Generalization of Some CFAR Detectors for MIMO Radars Baadeche, Faouzi Soltani, and Amar Mezache Laboratoire Signaux et Systèmes de Communication, Département d’électronique, Université Constantine 1, Constantine 25000, Algeria Email: {baadechemohamed, f.soltan, mezache_a}@yahoo.fr Abstract—In this paper we generalize the GOSCA-CFAR, the OSGO-CFAR and the OSSO-CFAR detectors for the MIMO (Multi Input Multi Output) radars. We derive close- form expressions of the probability of false alarm (Pfa) and the probability of detection (Pd) in homogeneous environment. The comparison of these detectors for a non- homogeneous clutter environment showed that the OSSO- CFAR has better performance when the number of interfering is high. Index Terms— , GOSCA-CFAR, OSGO- CFAR, OSSO-CFAR I. INTRODUCTION MIMO radar is characterized by using multiple antennas to simultaneously transmit diverse (possibly linearly independent) waveforms and by utilizing multiple antennas to receive the reflected signals [1]. This concept has the ability to improve radar performance, in terms of false alarm rate and detection, by exploiting radar cross section (RCS) diversity [2]. Based on antennas spatial diversity, two types of MIMO radars have been broadly discussed in literature; the coherent MIMO radar “co-located” and the statistical one “widely separated”, as illustrated in Fig. 1. One important difference between the two configurations is the signal model: Widely Separated antennas take advantage of the spatial properties of extended targets while the target is modeled as a point with no spatial property for co-located antennas. In other words, Widely Separated antennas see different independent aspects of the target while the co- located antennas see the same target RCS up to some known relative delay due to the geometries of the antennas [3]. In radar automatic detection system, adaptive threshold is a necessary unit to keep the false alarm rate constant under the noise level variation and interferences. Cell-Averaging (CA-CFAR) gives the optimum estimation for independent and identical exponential distributed cell samples in homogenous environments. The presence of interfering targets is a case of the non- homogeneous background which can occur in a real situation. In this situation the performance of the CA- Manuscript received September 5, 2014; revised November 21, 2014. 22 doi: 10.12720/ijsps.4.1.22-26 Mohamed CFAR degrades seriously. In order to improve the performance detection in this situation, several detectors are proposed based on the order statistic technique: the OS-CFAR [4], the GOSCA-CFAR [5], the OSGO-CFAR and the OSSO-CFAR [6]. Figure 1. MIMO radar concept. Target detection in MIMO radars has the interest of several works in the Neyman-Pearson sense [7], [8] or using the Generalized Likelihood Ratio [9]. Recently, Janatian [10] has generalized the CA-CFAR, the SO-CFAR, the OS-CFAR and the ACMLD (Automatic Censored Mean-Level Detector) for the Widely Separated MIMO radars in homogeneous and non-homogeneous clutter (presence of interfering targets). In this paper, we generalize the GOSCA-CFAR, OSGO-CFAR and the OSSO-CFAR for Widely Separated MIMO radars in the Neyman-Pearson sense. Close form expressions for Pfa and Pd of these detectors in a homogeneous background are derived, assuming a white Gaussian noise model. The paper is organized as follow: signal model in MIMO radar is developed in section 2, in section 3 we define mathematical models for MIMO radars of the proposed detectors in a homogeneous background. Results and discussions for the performance of the generalized detectors in homogeneous and in the presence of interfering targets are presented in section 4. Conclusions are drawn in section 5. II. SIGNAL MODEL IN MIMO RADARS We consider a MIMO radar system that has M transmit antennas and N receive antennas with all the antennas International Journal of Signal Processing Systems Vol. 4, No. 1, February 2016 ©2016 Int. J. Sig. Process. Syst. MIMO radars