Title :
Spectral estimation through parallel filtering
Author :
Khanshan, Amir H. ; Amindavar, Hamidreza
Author_Institution :
Islamic Azad Univ. (IAU), Tehran
Abstract :
This paper, is a study of a recent approach to spectral estimation. The key is to define a complex function f which its value on the unit circle provides an estimate to the power spectral density (PSD). The value of this complex function on some desired points outside but near the unit circle can be obtained through a bank of filters. Then the value of f on the unit circle is acquired by interpolation. We consider both nearest neighborhood (NN) and Nevanlinna-Pick interpolation (NPI). We will compare the performance of NPI approach with existing estimators such as MUSIC or Welch in different cases. Simulation results show that NPI performs much better but requires a proper selection of interpolation points which may depend on the PSD itself.
Keywords :
filtering theory; interpolation; signal classification; MUSIC; Nevanlinna-Pick interpolation; Welch; complex function; filter bank; nearest neighborhood; parallel filtering; power spectral density; spectral estimation; Circuit theory; Colored noise; Filter bank; Filtering; Interpolation; Iterative algorithms; Multiple signal classification; Neural networks; Paper technology; Robust control; Nevanlinna-Pick interpolation; Spectral estimation; filter bank; nearest neighborhood;
Conference_Titel :
Telecommunication Networks and Applications Conference, 2007. ATNAC 2007. Australasian
Conference_Location :
Christchurch
Print_ISBN :
978-1-4244-1557-1
Electronic_ISBN :
978-1-4244-1558-8
DOI :
10.1109/ATNAC.2007.4665294
