Stability Exploitation and Subspace Array Processing

摘要

Detecting and characterizing signals arriving at a sensor array is a problem ofpractical importance in aerospace, biomedical, geological, and sonar signal processing. If the simplifying assumption of narrowband distinct signal sources can be made, the so-called high-resolution techniques, which are also known as eigenvalue methods or signal subspace methods, may be applied and offer a promise of complete and unambiguous assessment of the environment. One of the impediments to practical application of these concepts has been implicit requirement for well-known noise structures and precise array calibration. Herein we introduce a class of techniques termed subspace stability methods which relax those restrictions by exploiting the temporal stability of the signal subspace. These are demonstrated to effectively process sonar array data against which conventional subspace processing fails. The most promising variation is the Subspace Stability Exploitation Tracker (SSET) which couples signal subspace DOA estimation algorithms with multiple target tracking techniques for accurate signal enumeration and characterization. A novel proof of the validity of spatial smoothing to permit processing of coherent wavefronts is offered using a vector subspace perspective. This viewpoint then suggests new algorithms for coherent signal processing. Finally, in addition to developing an array calibration algorithm amenable to on-line processing, prominent array signal processing techniques are described in a tutorial fashion and practical aspects of their performance and implementation discussed.