Direct state determination of multiple sources with intermittent emission

This paper investigates the direct state determination problem from passive measurements made with a moving antenna array in the case of a time-varying number of emitting sources. We derive the Cramér-Rao Bound (CRB) for the estimation problem and find an approximation that is applicable for a large number of observations. We use two Subspace Data Fusion (SDF) approaches to solve the estimation problem. Therein, subspaces are formed in the pre-processing step from the raw antenna outputs at all positions of the moving array. Then the state parameters of interest (e.g. position, velocity) are estimated directly from a cost function that results from fusing all subspaces. The SDF approaches are based on the Multiple Signal Classification (MUSIC) and on the Subspace Fitting (SSF) method using a low- and high-dimensional optimization, respectively. In simulations, we find that the SSF-SDF approach outperforms the MUSIC-SDF approach.