Optimal array signal processing in unknown noise environments via eigendecomposition approaches

In array signal processing, we most commonly use the spatially white noise model, and most of the high resolution methods are established on such a noise model. However, in real environments, the noise model is often either unknown or undeterminable. This may cause the high resolution methods to suffer severe performance degradation. In this paper, under the assumption that noise correlation is spatially limited, using two separated arrays, we propose a new approach for consistent directions of arrival (DOA) estimations in unknown noise environments. This new method can be also applied in radar or sonar tracking and time series analysis. For a single array the new method is also applicable.