Stability and Super-resolution of MUSIC and ESPRIT for Multi-snapshot Spectral Estimation

This paper studies the spectral estimation problem of estimating the locations of a fixed number of point sources given multiple snapshots of Fourier measurements collected by a uniform array of sensors. We prove novel non-asymptotic stability bounds for MUSIC and ESPRIT as a function of the noise standard deviation, number of snapshots, source amplitudes, and support. Our most general result is a perturbation bound of the signal space in terms of the minimum singular value of Fourier matrices. When the point sources are located in several separated clumps, we provide an explicit upper bound of the noise-space correlation perturbation error in MUSIC and the support error in ESPRIT in terms of a Super-Resolution Factor (SRF). The upper bound for ESPRIT is then compared with a new CramérRao lower bound for the clumps model. As a result, we show that ESPRIT is comparable to that of the optimal unbiased estimator(s) in terms of the dependence on noise, number of snapshots and SRF. As a byproduct of our analysis, we discover several fundamental differences between the single-snapshot and multi-snapshot problems. Our theory is validated by numerical experiments.

[1]  D. Donoho.  Superresolution via sparsity constraints , 1992 .

[2]  O. Papaspiliopoulos.  High-Dimensional Probability: An Introduction with Applications in Data Science , 2020 .

[3]  Laurent Demanet,et al.  Conditioning of Partial Nonuniform Fourier Matrices with Clustered Nodes , 2018, SIAM J. Matrix Anal. Appl..

[4]  Boaz Nadler,et al.  Non-Parametric Detection of the Number of Signals: Hypothesis Testing and Random Matrix Theory , 2009, IEEE Transactions on Signal Processing.

[5]  Ilan Ziskind,et al.  Detection of the number of coherent signals by the MDL principle , 1989, IEEE Trans. Acoust. Speech Signal Process..

[6]  Claudia Biermann,et al.  Mathematical Methods Of Statistics , 2016 .

[7]  F. Li,et al.  Performance analysis for DOA estimation algorithms: unification, simplification, and observations , 1993 .

[8]  Weilin Li,et al.  Stable super-resolution limit and smallest singular value of restricted Fourier matrices , 2017, Applied and Computational Harmonic Analysis.

[9]  F. Li,et al.  Sensitivity analysis of DOA estimation algorithms to sensor errors , 1992 .

[10]  R. O. Schmidt,et al.  Multiple emitter location and signal Parameter estimation , 1986 .

[11]  Thomas Kailath,et al.  Detection of number of sources via exploitation of centro-symmetry property , 1994, IEEE Trans. Signal Process..

[12]  Florian Roemer,et al.  Performance Analysis of Multi-Dimensional ESPRIT-Type Algorithms for Arbitrary and Strictly Non-Circular Sources With Spatial Smoothing , 2016, IEEE Transactions on Signal Processing.

[13]  Arthur Jay Barabell,et al.  Improving the resolution performance of eigenstructure-based direction-finding algorithms , 1983, ICASSP.

[14]  Anru R. Zhang,et al.  Rate-Optimal Perturbation Bounds for Singular Subspaces with Applications to High-Dimensional Statistics , 2016, 1605.00353.

[15]  M. Viberg,et al.  Two decades of array signal processing research: the parametric approach , 1996, IEEE Signal Process. Mag..

[16]  S. Unnikrishna Pillai,et al.  Performance analysis of MUSIC-type high resolution estimators for direction finding in correlated and coherent scenes , 1989, IEEE Trans. Acoust. Speech Signal Process..

[17]  Gil Goldman,et al.  The spectral properties of Vandermonde matrices with clustered nodes , 2019, ArXiv.

[18]  Stefan Kunis,et al.  On the smallest singular value of multivariate Vandermonde matrices with clustered nodes , 2019, Linear Algebra and its Applications.

[19]  P. Wedin.  Perturbation bounds in connection with singular value decomposition , 1972 .

[20]  Piya Pal,et al.  Cramér–Rao Bounds for Underdetermined Source Localization , 2016, IEEE Signal Processing Letters.

[21]  Wenjing Liao,et al.  Super-Resolution Limit of the ESPRIT Algorithm , 2019, IEEE Transactions on Information Theory.

[22]  Anne Ferréol,et al.  Statistical Analysis of the MUSIC Algorithm in the Presence of Modeling Errors, Taking Into Account the Resolution Probability , 2010, IEEE Transactions on Signal Processing.

[23]  Yuejie Chi,et al.  Spectral Methods for Data Science: A Statistical Perspective , 2021, Found. Trends Mach. Learn..

[24]  Gabriel Peyré,et al.  Exact Support Recovery for Sparse Spikes Deconvolution , 2013, Foundations of Computational Mathematics.

[25]  Keith Q. T. Zhang.  Probability of resolution of the MUSIC algorithm , 1995, IEEE Trans. Signal Process..

[26]  Stefan Kunis,et al.  On the condition number of Vandermonde matrices with pairs of nearly-colliding nodes , 2018, Numerical Algorithms.

[27]  Phillip A. Regalia,et al.  On the behavior of information theoretic criteria for model order selection , 2001, IEEE Trans. Signal Process..

[28]  Anthony J. Weiss,et al.  Effects of model errors on waveform estimation using the MUSIC algorithm , 1994, IEEE Trans. Signal Process..

[29]  Bhaskar D. Rao,et al.  Effect of spatial smoothing on the performance of MUSIC and the minimum-norm method , 1990 .

[30]  J. A. Nossek,et al.  Comparison between unitary ESPRIT and SAGE for 3-D channel sounding , 1999, 1999 IEEE 49th Vehicular Technology Conference (Cat. No.99CH36363).

[31]  Harry B. Lee,et al.  The Cramer-Rao bound on frequency estimates of signals closely spaced in frequency , 1992, IEEE Trans. Signal Process..

[32]  Emmanuel J. Candès,et al.  Super-Resolution from Noisy Data , 2012, Journal of Fourier Analysis and Applications.

[33]  Adriaan van den Bos,et al.  Resolution: a survey , 1997 .

[34]  Benedikt Diederichs.  Well-Posedness of Sparse Frequency Estimation , 2019 .

[35]  A. Lee Swindlehurst,et al.  A Performance Analysis ofSubspace-Based Methods in thePresence of Model Errors { Part I : The MUSIC AlgorithmA , 1992 .

[36]  Petre Stoica,et al.  MUSIC, maximum likelihood, and Cramer-Rao bound , 1989, IEEE Transactions on Acoustics, Speech, and Signal Processing.

[37]  Wenjing Liao,et al.  MUSIC for Single-Snapshot Spectral Estimation: Stability and Super-resolution , 2014, ArXiv.

[38]  Ankur Moitra,et al.  Super-resolution, Extremal Functions and the Condition Number of Vandermonde Matrices , 2014, STOC.

[39]  V. N. Bogaevski,et al.  Matrix Perturbation Theory , 1991 .

[40]  Benjamin Friedlander.  A sensitivity analysis of the MUSIC algorithm , 1990, IEEE Trans. Acoust. Speech Signal Process..

[41]  Petre Stoica,et al.  MUSIC, maximum likelihood, and Cramer-Rao bound: further results and comparisons , 1990, IEEE Trans. Acoust. Speech Signal Process..

[42]  Thomas Kailath,et al.  ESPRIT-estimation of signal parameters via rotational invariance techniques , 1989, IEEE Trans. Acoust. Speech Signal Process..

[43]  Emmanuel J. Candès,et al.  Super-Resolution of Positive Sources: The Discrete Setup , 2015, SIAM J. Imaging Sci..

[44]  B. Hofmann-Wellenhof,et al.  Introduction to spectral analysis , 1986 .

[45]  El-Hadi Djermoune,et al.  Perturbation Analysis of Subspace-Based Methods in Estimating a Damped Complex Exponential , 2009, IEEE Transactions on Signal Processing.

[46]  J. Benedetto,et al.  Super-resolution by means of Beurling minimal extrapolation , 2016, Applied and Computational Harmonic Analysis.

[47]  Florian Roemer,et al.  Analytical Performance Assessment of Multi-Dimensional Matrix- and Tensor-Based ESPRIT-Type Algorithms , 2014, IEEE Transactions on Signal Processing.

[48]  Bhaskar D. Rao,et al.  Performance analysis of ESPRIT and TAM in determining the direction of arrival of plane waves in noise , 1989, IEEE Trans. Acoust. Speech Signal Process..

[49]  Charles R. Johnson,et al.  Topics in Matrix Analysis , 1991 .

[50]  Albert Fannjiang,et al.  The MUSIC algorithm for sparse objects: a compressed sensing analysis , 2010, ArXiv.

[51]  Tapan K. Sarkar,et al.  Matrix pencil method for estimating parameters of exponentially damped/undamped sinusoids in noise , 1990, IEEE Trans. Acoust. Speech Signal Process..

[52]  Z. Bai,et al.  On detection of the number of signals in presence of white noise , 1985 .

[53]  Thomas Kailath,et al.  On the sensitivity of the ESPRIT algorithm to non-identical subarrays , 1990 .

[54]  Weilin Li,et al.  Conditioning of restricted Fourier matrices and super-resolution of MUSIC , 2019, 2019 13th International conference on Sampling Theory and Applications (SampTA).