Optimum array signal processing in the presence of imperfect spatial coherence of wavefronts

Traditionally optimum-adaptive beamforming algorithms have been developed assuming fully coherent plane wavefronts, i.e., assuming a data model of point sources. In most applications this assumption is inappropriate, since the channel model has to account for different kinds of dispersion phenomena due to both the propagation environment and the array itself. Significant examples are sonar and underwater communication systems. Indeed, in such circumstances, the resulting wavefronts can be randomly distorted, usually suffering a loss of spatial coherence. Here, assuming a more realistic stochastic channel model, we analyze the performance of a traditional optimum adaptive beamformer for point sources, when the signal or the interference undergo a spatial coherence degradation. It is shown, with analytical details, that the same coherence loss, for the interference results in larger performance degradation than for the signal, Furthermore, we provide a theoretical comparison among different beamforming algorithms, based on the estimate of the channel parameters and on spatial smoothing methods.