Genetic algorithm based maximum likelihood DOA estimation

Author

Li, M. ; Lu, Y.

Author_Institution

Nanyang Technol. Univ., Singapore

fYear

2002

fDate

15-17 Oct. 2002

Firstpage

502

Lastpage

506

Abstract

The maximum likelihood (ML) direction-of-arrival (DOA) estimation method was one of the first to be investigated. For a long time, the complexity and computational load of maximizing the multivariable, highly nonlinear likelihood function prevented it from popular. We present the genetic algorithm (GA) for computing exact solutions to the likelihood function with almost guarantee of global convergence. The performance of GA-based ML and multiple signal classification (MUSIC) algorithm have been compared for a variety of scenarios of SNR, DOA separation, number of snapshots, and computational cost. The relationship between the ML technique and MUSIC is also investigated.

Keywords

array signal processing; computational complexity; convergence of numerical methods; direction-of-arrival estimation; genetic algorithms; maximum likelihood estimation; nonlinear functions; signal classification; source separation; DOA separation; GA-based ML algorithm; MLE; MUSIC algorithm; SNR; computational complexity; computational cost; computational load; direction-of-arrival estimation; exact solutions; genetic algorithm; genetic algorithm based DOA estimation; global convergence; maximum likelihood estimation; multiple signal classification; multivariable likelihood function; narrow-band signals; nonlinear likelihood function complexity; sensor array systems; snapshots; Classification algorithms; Computational efficiency; Computational modeling; Convergence; Data models; Direction of arrival estimation; Genetic algorithms; Maximum likelihood estimation; Multiple signal classification; Sensor arrays;

fLanguage

English

Publisher

iet

Conference_Titel

RADAR 2002

Conference_Location

Edinburgh, UK

ISSN

0537-9989

Print_ISBN

0-85296-750-0

Type

conf

DOI

10.1109/RADAR.2002.1174766

Filename

1174766