Sinusoidal signals with random amplitude: least-squares estimators and their statistical analysis

The asymptotic properties of constrained and unconstrained least-squares estimates of the parameters of a random-amplitude sinusoid are analyzed. An explicit formula for the asymptotic covariance matrix of the estimation errors is derived for both the constrained and unconstrained estimators. Accuracy aspects are investigated with the following main results. For a certain weighting matrix, which is shown to be the same for the constrained and unconstrained methods, the estimation errors achieve their lower bounds. It is proven that in the optimal case, the constrained method always outperforms the unconstrained method. It is also proven that the accuracy of the optimal estimators improves as the number of least-squares equations increases. A formula for the sample length needed for the asymptotic theory to hold is derived, and its dependence on the lowpass modulating sequence is stressed. Simulations provide illustrations of the difference between the constrained and unconstrained estimators as well as the difference between the optimal and basic estimates. The influence of the number of least-squares equations and the characteristics of the lowpass envelope on the estimation accuracy is also investigated.

[1]  Olivier Besson,et al.  On estimating the frequency of a sinusoid in autoregressive multiplicative noise , 1993, Signal Process..

[2]  Boualem Boashash,et al.  Analysis of FM signals affected by Gaussian AM using the reduced Wigner-Ville trispectrum , 1993, 1993 IEEE International Conference on Acoustics, Speech, and Signal Processing.

[3]  T. W. Anderson,et al.  On the Asymptotic Distribution of the Autocorrelations of a Sample from a Linear Stochastic Process , 1964 .

[4]  Roger F. Dwyer,et al.  Fourth‐order spectra of Gaussian amplitude‐modulated sinusoids , 1991 .

[5]  S. Bruzzone,et al.  Information tradeoffs in using the sample autocorrelation function in ARMA parameter estimation , 1984 .

[6]  Ken Sharman,et al.  Performance evaluation of the modified Yule-Walker estimator , 1985, IEEE Trans. Acoust. Speech Signal Process..

[7]  T. Söderström,et al.  Optimal instrumental variable estimates of the AR parameters of an ARMA process , 1985 .

[8]  Petre Stoica,et al.  Overdetermined Yule-Walker estimation of the frequencies of multiple sinusoids: Accuracy aspects , 1989 .

[9]  T. Söderström,et al.  High-order Yule-Walker equations for estimating sinusoidal frequencies: the complete set of solutions , 1990 .

[10]  Torsten Söderström,et al.  On estimating the noise power in array processing , 1992, Signal Process..

[11]  Olivier Besson.  Improved detection of a random amplitude sinusoid by constrained least-squares technique , 1995, Signal Process..

[12]  B. Friedlander,et al.  The Modified Yule-Walker Method of ARMA Spectral Estimation , 1984, IEEE Transactions on Aerospace and Electronic Systems.

[13]  Ananthram Swami,et al.  Multiplicative noise models: parameter estimation using cumulants , 1994, Fifth ASSP Workshop on Spectrum Estimation and Modeling.

[14]  A. A. Beex.  Efficient generation of ARMA cross convariance sequences , 1985, ICASSP '85. IEEE International Conference on Acoustics, Speech, and Signal Processing.

[15]  P. Stoica.  On a procedure for testing the order of time series , 1981 .

[16]  M. Bartlett.  On the Theoretical Specification and Sampling Properties of Autocorrelated Time‐Series , 1946 .

[17]  C. K. Yuen,et al.  Digital spectral analysis , 1979 .

[18]  Petre Stoica,et al.  Asymptotic properties of the high-order Yule-Walker estimates of sinusoidal frequencies , 1989, IEEE Trans. Acoust. Speech Signal Process..

[19]  Petre Stoica,et al.  On instrumental variable estimation of sinusoid frequencies and the parsimony principle , 1986 .

[20]  J. Cadzow,et al.  Spectral estimation: An overdetermined rational model equation approach , 1982, Proceedings of the IEEE.