2000 2"d International Conference on Microwave and Millimeter Wave Technology Proceedings

RANGE PERFORMANCE ANALYSIS IN LINEAR FMCW RADAR

CHEN Weidong, XU Shanjia, WANG Dongjin, LIU F a h Dept. of Electronic Engineering and Information Science, University of Science and Technology of China, Hefei 230027, China

Absfract-From the ambiguity function of the linear FMCW signal, the general effects of the frequency sweep nonlinearity to the range performance are discussed. Based on the criteria of determining the Carson Bandwidth, an approximate relationship among the range resolution, the target range delay and the mean square root of the nonlinear frequency errors is presented at small phase modulation index. Then the problems of the self-clutter interference in this radar are also described, and an estimation expression of the selfclutter to signal ratio is given. Finally, the measured data of nonlinear frequency errors are utilized for the simulation and comparison of the range performance in this radar.

can be expressed as
0 I t 5 T (1) u(t) = exp(j27r(-st2 + j f , ( t ' ) d t ' ) ) 2 0 where s is the slope of the ideal linear sweep,.and f,(t) is
1 1

the nonlinear frequency error fimction. The range ambiguity function of u(t) is defined [2] as
m

X(Z,O) =

[u(t)u'(t+ Z ) d t
-m

(2)

According to the Correlation Theorem, the Fourier transform of Eq. (2) can be obtained by
m

3(X(Z,O)) = 3( Ju(t)u*(t+ Z)dt)

=G

U (f)u*&-Jlr
( f )* 3 ( e
,+,V)dI~ 12n /,(l')dl'

Index Terms--linear FMCW radar, range resolution, frequency sweep nonlinearity, self-clutter interference.

(3 1

= -[U,,

e-ilr

)IW,,( f )* x

' "

j.

>I*

where U(f) =3(u(t))

U,](f= ) 3(e'"' )

I. INTRODUCTION

Eq. (3) offers a description for the spectrum expansion effects produced by the frequency sweep nonlinearity. The effects can expand the mainlobe width of the signal response, and raise the level of the response sidelobes. So the range resolution will be degraded, and the interference of self-clutter becomes more appreciable. Delay resolution constant 123 is often adopted for the description of the spectrum expansion effects as follows
m

s is well known, linear FMCW radar is often used

in such diverse fields as smart ammunition sensors, vehicle collision avoidance and industry control [ 11. Fine

range resolution is important to target-to-clutter contrast by reducing the clutter area and to support target discrimination. Range resolution in linear FMCW radar systems depends on the transmit bandwidth, the overlap between the transmit and receive frequency sweeps, the frequency sweep linearity, and the receiver frequency resolution. Frequency sweep nonlinearity has often been the main limiting factor for this radar range resolution. According to the theory of modulation, this paper presents an analysis of the spectrum expansion effects that result from the frequency sweep nonlinearity in this radar. Several approximate formulae about range resolution and self-clutter to signal ratio are provided for simulation and practical applications.

A, =

) X ( Z , O ) I2 dr
-m

(4)

However, the delay resolution constant only shows the general contribution of the mainlobe and the floor of signal response to the range performance. The relationship among the range resolution, range delay and frequency sweep nonlinearity cannot be indicated directly by Eq. (4).

So Eq. (4) is unsuitable to practical needs.

A. RANGE RESOLUTiON
For homodyne linear FMCW radar, the echo of a point target can be expressed as
e , ( t )= A COS(275f-d + 2i-C

11. RANGE PERFORMANCE ANALYSIS

Including the frequency sweep nonlinearity, the iiormslized complex envelope of the linear FMCW signal 654

j,

I-i

(Wt'+P,,, ) s<tlT

(5)

where: f, = s r , 5 is the round trip propagation time. For short-range linear FMCW radar applications,
T

is

Insert Eq. (8) and Eq. (10) into Eq. (12), Eq. (12) can be rewritten as

often far smaller than the transmit signal period T. If the highest frequency of the main component off,(t) is far lower than the inversion of 5 , Eq. ( 5 ) can be rewritten as e,AO = Acos(2@ + 2nTt(t) + 9) (6) As a periodical function, f,(t) can be expanded with Fourier series

hRlods = R--J8aZ(P)(2n7f_, Af
f,

+ 1) + 1

(13)

So the range resolution is associated with the range

delay and the frequency sweep nonlinearity.

B. INTERFERENCE OF SELF-CLUTTER
Due to the existence of the frequency sweep nonlinearity in linear FMCW radar, the sidelobe level of the radar response will be boosted. For multiple targets, the response floor of a strong target may smear the mainlobe of a weaker target, and the combination of the floors of the multiple targets may even conceal the mainlobe of a stronger target. In the presence of terrestrial or sea clutter, the target response would also be contaminated by the combination of the floors of these clutters. Next the terrestrial clutter is used as an example to analyze the self-clutter interference in linear FMCW radar, while the case of multiple targets can be regarded as a special example of this analysis. Firstly, a single small clutter element will be considered. The echo spectrum of this clutter element will be broadened at the beat frequency position corresponding to a range delay
T.

(7)
So e,F(t) can be regarded as a phase modulation signal with N modulated frequencies ( Q , k=l , 2,. . .,N), and that the phase modulation indexes corresponding to each modulated frequency is P, = 2nrFk . According to the criteria of determining of the Carson Bandwidth [3], when B < l , the -20dB beat frequency bandwidth of a point target echo can be approximately expressed under the condition of small phase modulation indexes (Bk<l) as where

f , , . is the root mean square of the nonlinear

frequency errors, and can be expressed as

Assume that

and

fm

is the equivalent modulated frequency whose

the range delay of all the phase modulation sidebands of the clutter element echo is almost the same as that of the mainlobe of the clutter element echo, and that the offset angular frequency of the phase modulation sideband is Y , then the effective modulation index [SI of the phase

relation with the coherent processing interval of the receiver is as follows

a( B ) is the function of the phase modulation index P = 2mf,,, , which can be chosen between 1 and 2 under

modulation sideband (1 Hz bandwidth) at the offset angular frequency uis as follows.

rr (14) 2 whereO,(y) is the effective modulation index of linear

B<1. td is often the maximum value of the maximum

range delay
rmClx and

8,,(r)= 28,,(r)sin-

the locked time of the closed-loop

linearizer. Thus, T, is also equal to the inversion of the receiver frequency resolution. According to [4], the range resolution including the frequency sweep nonlinearities is given by: A R
f b

FMCW signal at Y . Assume that the receiver frequency resolution is 6, and all the ground returns at the bHz bandwidth have almost the same range delay
T .

Then the peak phase modulation

(12)

index generated by the returns of this so-called clutter group element can be written as
(15) Bc2 (T) = 8: (r)b Furthermore, let the carrier power of the clutter group

The square root term accounts for the convolution between the beat frequency bandwidth B, of a point target and the receiver frequency resolution A f

655

element be C ., Then the phase modulation power at the offset angular frequency element is (16) 4 2 If the target is just at this position, NP becomes the interference power. Suppose that there are N clutter group elements at this targets frequency, the total interference power will be
N

0.005% can be achieved.

of this clutter group

10 5-

N p = c,

[ew3b =C,0,(r)bsin2- r r

2
g
0 2.

0-

-5-10-15 -

N
k=I

(17) 2 Let the echo power of the target be S, then the clutter-tok=l

IC = X N p k =XCgk6J,(rk)bsin 2 rkzk

-20 -251

t !II
2co
400

-30

signal ratio can be obtained with (1 7) as

Boo Tme(us)

loo0

1200

Fig. 1. Measured data of the nonlinear frequency errors This is the self-clutter interference problem in linear FMCW radar. When the frequency sweep nonlinearity exists, this interference becomes more appreciable. Assuming uniformly distributed ground returns in the area of radar illumination, then C ,, C,. For short-range applications sin2( y k r J 2 )
-151

II

=( y k r k / 2 )

2 -251

can be always
-40
-45

i , , , , , . Eq. (18) can be satisfied. And then let Oi(rA) = O derived as

-50
315

320

325

330

335

340

345

Range(m)

Then

Fig.2 The echo spectrum of a point target, R=330m

By using Eq. (20), the relative level of the phase

modulation sideband of the linear FMCW signal can be strictly limited.

- Ltnear Sweep

111. SIMULATION RESULTS

As was mentioned previously, the frequency sweep
nonlinearity is the main factor, which influences the range resolution and produces the self-clutter interference in linear FMCW radar. Fig. 1 shows the measured data of the nonlinear frequency errors. The maximum nonlinear frequency error is 25.63KHz, and the root mean square of the errors is 5.9KHz. So a fine sweep linearity about
-501 900

I
905
910

915

920

925

930

Range(m)

Fig.3 The echo spectrum of a point target, R=915m The simulations of three point targets with different range delays are performed in terms of the practical data in

656

Fig. 1 (excluding the receiver noise). Fig.2, Fig.3 and Fig.4 show the beat frequency spectra of point targets with 300m, 915m and 1215m respectively. The dashed lines in these figure show the cases of the linear frequency sweep. As shown in Table 1, the -20dB bandwidth of the echo spectra will be expanded with increasing ranges. Table 1 Range (m) 330 915 . 1215
~~

in Fig.5, target T2 and target T3 can be clearly resolved for ideal linear sweep (the dashed line). But for the nonlinear sweep (the real line), T3 has been covered by T2. By using the parameters from Table 2, Eq. (19) and Eq. (20) can calculate the allowable maximum level of the phase modulation sideband in linear FMCW signal. Fig.6 shows the phase modulation power spectrum caused by

a(B)

B,,

1.04 1.44 1.86

(kHz) 3.67 5.69 7.7 .

KZOdB (m) 1.38 2.13 2.89

frequency sweep nonlinearity and three maximum level limits corresponding to 400m, 900m and 1200m. In this plot, the levels of some phase modulation components have exceeded the limits at 900m and 1200m. Table 2 400 900 1200

I I

-27 -15 -10

1 1 I

27 15 10

80 450 800

-40

450m -39 IdBc/Hz

I
350

1200

1205

1210

1215
Range(m)

1220

1225

1230

Fig.4 The echo spectrum of a point target, R=1215m

0 50

100

150 200 Frequency(KHz)

250

300

-15}

P II

Fig.6 The phase modulation power spectrum of the linear FMCW signal with the frequency sweep nonlinearity as shown in Fig. 1 , and the maximum limits of the allowable phase modulation level at three different ranges. REFERENCES [l] A. G. Stove, Linear FMCW Radar Techniques, IEE Proceedings F (Radar and Signal Processing), vol. 139, pp. 343-350, Oct. 1992. [2] LIN Maoyong and KE Youan, Theory of Radar Signal. Beijing: National Defense Industry Press, 1984. [3] J. R Carson, Notes on the Theory of Modulation, P.IEE, vol.10, pp.57-64, Feb. 1922. [4] James A. Scheer and James L. Kurtz, Coherent Radar Performance Estimation. Boston, MA: Artech House, 1993. [5] W. P. Robins, Phase Noise in Signal Sources (Theory and Applications). Beijing: Peoples Telecommunications Press, 1982.

1200

1205

1210

1215

1220

1225

Range(m)

Fig.5 The echo spectra of four point targets Fig.5 shows the simulated beat frequency spectrum of four adjacent point targets whose locations are 1210m (Tl), 1211.5m (T2), 1213m (T3) and 1216m (T4) with the relative levels of 0.3, 1, 0.1 and 0.2 respectively. As shown

657