Menzil Sıralı İstatistiklere Dayalı Sabit Yanlış Alarm Oranı İşlemcileri

Abstract

Constant false alarm rate (CFAR) processors are used to detect radar targets in the background in which parameters in the statistical distribution are unknown and may not be stationary. Signal returns from radar targets are usually embedded in thermal noise and clutter (unwanted signal echoes reflected back to the receiver by buildings, clouds, sea etc.). At any given location, since the clutter and thermal noise power cannot be known beforehand, a fixed threshold value determination process to control the false alarm rate cannot be performed by only looking at the radar return from a cell. One of the effective schemes that can be used to overcome the problem of thermal noise and clutter is CFAR processing schemes that can adaptively adjust the threshold based on local estimate of noise power. A CFAR detector has a test cell designated for examination and reference cells surrounding this test cell. The noise power of the background is estimated by processing the values in the reference cells of the CFAR detector. This noise power is used to determine an appropriate threshold value for the background and to determine whether a target exists in the test cell. The background in which the CFAR processor is located is defined in two different ways, either homogeneous or nonhomogeneous. It is assumed that there is only noise in homogeneous backgrounds. In this case, the values in the reference cells of the CFAR processor come only from the distribution of the noise. In nonhomogeneous backgrounds, the values in some reference cells of the CFAR processor come only from the distribution of the noise, while the values in some reference cells come from the distribution of the noise plus the interfering target. In the literature there are CFAR schemes with different processor configurations. Some of these highlights; Cell Averaging (CA) based CFAR, Ordered Statistics (OS) based CFAR, Censored Mean-Level (CML) based CFAR and Adaptive Detection Procedures (ADP) based CFAR processors. In this thesis, a CFAR processor based on Range Ordered Statistics (ROS) which can be presented as an alternative to the CFAR processors in the literature is proposed as a new CFAR processor. In this thesis, firstly, the probability of detection expression of ROS-CFAR processor for a given probability of false alarm in a homogeneous background was analytically derived. In addition, the target detection performance of ROS-CFAR processor was compared with the above-mentioned CFAR processors based on analytic analysis in homogeneous backgrounds and simulation analysis in nonhomogeneous (multiple target) backgrounds.