Matched subspace beamforming using signal subspace

In this paper, the manifold vector of the desired signal is assumed to lie in a known subspace, but the associated coordinates are random variables. This model is suitable for the fast varying environment. The so-called matched subspace beamforming is referred to as the beamformer resolving the signal that are drawn from this multi-dimension subspace. By jointly using such subspace and the signal subspace, we can estimate the multi-rank effect due to the desired signal on the received signal covariance. It is also found that this estimator is insensible to the overestimation of the signal dimension. Numerical results demonstrate the robustness of the proposed beamformer to the subspace perturbation and overestimation of signal dimension.