Bootstrap Methods And Its Application In Array Signal Processing

The parameter and non-parameter estimate methods are widely used in signal processing, especially in array signal processing and statistic signal processing. Of which, the array signal processing is used in radar, sonar, communication. In this paper, we try to study array signal processing in the view of statistic application.Beam-forming and source number estimate are two main subjects in signal processing. Beam-forming, which maximize the signal while minimize the jammings and noise. Source number estimate is the basis of super resolution processing.There are two types of beam-forming, normal beam-forming and super resolution beam-forming. In this paper, we overview the development of beam-forming, introduce the normal beam-forming, discuss the rule of beam-forming, and finally describe the most usually used algrithms.As a statistic method, Bootstrap has advantages when the sample size is small or the distribution of the sample is unknown, and is being used more and more in signal processing, finance, statistic analysis, and so on.In this paper, we also introduce the percepts,development, and basic methods of bootstrap, and application and shortcomings as well. Meanwhile, Jackknife method is discussed as well.At the end of the paper, Bootstrap is used in estimating source numbers, the results are compared with classic methods such as MDL, AIC. Simulation results confirms the advantages, especially when sample size is small or the background is non-Guassian. Based on this, a new method which firstly segment the data, then use the bootstrap method is proposed, simulation confirms the validity of the method.