Abstract

Abstract : Abstract report addresses the statistical analysis of the non-homogeneity detector (NHD) for non-Guassian interference scenarios. An important issue in STAP is that of homogeneity of training data. Non-homogeneity of the training data has a deleterious effect on STAP performance in that undernulled clutter significantly degrades detection and false alarm characteristics. Previous work in this area has proposed the use of non-homogeneity detector based on a generalized inner product (GIP). The unsuitability of the GIP based test for non-Guassian interference scenarios is noted. We present a new non-homogeneity detector for non-Guassian interference scenarios which can be modeled by a spherically invariant random process (SIRP). Our work includes a statistical analysis of the NHD for non-Guassian interference taking into account the fact that finite sample support is used for covariance estimation. In particular, exact theoretical expressions for the NHD test statistic PDF and the mean of a related test statistic are derived. We also note that the related test statistic admits a remarkably simple stochastic representation as a ratio of an F-distributed random variable and a beta-distributed loss factor. Based on this development, a formal goodness-of-fit test is presented. Performance analysis is carried out using simulated and measured data from the MCARM Program.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call