Stein's Method and Its Application in Radar Signal Processing

摘要

In the analysis of radar signal processing, many approximations are made in order to simplify systems analysis. In the context of radar detection, environmental assumptions, such as in the modelling of clutter and noise, are often made. Similar assumptions are employed in the analysis of speckle in radar imaging systems. In addition to this, further statistical approximations of random processes of interest are made. In radar detection theory, an assumption on the distribution of a clutter statistic is often made. An important part of the model validation process is to assess the validity of employed assumptions. The validity of environmental assumptions is often assessed on the basis of field trials. In the case of approximations of random processes of interest, there are a number approaches. This study introduces a general scheme, known as Stein's Method, for assessing distributional approximations of random processes. It is described in general terms, and its application to the Poisson and Gaussian approximation of a given random variable is outlined. A new development of Stein's Method for Exponential approximation is included. The importance of the latter is that it can be used as na approximating distribution for a number of random variables of interest in radar.