LL Logo

FIFTH ANNUAL
ASAP '97 WORKSHOP


____

Statistical Performance
Analysis of Clutter Editing
Adaptive Detection
Algorithm in a
Homogeneous Environment

Christ D. Richmond
MIT Lincoln Laboratory
244 Wood Street
Lexington, MA 02173-9108
tel: (781) 981-5954
email: christ@ll.mit.edu


Abstract
An adaptive detection algorithm known as clutter editing (CLED) was recently proposed as a means of overcoming the high probability of false alarm (PFA) rate associated with the adaptive matched filter (AMF) detector in the presence of undernulled clutter and/or clutter discretes. The AMF is known to be a constant false alarm rate (CFAR) detector under the assumptions of homogeneous clutter with complex Gaussian statistics. In practice, the inhomogeneity of radar clutter (especially in airborne radar systems), and the resulting difficulties in estimating the data covariance matrix significantly mitigate the AMF's inherent CFAR property. The CLED adaptive detection algorithm attempts to remedy this situation. Essentially, this algorithm can be described as a two-stage adaptive sequential detection consisting of a first stage AMF detection followed by a second stage detector known as an adaptive coherence estimator (ACE). The ACE statistic provides a measure of the correlation between the test cell and the assumed target array response vector in a whitened space. Only those range-Doppler test cells that survive both detection thresholdings are declared as target bearing. This present analysis of the CLED algorithm assumes the traditional complex Gaussian statistics in a homogeneous clutter environment. In particular, we provide exact novel closed form expressions for the resulting probability of detection (PD) and PFA of the CLED adaptive detection algorithm and demonstrate that (1) the CLED has a higher or commensurate PD for a given PFA than both the AMF and the ACE, (2) the CLED has a lower or commensurate PFA for a given PD than both the AMF and the ACE, (3) the CLED has an overall performance which is commensurate with the benchmark generalized likelihood ratio test (GLRT), and (4) the CLED is computationally more efficient than a straight GLRT.

____

 


LL Logo Disclaimer

Direct comments and questions to: webmaster@ll.mit.edu

MIT Lincoln Laboratory. All rights reserved.