Source number estimation methods base on suooprt vector machine algorithm

The source number estimation is a basic problem in the smart antenna technology. Some classic estimation algorithms have been developed in past twenty years like `AIC', `MDL', hypothesis test (`HPY'), Gerschgorin Radii (`GDE'), etc .But the estimation error will be great in the c...

Full description

Saved in:
Bibliographic Details
Main Authors: Zhao hui-qiang, Zhao bo, Quan Houde, Zhang Yu-ping
Format: Conference Proceeding
Language:English
Subjects:
Antenna arrays
characteristic vector
Colored noise
Covariance matrix
Data mining
Gaussian processes
Gerschgorin Radii
Sensor arrays
Signal to noise ratio
Source number estimation
Support vector machine classification
Support vector machines
SVM
Transmission line matrix methods
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The source number estimation is a basic problem in the smart antenna technology. Some classic estimation algorithms have been developed in past twenty years like `AIC', `MDL', hypothesis test (`HPY'), Gerschgorin Radii (`GDE'), etc .But the estimation error will be great in the circumstances of low S/N, small sample with these algorithms. This paper develops a novel method based on support vector machine (SVM), which offers more precise result than these classic algorithms .The novel algorithm extracts several classification characteristic vectors from the data matrix received by the array with GDE at first, and then constructs and trains the SVM, gets signal subspace from the output of the SVM and sources signal number from the number of vectors of the signal subspace. The simulation result confirms the conclusion.
DOI:10.1109/PACIIA.2009.5406439