Subspace stability in high resolution direction finding and signal enumeration

Subspace-based array signal processing methods implicitly require an accurate partitioning of the sampled data into signal and orthogonal ('noise') subspaces. The authors exploit the stability of the signal subspace for persistent signal sources coupled with the a priori knowledge of the array manifold. In-water sonar data is used to demonstrate the effectiveness of this approach in situations where conventional information-theoretic criteria fail. Although a variety of formulations are possible which exploit the model validity assessment and subspace stability, the proposed SSET (Subspace Stability Exploitation Tracker) approach presented in this paper is attractive due to the relatively low computational demands. Essentially, the approach involves applying multiple hypothesis target tracking algorithms to the movement of potential signal roots in the complex plane derived from matrix-shifting implementations of subspace DOA estimation algorithms. Due to the relatively low computational demands and indifference to the noise covariance structure, SSET is appropriate for real-time in-water implementation.