_0 are independent, with n0∼Nnm(0,I)\mathbf{n}_0\sim\mathcal{N}_{nm}(\mathbf{0},\mathbf{I}). Under the normalization condition in (2), the texture variable satisfies E(τ)=1\mathbb{E}(\tau)=1. The density generator gnm(⋅)g_{nm}(\cdot), or equivalently the c.d.f. FτF_{\tau} of τ\tau, may be fully known, parameterized by a finite-dimensional nuisance parameter [14], or completely unspecified [15].

In the following, the above definition is applied to stationary and cyclostationary processes, with the covariance matrix Σ\boldsymbol{\Sigma} and the corresponding density generator determined by the temporal structure of the sampled observations.

Preliminaries on multidimensional stationary CGP

A CGP is said to be strictly stationary if the p.d.f. in (1) of any finite collection of samples y\mathbf{y} is invariant under arbitrary translations of the sampling indices (k1,…,kn)(k_1,\ldots,k_n). Equivalently, the v.s (x1+k,…,xn+k)(\mathbf{x}_{1+k},\ldots,\mathbf{x}_{n+k}) and (x1,…,xn)(\mathbf{x}_1,\ldots,\mathbf{x}_n) share the same joint distribution for all n∈N∗n \in\mathbb{N}^* and k∈Zk \in\mathbb{Z}. As for Gaussian processes, the compound-Gaussian structure implies that stationarity is equivalent to wide-sense stationarity. More precisely, a CGP is stationary if and only if E(xℓ)\mathrm{E}(\mathbf{x}_{\ell}) and E(xℓxℓ+kT)\mathrm{E}(\mathbf{x}_{\ell}\mathbf{x}_{\ell+k}^{T}) are independent of ℓ∈Z\ell\in\mathbb{Z} for every k∈Zk \in\mathbb{Z}. Consequently, for a stationary CGP, the random vector formed by stacking nn consecutive observations of xk\mathbf{x}_k is given by

yn≡def(x1T,…,xnT)T∈Rmn∀n∈N∗,(5)\mathbf{y}_n \mathrel{\mathop{\equiv}^{\mathrm{def}}} (\mathbf{x}_1^T,\ldots,\mathbf{x}_n^T)^T \in\mathbb{R}^{mn}\quad\forall n \in\mathbb{N}^*, \tag*{(5)}

has mean μyn≡defE(yn)=(1n⊗Im)μx\boldsymbol{\mu}_{\mathbf{y}_n} \mathrel{\mathop{\equiv}^{\mathrm{def}}} \mathrm{E}(\mathbf{y}_n) = (\mathbf{1}_n \otimes\mathbf{I}_m)\boldsymbol{\mu}_{\mathbf{x}} with μx≡defE(xℓ)\boldsymbol{\mu}_{\mathbf{x}} \mathrel{\mathop{\equiv}^{\mathrm{def}}} \mathrm{E}(\mathbf{x}_{\ell}) and covariance matrix Σyn≡defCov⁡(yn)\boldsymbol{\Sigma}_{\mathbf{y}_n} \mathrel{\mathop{\equiv}^{\mathrm{def}}} \operatorname{Cov}(\mathbf{y}_n), which is symmetric positive definite and has a block Toeplitz structure with block size mm. Its (i,j)(i,j)-th block is given by Rx(j−i)\mathbf{R}_{\mathbf{x}}(j-i) with Rx(k)≡defE[(xℓ−μx)(xℓ+k−μx)T]∈Rm×m\mathbf{R}_{\mathbf{x}}(k) \mathrel{\mathop{\equiv}^{\mathrm{def}}} \mathrm{E}[(\mathbf{x}_{\ell}-\boldsymbol{\mu}_{\mathbf{x}})(\mathbf{x}_{\ell+k}-\boldsymbol{\mu}_{\mathbf{x}})^T] \in\mathbb{R}^{m \times m}. We assume that the sequence Rx(k)\mathbf{R}_{\mathbf{x}}(k) is absolutely summable, i.e., ∑k=−∞∞∥Rx(k)∥2<∞\sum_{k=-\infty}^{\infty}\|\mathbf{R}_{\mathbf{x}}(k)\|_2 < \infty, which guarantees that the spectrum1 Rx(f)≡def∑kRx(k)e−i2πkf\mathbf{R}_{\mathbf{x}}(f) \mathrel{\mathop{\equiv}^{\mathrm{def}}} \sum_k \mathbf{R}_{\mathbf{x}}(k)e^{-i2\pi kf} is well-defined.

Asymptotic Frequency-Domain FIM Rate for Stationary CGP

Suppose that the mean μx\boldsymbol{\mu}_{\mathbf{x}} and the sequence Rx(k)\mathbf{R}_{\mathbf{x}}(k) depend on a parameter θ=(θ1,…,θr)T∈Θ⊂Rr\boldsymbol{\theta} = (\theta_1,\ldots,\theta_r)^T \in\Theta\subset\mathbb{R}^r, and consequently, so does the spectrum Rx(f)\mathbf{R}_{\mathbf{x}}(f). For notational simplicity, this dependence on θ\boldsymbol{\theta} is omitted whenever no ambiguity arises. Assume that the mapping θ↦(μx,[Rx(k)]k∈Z,Rx(f))\boldsymbol{\theta} \mapsto(\boldsymbol{\mu}_{\mathbf{x}},[\mathbf{R}_{\mathbf{x}}(k)]_{k\in\mathbb{Z}},\mathbf{R}_{\mathbf{x}}(f)) is differentiable in a neighborhood of the true parameter value and that the regularity conditions stated in [4] hold, among which is E(τ−1)<∞\mathrm{E}(\tau^{-1}) < \infty. Then, combining the Slepian-Bangs expression for the finite-sample FIM with the asymptotic theory of block Toeplitz matrices yields the following asymptotic frequency domain FIM rate for a multidimensional real-valued stationary CGP:

References

  1. [1]P. Whittle, “The analysis of multiple stationary time series,” Journal of the Royal Statistical Society, vol. 15, no. 1, pp. 125-139, 1953.DOI
  2. [2]D. Slepian, “Estimation of signal parameters in the presence of noise,” Trans. IRE Prof. Group Inform. Theory PG IT-3, pp. 68-89, 1954.DOI
  3. [3]W. J. Bangs, “Array processing with generalized beamformers,” Ph.D. dissertation, Yale Univ., New Haven, CT, USA, 1971.
  4. [4]J.-P. Delmas, H. Abeida, and S. Fortunati, ”On the parametric and semiparametric Fisher information matrix for non-zero mean stationary spherical invariant random processes,” submitted to IEEE Trans. Signal Process. Sept. 2026, (arXiv:2609.20469v1).arxiv.org/abs/2609.20469
  5. [5]E. Serpedin, F. Pandurua, I. Saria, and G. B. Giannakis, ”Bibliography on cyclostationarity,” Signal Processing, no. 85, pp. 2233-2303, 2005.
  6. [6]A. Napolitano, ”Cyclostationarity: New trends and applications,” Signal Processing, no. 120, pp. 385-408, 2016.DOI
  7. [7]J. Antoni, F. Bonnardot, A. Raad, and M. El Badaoui, ”Cyclostationary modelling of rotating machine vibration signals,” Mechanical Systems and Signal Processing, vol. 18, no. 6, pp. 1285-1314, 2004.DOI
  8. [8]R. Sun, Z. Yang, K. C. Gryllias, and X. Chen, ”Cyclostationary modeling for local fault diagnosis of planetary gear vibration signals,” Journal of Sound and Vibration, vol. 471, 115175, 2020.DOI
  9. [9]X. Zhang, M. Nabil El Korso, and M. Pesavento, ”MIMO radar target localization and performance evaluation under SIRP clutter,” Signal Processing, vol. 130, pp. 217-232, January 2017.arxiv.org/abs/1603.08969
  10. [10]F. Gini, M. Montanari, and L. Verrazzani, ”Estimation of chirp radar signals in compound-Gaussian clutter: A cyclostationary approach,” IEEE Trans. on Signal Process., vol. 48, no. 4, pp. 1029-1039, April 2000.DOI
  11. [11]Y.-R. Chien, J.-L. Lin, and H.-W. Tsao, ”Cyclostationary Impulsive Noise Mitigation in the Presence of Cyclic Frequency Offset for Narrowband Powerline Communication Systems,” Electronics, vol. 9, no. 6, 2020.DOI
  12. [12]S. Moaveninejad, A. Kumar, M. Elgenedy, N. Al-Dhahir, A. M. Tonello, and M. Magarini, ”Simpler Than FRESH Filter: A Parametric Approach for Cyclostationary Noise Generation in NB-PLC,” IEEE Commun. Letters, vol. 24, no. 7, pp. 1373-1377, April 2020.DOI
  13. [13]K. Yao, ”A representation theorem and its applications to spherically invariant random processes,” IEEE Trans. Info. Theory, vol. 19, no. 5, pp 600-608, Sept. 1973.DOI
  14. [14]H. Abeida and J.-P. Delmas, ”Slepian-Bangs formulas for parameterized density generator of elliptically symmetric distributions,” Signal Processing, vol. 205, Jan. 2023.DOI
  15. [15]S. Fortunati, J.-P. Delmas, and E. Ollila, ”Nuisance parameters and elliptically symmetric distributions: a geometric approach to parametric and semiparametric efficiency,” IEEE Trans. Info. Theory, vol. 72, no. 9, Sept 2026.arxiv.org/abs/2506.23213
  16. [16]W. A. Gardner, Cyclostationarity in communications and signal processing, New York, IEEE Press, 1994.
  17. [17]A. Napolitano, Generalizations of cyclostationary signal processing: Spectral analysis and applications John Wiley and Sons, IEEE Press, New York, 2012.DOI
  18. [18]E.D. Gladyshev, ”Periodically correlated random processes,” Soviet. Math. Dokl., vol. 2, pp. 385-388, 1961.
  19. [19]A.R. Nematollahi and T. Subba Rao, ”On the spectral density estimation of periodically correlated (cyclostationary) time series,” The Indian Journal of Statistics, vol. 67, no. 3, pp 568-589, 2005.
  20. [20]D. Ramirez, P. J. Schreier, J. Via, I. Santamaria, and L. L. Scharf, ”Detection of multivariate cyclostationary,” IEEE Trans. Signal Processing, vol. 63, no. 20, pp. 5395-5408, Oct. 2015.
  21. [21]H. L. Hurd and A. Miamee, Periodically Correlated Random Sequences: Spectral Theory and Practice, John Wiley & Sons, Book Series:Wiley Series in Probability and Statistics, March 2007.
  22. [22]S. V. Schell, ”Generalization of Whittle’s theorem to cyclostationary signals,” Proc. sixth IEEE SP Workshop on Statistical Signal and Array Processing, (Victoria, BC), Oct. 1992.
  23. [23]P. Stoica and B. C. Ng, Chap. 2: Performance bounds for blind channel estimation, Signal Processing Advances in wireless and mobile communications, trends in channel estimation and equalization, vol. 1. Prentice Hall PTR, Upper Saddle River, 2001.
  24. [24]G.B. Giannakis, ”Cyclostationary signal analysis”, chap. 17 in The digital signal processing handbook, Second edition, CRC Press, 2010.
  25. [25]P. Chevalier, R. Chauvat, and J.-P. Delmas, ”Enhanced widely linear filtering to make quasi-rectilinear signals almost equivalent to rectilinear ones for SAIC/MAIC,” IEEE Trans. Signal Processing, vol. 66, no. 6, pp. 1438-1453, March 2018.DOI
  26. [26]H. Abeida and J.-P. Delmas, ”Slepian-Bangs formula and Cramér Rao bound for circular and non-circular complex elliptical symmetric distributions,” IEEE Signal Process. Letters, vol. 26, no. 10, pp. 1561-1565, Oct. 2019.DOI
  27. [27]A. Ladaycia, K. Abed-Meraim, A. Bader, and M. S. Alouini, ”CFO and channel estimation for MISO-OFDM systems,” Processing EUSIPCO, Kos island, Greece, August 2017.
  28. [28]J.-P. Delmas and H. Abeida, “Stochastic Cramér-Rao bound for non-circular signals with application to DOA estimation,” IEEE Trans. Signal Process., vol. 52, no. 11, pp. 3192-3199, Nov. 2004.
  29. [29]H. Abeida and J.-P. Delmas, ”Refinement and derivation of statistical resolution limits for circular or rectilinear correlated sources in CES data models,” Signal Processing, vol. 195, June 2022.DOI
  30. [30]S. V. Schell and W. A. Gardner, ”The Cramer-Rao lower bound for directions of arrival of Gaussian cyclostationary signals,” IEEE Trans. Inform. Theory, vol. 38, no. 4, pp. 1418-1422. July 1992.DOI
  31. [31]S. Boyd and L. Vandenberghe, Convex optimization, Cambridge University Press, 2009.

Paper details

vlnc:W_b4b3vz2t
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Engineering

Not recorded for this paper

View statement
Looking for related papers…
Contents