PERFORMANCE COMPARISON OF HIGH-RESOLUTION BEARING ESTIMATION ALGORITHMS USING SIMULATED AND SEA TEST DATA

Many high resolution bearing estimation algorithms that have been deve loped in recent years require the explicit calculation of the eigenvec tors and the eigenvalues of the cross spectral matrix of sensor output s. The major differences in these algorithms are the way in which they alter the eigenvalues to give a reweighting of the eigenvectors. For example, in the MUSIC algorithm the eigenvalues corresponding to those eigenvectors in the signal subspace are set to zero while the eigenva lues in the noise subspace are set to unity. These high resolution tec hniques, while providing good bearing resolution, are however, sensiti ve to the presence of system errors and deviations from the noise mode l that has been assumed in the algorithm development. One technique th at has been successfully applied to improve the performance of these a lgorithms is to preprocess the data. Various preprocessing techniques are currently available. The performance of two preprocessing algorith ms, namely the beam space and the recently developed sector focused st ability algorithms are compared with both the Capon and the MUSIC esti mators. This performance assessment is undertaken first using simulate d data, where the performance of each algorithm is assessed in the pre sence of finite averaging errors and decreasing target signal strength . Secondly, the performance of each algorithm is assessed using sea te st data collected with an experimental planar array.