This thesis concerns the problem of sensor array source localization and power estimation by an acoustical array of sensors. In first the acoustical array directivity is treated. It is shown that such array is not useful for the localization of multiple sources. Adaptive arrays and high resolution methods are then introduced. They are based on the estimation of the sensor output covariance matrix and their performances overcome the natural limitations of the weighted beamforming processing. However, these methods require the use of a propagation model and are not robust to model errors. We present a new method which is an application of sparse regularization methodology to acoustical source localization using an acoustical array. Its performances are better than high-resolution methods and this method works very well in the case of correlated or uncorrelated signals, narrow band or wideband signals, near field or far field environments. Finally, a power estimation of sound sources by an acoustical array is presented. Numerical and experimental results in an anechoic room are presented showing the effectiveness of theoretical results
