Adaptive noise subspace processing for direction finding in sensor arrays

High-resolution algorithms for the detection and estimation of Directions Of Arrival (DOA) such as MUSIC, lead to accurate results but require the computation of the noise-subspace through an expensive covariance matrix eigendecomposition. Adaptive estimators of the noise-subspace can be very useful in a non-stationary environment when the convergence is possible with a few number of snapshots. Some adaptive methods are presented showing that an indirect noise-subspace estimation through a signal subspace estimation can be advantageous both in terms of convergence rate and computation complexity during each update. Some computer simulations examples showing performances are provided.