An improved CFAR algorithm for target detection

Abstract

The constant false alarm rate (CFAR) technique plays a key role in radar automatic detection process. The cell averaging (CA) CFAR procedure suffers from the masking effect in almost all the multitarget situations. The smallest of cell averaging (SOCA) CFAR has a better performance only when the interfering targets are present in the front or the rear reference window. The ordered statistic (OS) CFAR is rather robust in multitarget situation but at a cost of high computation complexity. However, when a large target continuously occupies several spectral cells, either SOCA-CFAR or OS-CFAR cannot avoid the masking effects. Therefore, an improved CFAR algorithm based on SOCA-CFAR is proposed to tackle these problems. The simulations reveal that the improved CFAR algorithm can alleviate the masking effects at a low computation complexity. A two-dimensional (2D) extension of the proposed CFAR algorithm is also applied for range-doppler-matrix (RDM) and simulation results demonstrate its performance advantages.