Signal Processing 86 (2006) 2115–2122 Performance analysis of some CFAR detectors in homogeneous and non-homogeneous Pearson-distributed clutter Hilal Abdenour Meziani, Faouzi Soltani à De´partement d’Electronique, Faculte´des Sciences de l’Inge´nieur, Universite´de Constantine,Route d’Ain El Bey, Constantine25000, Algeria Received 2 December 2004; received in revised form 10 July 2005; accepted 6 October 2005 Available online 3 April 2006 Abstract In this paper, we analyse the performance of the Greatest Of-Constant False Alarm Rate (GO-CFAR) and the Smallest Of (SO)-CFAR detectors in the presence of clutter environments whose amplitude statistics are modelled by the Pearson V distribution and where the clutter dominates the receiver noise. The performance of these detectors are evaluated both in homogeneous and non-homogeneous clutter. The non-homogeneity is modelled as a step function discontinuity in the reference window. The target in the test cell is assumed to fluctuate according to the Swerling I model. We derive closed form expressions for the probability of false alarm (Pfa) and the probability of detection (Pd) in homogeneous and non- homogeneous Pearson V distributed clutter. The comparison of the two detectors for a non-homogeneous clutter environment showed that the best false alarm rate performance at clutter boundary is obtained for the SO-CFAR detector when the test cell is in low level clutter, while the GO-CFAR detector exhibits better performance when the test cell is from the higher level clutter as expected. r 2006 Elsevier B.V. All rights reserved. Keywords: Cfar detection; Pearson V distribution; Non-homogeneous clutter 1. Introduction In radar systems, the detection procedures involve the comparison of the received signal with a threshold under the constraint of constant false alarm rate. This constraint is impaired by the presence of clutter returns which arise from reflec- tions from the sea, the land, etc. Since the clutter power is unknown, fixed thresholding techniques cannot be applied. One solution to this problem is to set the detection threshold adaptively. For this, the received signal is sampled in range by the range resolution cells. The clutter level in the test cell is estimated by averaging the outputs of the nearby resolution cells. The detection threshold is obtained by scaling the noise level estimate with a constant T to achieve the design probability of false alarm (Pfa). This is the conventional CA-CFAR (Cell- Averaging CFAR) detector proposed by Finn and Johnson [1] as shown in Fig. 1. This detector is optimal when the clutter powers in the reference window are independent, identically distributed and Rayleigh modelled. In some practical applications, the clutter returns may not be uniformly distributed. In the presence of ARTICLE IN PRESS www.elsevier.com/locate/sigpro 0165-1684/$ - see front matter r 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.sigpro.2006.02.036 à Corresponding author. Tel.: +1 213 31 61 42 07; fax: +1 213 31 93 22 47. E-mail addresses: hameziani@yahoo.fr (H.A. Meziani), f.soltani@caramail.com (F. Soltani).