DOA tracking via simultaneous angle-source update (SASU)

This paper focuses on the DOA (direction of arrival) tracking problem with uniform linear arrays. In the literature, there are two common approaches: one is to first use MUSIC based algorithms to estimate the signal subspace and then calculate the arrival angles from it; the other is to use Kalman filtering (KF) techniques to track the angles directly. This paper emphasizes on the fast movement case and hence follows the second approach. The key feature of this work is to simultaneously update the arrival angles and the source signals in the KF step by treating the source signals as a function of the arrival angles. Hence the term Simultaneous Angle-Souce Update (SASU). On the theoretical front, we show the deep connections between SASU and the two common approaches in the literature. On the more practical side, a Newton method is developed to solve the optimization problem in the KF step. Simulations demonstrate that SASU is more robust and accurate than the benchmark algorithms, especially when uncertain and fast moving sources are concerned.