An Adaptive Diagonal Loading Covariance Matrix Estimator in Spatially
Heterogeneous Sea Clutter
Yanling Shi, Xiaoyan Xie
College of Telecommunications & Information Engineering
Nanjing University of Posts and Telecommunications
Nanjing 210003, China
Email:ylshi@njupt.edu.cn m18625156782@163.com
Keywords: diagonal loading; covariance matrix estimator; heterogeneous sea clutter; radar target
detection
Abstract. The most typical method to estimate the covariance matrix of sea clutter by secondary
samples is the sample covariance matrix (SCM), implying secondary samples with the same
statistical property. The hypothesis established or not depends on whether or not sea clutter is
spatially homogeneous. In order to get rid of the hypothesis on secondary samples, a novel
covariance matrix estimation algorithm based on adaptive diagonal loading technique is presented.
The loading coefficient is a factor to measure the statistical consistency of secondary samples.
Experimental results show that, to detect the distributed targets in real sea clutter by generalized
likelihood ratio detector (GLRT), the proposed estimator is robust to jamming and the performance
improves in different range resolution sea clutter, especially as high as 15dB in 3m range resolution.
Introduction
It is well known that sample covariance matrix (SCM) estimator is the maximum likelihood
estimate of covariance matrix in Gaussian white noise [1]. It is simple in operation and small in
computation. For target detection in sea clutter, SCM estimator works well in the low range
resolution where sea clutter can be modelled as complex Gaussian random process. Secondary
samples are with the same statistical properties, such as share the same covariance matrix. However,
with the improvement of radar range resolution, sea clutter in the range cell is the vector
superposition of only a few scatterers. Sea clutter is spatially heterogeneous [2], which is mainly
reflected in the inconsistency of covariance matrix of sea clutter. For the high range resolution, due
to the statistical inconsistency of secondary samples, SCM estimator has a large error in estimating
covariance matrix, which will lead to a performance loss of detector.
In this letter, an adaptive covariance matrix estimator based on the diagonal loading (DL) is
proposed. It combines SCM with the unit matrix adaptively, where the coefficient of combination
changes with the statistical property of secondary samples adaptively. Experimental results show
that, to detect the distributed target in real sea clutter by generalized likelihood ratio test (GLRT)
detector [3], the proposed estimator is robust to jamming, and its performance has a significant
improvement in high range radar resolution sea clutter, compared with SCM.
Problem Description
Let
-dimensional complex vector
,
,
be the received echoes, sea clutter and Doppler
steering vector,
, where
is the number of range cells that target presences.
,
, are unknown constants relevant to target radar cross section.
, are
secondary samples. Target detection in sea clutter is a conventional binary hypothesis testing,
defined as
2nd International Workshop on Materials Engineering and Computer Sciences (IWMECS 2015)
© 2015. The authors - Published by Atlantis Press