Subspace tracking for time-varying direction-of-arrival estimation with sensor arrays

Abstract

Subspace estimation and tracking are of great use for high-resolution sensor array signal processing in aerospace and defense applications. In this chapter, several subspace tracking algorithms with different arithmetic complexities and tracking abilities are introduced. The application of these algorithms to time-varying direction-of-arrival (DOA) estimation has been presented. Particularly, by introducing a better estimate of the subspace to the conventional projection approximation subspace tracking (PAST) algorithm, two modified methods, namely modified PAST and modified orthonormal PAST, are developed for slowly varying subspace. For fast varying subspace, a Kalman filter with a variable number of measurements (KFVM) method is introduced. To improve the robustness against system model imperfections, two robust subspace tracking algorithms, that is, robust PAST and robust KFVM, are developed for scenarios where measurements are contaminated by impulsive noise. Numerical examples have been presented to demonstrate the flexibility, effectiveness, and robustness of these algorithms for subspace and DOA tracking.