Title :
A new use of ESPRIT for detection and model order estimation
Author :
Silverstein, Seth D.
Author_Institution :
GE Corp. Res. & Dev., Schenectady, NY, USA
Abstract :
This work studies the distribution of the ESPRIT roots in the complex z plane. Real source roots are tightly clustered about their true values, while pseudoroots corresponding to overestimated nonexisting sources are distributed broadly with a phase angle bias away from the real source roots. The ESPRIT roots distributions can be coupled with Neyman-Pearson tests for detection/system model order estimation. A mathematical derivation of the asymptotic distribution of the pseudoroots is given.<>
Keywords :
array signal processing; matrix algebra; parameter estimation; signal detection; statistical analysis; ESPRIT roots distribution; Neyman-Pearson tests; asymptotic distribution; complex z plane; frequency estimates; model order estimation; phase angle bias; pseudoroots; real source roots; signal detection; uniform linear receiving array; Covariance matrix; Eigenvalues and eigenfunctions; Frequency estimation; Narrowband; Research and development; Sensor arrays; Testing; Vectors;
Journal_Title :
Signal Processing Letters, IEEE
