A Novel Optimum SINR Beamforming in Flat Fading Macrocells

In the array processing literature several statistically optimum beamformers have been proposed, to cope with the problem of optimal source separation. Optimum signal to interference plus noise power ratio (SINR) beamforming is probably one of the best techniques because it can circumvent most of the drawbacks of other beamforming techniques, but, in many already proposed approaches, the knowledge of the correlation matrices of the desired and the interference signals, is simply assumed. Unfortunately, the estimation might not be easily evaluated in practical scenarios wherein both kind of signals are simultaneously present at the receiver. In this paper we propose a novel direction of arrival (DOA) driven statistically optimum SINR beamformer whose aim is to exploit the inherent structure of the signal space. The beamformer response is optimized in such a way that power contributions of signals belonging to the desired signal subset (DSS) are maximized at the beamformer output, while power contributions of signals belonging to the complementary subset, named the rejected signal subset (RSS) and to the noise are minimized. In addition, a practical way to estimate the correlation matrices of desired and rejected signals is also provided, along with a closed form of the optimum weight vector characterizing the beamformer. This proposed technique yields to a family of SINR optimum beamformers, parameterized by different partitions of the DOA set. Another important feature of our technique is its inherent modularity, that makes it suitable for a parallel hardware implementation, so that real-time applications can be carried ou