A structured low-rank matrix pencil for spectral estimation and system identification

In this paper we propose a new matrix pencil based method for estimating parameters (frequencies and damping factors) of exponentially damped sinusoids in noise. The proposed algorithm estimates the signal parameters using a matrix pencil constructed from measured data. We show that the performance of the estimation can be signiicantly improved by the combination of our proposed matrix pencil algorithm and the structured low rank approximation of the data matrix. Comparison of our matrix pencil method to existing matrix pencil methods as well as to polynomial methods show that our matrix pencil method is more accurate in estimating the signal parameters. It is found through computer simulations that, for exponentially damped sinusoids, our matrix pencil method is less sensitive to noise and has a lower signal-to-noise ratio (SNR) threshold.