Adaptive eigen-subspace algorithms for tracking suddenly disappearing sources

Adaptive eigen-subspace algorithms have advantages in tracking nonstationary signal parameters, for example, the directions of arrival in array signal processing. Most adaptive eigenstructure methods can track all moving and suddenly appearing sources. However, an interesting disadvantage of these algorithms is that they fail to track suddenly disappearing sources. As a result of this failure, false peaks will occur in the angular spectrum at the directions where the sources disappear. The authors show this inherent problem of adaptive eigen-subspace algorithms by theoretical derivations and simulation examples.