Title :
Estimation of the number of signals from features of the covariance matrix: a supervised approach
Author :
Costa, Pascale ; Grouffaud, Joel ; Larzabal, Pascal ; Clergeot, Henri
Author_Institution :
Lab. d´´Electr., Signaux et Robotique, Ecole Normale Superieure de Cachan, France
fDate :
11/1/1999 12:00:00 AM
Abstract :
The purpose of this paper is to provide a fast and simplified detection test for use in the presence of a small number of sources (from 0-2), which is able to accommodate correlated paths and nonwhite noise; conventional eigenvalue-based criteria are unable to do so. For a uniform linear array, using common sense arguments, a small set of significant features of the covariance matrix are used as inputs to a neural net. The nonlinear transfer function of the neural net is adjusted by supervised training to provide the discriminant functions for order selection in its outputs. Results from the net are then compared with conventional criteria and demonstrate superior performance, in particular, for correlated sources and small sample sizes. Training may be introduced for known nonwhite noise, which serves to maintain high performance for reasonable correlation lengths
Keywords :
array signal processing; correlation theory; covariance matrices; learning (artificial intelligence); multilayer perceptrons; transfer functions; correlated paths; covariance matrix; detection test; discriminant functions; features; neural net; nonlinear transfer function; nonwhite noise; order selection; small sample sizes; supervised approach; supervised training; uniform linear array; Antenna arrays; Array signal processing; Covariance matrix; Eigenvalues and eigenfunctions; Multilayer perceptrons; Neural networks; Sensor arrays; Signal processing; Testing; Transfer functions;
Journal_Title :
Signal Processing, IEEE Transactions on