Application of Digital Extrapolation in Array Processing

摘要

In the area of signal processing, considerable attention has been devoted to digital array processing in recent years. This attention is due to the increasingly wide use of array processing for both civilian and military purposes. Digital beamforming, for example, is popular because of its advantages in speed, accuracy, etc., over conventional analog beamforming. Many papers concerning digital beamforming have recently been published. It has been shown that the quality of performance of a beamformer, such as beam pattern, signal-to-noise ratio, etc., depends on a great extent on the number of sensors used, i.e., the more sensors used, the better the beam pattern becomes. In a practical situation, however, the number of sensors may be restricted by economical reasons or physical restrictions. In this situation one may weight the output of each sensor before beamforming. This helps to some extent, although the improvement is rather limited. A totally different issue, signal extrapolation, has also been drawing a great amount of interest recently, largely in the area of spectral estimation. It has been shown that a known portion of a signal can be extrapolated outside of the observation interval if the signal possesses certain property. Many algorithms, both iterative and non-iterative, have been proposed for both continuous and discrete cases. The purpose of this thesis is to use spatial signal extrapolation in digital beamforming to improve the beam pattern without adding more physical sensors. Effectively, the sensors are added synthetically through signal processing. This has the potential to improve performance considerably.