Genetic algorithms for spatial spectral estimation

The authors propose a class of algorithms for estimating the directional spectrum of a traveling wavefield from noisy data collected by an array of spatially distributed sensors that need not be regularly spaced. They consider the application of genetic algorithms to find the maximum likelihood parameter estimates of a set of narrowband point wavefield sources. They include some experimental results that show that maximum likelihood techniques in conjunction with genetic algorithms can lead to better parameter estimates than conventional techniques such as MUSIC, ESPRIT, and others