Вы находитесь на странице: 1из 4

A Robust CFAR Detector Based on

Ordered
,

Statistic

Wenlin Hu" Yongliang Wang', Shouyong Wang', Qianxue Fang'


1.Wuhan Radar institiute, China 2.National University of Defense Technology, China
Abstract-An improved version of VI-CFAR called OSVICFAR is presented. It is a combination of the ordered statistics (OS) CFAR, the smallest-of OS (SOOS) CFAR and the greatest-of OS (GOOS) CFAR. In homogeneous and clutter edge environments, the detection performance and false-alarm regulation properties for the OSVI-CFAR detector are almost consistent with the VI-CFAR based on mean level (called MLVI-CFAR). But in multiple target situations which the interfering targets are present in both the halves of the reference window, the detection performance for the OSVI-CFAR detector is improved greatly, so it is more robust than the MLVI-CFAR in practice. Finally, some important properties of the OSVICFAR detector are analyzed and discussed.

OSVI-CFAR DETECTOR The reference window of the OSVI-CFAR detector is divided into leading (Window A) and lagging (Window B) halves. The VI-CFAR defines two statistics VI and MR, they are written repeatedly below [3,4].
II.
VI=1+
/1

=1
n i=

(x

x)2 (x)2

(1)
(2)

MR = XA |XB= Yxi | Exi

Key words: CFAR,VI, OSVI, nonhomogeneous environments


I.

INTRODUCTION

xi is the samples of the leading or lagging half of the reference window. x is the arithmetic mean of xi in a half reference window. n =N 2 , N is the size of the whole reference window. &2 is estimated population variance and ,i is the estimated population mean.

Radar fast threshold constant false alarm rate (CFAR) detector keeps CFAR by estimating background and adjusting adaptive detect threshold, such as the cellaveraging CFAR (CA-CFAR) detector and ordered statistic CFAR (OS-CFAR) detector. It is well known that for a homogeneous Rayleigh environment, the CA-CFAR is optimal. Unfortunately, CA-CFAR suffers performance degradation in nonhomogeneous environments [1]. The detection performance of the OS-CFAR is improved in multiple target situations and the loss is also acceptable in homogeneous environments [2]. But the OS-CFAR is unable to prevent excessive false alarm rate at clutter edge[3]. M.E.Smith proposed the VI (variability index) CFAR detector which is a combination of the CA-CFAR, the smallest-of CA (SOCA) CFAR and the greatest-of CA (GOCA) CFAR[3,4], and is called MLVI-CFAR here. The MLVI-CFAR offers a low-loss in homogeneous environments and good false-alarm regulation properties in clutter edge environments. Unfortunately, the performance of the MLVI-CFAR detector may be seriously degraded when the interfering targets are not confined to a single half of the reference window. In this work, the robustness of the MLVI-CFAR detector is improved in multiple target situations by substituting the mean level CFAR for OS-CFAR, and it is called OSVICFAR. At the same time, the detection performance and false-alarm regulation properties of the OSVI-CFAR are almost consistent with the MLVI-CFAR in homogeneous and clutter edge environments.

The VI-CFAR estimates the background of the cell under test through the decision results of the two hypothesis testes below[3,4].

JVI < KV7 |VI > KV7


lMR <K-1
K 1 <MR< KMR or MR > K

Nonvariable > Variable


>
> Same Means > Different Means

(3) (4)

where K, is the threshold of the VI hypothesis testing, and KMR is the threshold of the MR hypothesis testing.
Two crucial environment decided problems are implemented through VI and MR hypothesis testes, i.e. whether the data in each half of reference window is homogeneous, and whether the power levels of the leading and lagging window are identical. The CFAR methods in different situations are listed in TABLE.I.
TABLE I.
Leading Window Vanable?
No No

OSVI-CFAR ADAPTIVE THRESHOLD


Different Mean?
No Yes
T0( TS

Lagging Window Vanable?


No No No

Adaptive Threshold
S=T-Z

Equivalent CFAR Method


OS-CFAR OSGO-CFAR OS-CFAR OS -CFAR

Tos (N,ko )OSB (ko )


NN N -Os (. (y 2' B 2 2

Yes

OS (k) Tos ( ,k) -O


T
02

(N

k)

OS, (k)
()

Yes

Yes

TO0( k)O
2

OSSO-CFAR

0-7803-9582-4/06/$20.00 C2006 IEEE

In TABLE.I., T, (n , n) is the threshold multiplier factor of the OS-CFAR, with a reference window size n, and order value n . OS _(A,B,(k) and OS.i (A,B,(k) are the k-th order statistic of the half reference window with a bigger and smaller sum respectively. OSA(k) and OS (k) are the k-th order statistic of the half reference window A and B respectively, k < N12 .

3.3 X 10-4, the confidence level of MR hypothesis test 8 was set equal to 0.08, and corresponding KVI and KMR are equal to 4.76,1.806 respectively.
A. Performance in Homogeneous Environment Figure. 1 shows the probability of detection (PD) for the OSVI-CFAR, the optimal detector, the CA, GO and OS-CFAR in homogeneous environment. The symbol "OSVI(10)" denotes that the order value of the OSVI-CFAR is equal to 10, the symbol "OS(30)" denotes that the order value of the OS-CFAR is equal to 30. It can be seen from Figure. 1 that all of the CFAR detector perform similarly and exhibit some CFAR loss relative to the optimal detector. When PD=0.5, the losses for the OSVI-CFAR relative to the CA,MLVI,OS-CFAR are about equal to 0.2,0.25, 0.5dB respectively.

The worst clutter edge situation is that the leading window is filled with clutter add noise, the lagging window is filled with noise only, and the cell under test is in clutter region. This situation corresponds to the second row of TABLE.I. The OSVI-CFAR uses GOOS-CFAR to control false alarm peak. If the order value of the OSCFAR is taken on the maximal value., the false-alarm regulation properties for the OS-CFAR are best [5], so the parameters k of the OS-CFAR is set equal to N 2. In homogeneous environment, the goal of the OSVICFAR is to approximate the performance of the CACFAR. This requires that the values of KVI and KMR were selected to ensure low error probability of hypothesis tests expressed by (3), (4). The error probability of hypothesis tests is also called the confidence level. For the VI hypothesis test, it is denoted by ac, and given in (5), for the MR hypothesis test, it is denoted by,8, and given in (6) [4]. a =P[VI > KVI Homogeneous Env.] (5)
=1

0.9
0.8 0.7 0.6

opt CA GO MLVI OSVI(1 0) OS(30)

o 0.5
0.4

0.3 0.2
0.1 _

-P[KLR

< MR <KMR

Homogeneous Env.] (6)

The confidence level a and,8 decrease with increasing KVI and K. . However, the sensitivity for detecting nonhomogeneous environment decreases at the same time. For reducing the false alarm peak that clutter edge locates in N12 (called the primary false alarm peak), a is taken the order of nominal PFA(probability of false alarm) [4]. In an analogous fashion, , should be set small enough. However, another factor which must be considered in setting , is the probability of switching from GOOSCFAR mode to OS-CFAR mode when clutter has filled Window A completely and begins to fill Window B. In this case, there will be an increase in PFA (called the secondary false alarm peak) as the combined Window (AB) is selected more frequently. If , is too small, the sensitivity is decrease in detecting "different mean", this will result in the secondary false alarm peak increasing. Hence, f8 should not be too small. In practice, typical values of f8 will not exceed 0.1 [4].
OSVI-CFAR PERFORMANCE ANALYSIS In our analysis below, the commonly used Rayleigh model for the clutter and target statistics and an even value for the reference window size N will be assumed. The operation of the OSVI-CFAR was simulated for a variety of homogeneous, interfering targets, and clutter edge environments. All performance results of the OSVI-CFAR in this paper were based on 100,000 Monte-Carlo simulation trials. The reference window size N is 36. The confidence level of VI hypothesis test a was set equal to

0N

10

15 SNR /dB

20

25

30

Figure 1. Detection performance of the MLVI,OSVI,CA,GO,OSCFAR in homogeneous environment.

III.

B. Performance in Multiple Target Situations It is assumed that both the interfering target and main target (target in cell under test) agree with Swerling II fluctuation, and their powers are equal to each other (i.e. SNR=INR, INR denotes interfering-to-noise ratio). There is no interfering target in cell under test. For analyzing the different effect of location of interfering target in the reference window on the detection performance of the MLVI and OSVI-CFAR, two kinds of cases are discussed below, i.e. the interfering targets are present in both the halves of the reference window and the interfering targets are confined to a single half of the reference window. Figure.2 and Figure.3 show the PD of the OSVI, CA, GO, SO, OS, and MLVI-CFAR in the case of two interfering targets in the reference window. Figure.2 corresponds to both interfering targets are located in Window A, and Figure.3 corresponds to one interfering target in each half of reference window. It can be seen that the MLVI-CFAR exhibits a low-loss relative to the CACFAR and outperforms the OSVI-CFAR when both interfering targets are located in Window A. When interfering targets are located in both Window A and Window B at the same time, the thresholds of both the SO and MLVI-CFAR are overestimated, so the detection performances degrade quickly while the OS and OSVI-

CFAR perform well. For the OSVI-CFAR, even if the selected reference window contained the interfering targets, the detection performance would not seriously degrade. It can also be seen from Figure.2 and Figure.3 that the increase in using a half of reference Window with bigger mean leads to somewhat CFAR loss compared with the OS-CFAR.
0.9
so MLVI OS(30) OSVI(10) CA GO

0.8 0.7 0.6 2 0.5

04
0.3 0.2 0.1 0
5

Figure 2. PD comparison of CA,GO,SO,OS, MLVI, OSVI-CFAR when two interfering targets in Window A.

0.9

0D8

MLVI OS(30)
E

07

CA GO

OSVI(1 0)

0.3

04~

0.4

031 0 0.2

--

0k

Figure 3. PD comparison of CA,GO,SO,OS, MLVI, OSVI-CFAR when one interfering target in each half Window.

C. Performance in Clutter Edge Environment It is assumed that clutter edges progressed from left to right (Window A to Window B), and clutter-to-noise ratio is CNR.
-2 --

-4

-68>

8-6

m-10
-12
-14
-16 _ -18 _
0

-8t\*i~so
10 15 SNR \ dB 20 25 ~

Figure 4. PFA comparison of OSVI, MLVI, CA, GO, OS, SO-CFAR in clutter edge (CNR= 1odB).

-4

CNR=10dB CNR=20dB

CL

-6_

!o
0

10

15 20 25 Last Cell Containing Clutter

30

35

Figure 5. PFA comparison of OSVI-CFAR indifferent CNRs (k= 16).


30

-------20

10

15 SNR \ dB

25

30

Figure.4 shows the PFA for the OSVI, MLVI,CA,GO, OS, SO-CFAR in the clutter edge environments when the CNR=lOdB. The order values of OSVI and OS-CFAR equal to 16 and 30 respectively. It can be seen that the false-alarm regulation properties of the OSVI-CFAR and MLVI-CFAR are almost consistent, and both outperform the GO-CFAR. The false-alarm regulation properties of the CA-CFAR and OS-CFAR are poor relatively, and SOCFAR is the worst. Figure.5 shows the PFA for the OSVI-CFAR with different CNRs. The order value k is 16 in the simulation. It can be seen that the primary false alarm peak for the OSVI-CFAR when CNR is equal to 20dB is higher than that CNR is equal to 10dB. However, as the clutter edge works its way into more cells of Window B, the bigger CNR results in a more accurate decision for nonhomogeneous environments, so the secondary false alarm peak for the OSVI-CFAR reduces on the contrary.
IV. CONCLUSIONS In this paper, we presented an improved version of VI-CFAR, called OSVI-CFAR detector. The OSVI-CFAR is more robust than the MLVI-CFAR in multiple target situations where the interfering targets are present in both the halves of the reference window, while their performances in homogeneous and clutter edge environments are almost consistent. So there is a better applied potential for the OSVI-CFAR. In multiple target situations, the OSVI-CFAR offers an additional detection loss compared with OS-CFAR, but the loss is endurable. In clutter edge environments, the false-alarm regulation properties of the OSVI-CFAR is even better than the GO-CFAR, its primary false alarm peak is lower than that of the GO-CFAR, and the secondary false alarm peak is lower than or equal to that of the GO-CFAR. The primary false alarm peak increases with increasing CNR, but the secondary false alarm peak decreases by contraries.
REFERENCES

-' F- -:

,,

<? '

~~~~~~~~~~~~~~~OS(30)
GO

oCA

OSVI

MLVI

10

Last Cell Containing Clutter

15

20

25

30

35

[1]

[2] [3]

Mohammad Ali Khalighi, Mohammad Hasan Bastani, "Adaptive CFAR Processor For Nonhomogeneous Environments,"IEEE Trans on AES, 2000,36 (3),pp. 889-897. Rohling H, "Radar CFAR Thresholding in Clutter and Multiple Target Situation," IEEE Transaction on AES, 1983, 19(4),pp. 608621. M E.Smith, P K.Varshney, "VI-CFAR. A Novel CFAR Algorithm Based on Data Variability,"IEEE International Radar conference, Edinburgh,UK: IEEE, 1997,pp. 263-268.

[4] [5]

M E.Smith,P K.Varshney, "Intelligent CFAR Processor Based on Data Variability,"IEEE Trans on AES,2000,36 (3) pp. 837-847. P. P. Gandhi, S. A.Kassam, "Analysis of CFAR Processors in Nonhomogeneous Background," IEEE Trans on AES, 1988,36(3), pp. 427-445.

Вам также может понравиться