Вы находитесь на странице: 1из 5

RESOLUTION OF RANGE AND VELOCITY AMBIGUITY FOR A MEDIUM

PULSE DOPPLER RADAR


Wen Lei, Beijing Institute of Technology, Beijing, I?R.China
Teng Long, Beijing Institute of Technology, Beijing, I? R.China
Yueqiu Han, Beijing Institute of Technology, Beijing, I? R.China
Abstract
In medium pulse repetition frequency (MPRF)
radars, ambiguities exist both in range and Doppler
measurements. Some efficient techniques have been
established to resolve the range and velocity ambiguity
of the target using multiple PRFs.
In this paper, a simple algorithm is proposed to
resolve both range and velocity ambiguity based on residue
arithmetic. This algorithm makes out the unambiguous
result by using a residue look-up table. The particular
problem about filter bandwidth unitary in frequency is
settled and assessed with another algorithm. An example
for the generation of the residue look-up table is presented
to resolve range ambiguity.

1. Introduction
In order to get unambiguous range and velocity value
simultaneously, MPRF waveform is chosen for modem
airbome radar. As MPRF waveform produces ambiguous
measurements for both range and velocity, a common
technique is to use multiple PRFs to settle this problem,
which needs an algorithm to resolve the ambiguity of
range and frequency in multiple PRFs.
For a given set of PRFs, Chinese Remainder Theorem
has been established to resolve ambiguity [1-31. However,
when the measurement error exists, the result error is
usually very large. Clustering algorithm is also suggested
with the minimum squared error criterion [4,5]. Apparently,
it has good anti-error ability and expensive computational
throughout. An algorithm based on the choice of particular
values for the PRFs is provided for velocity ambiguity
resolution, where a quasi-maximum likelihood criterion is
maximized for ambiguity order estimation [6]. Considering
the blind area both in time and frequency, this algorithm is
so limited by particular PRFs that it is not fit for other
combinations of PRFs to resolve velocity ambiguity.
In this paper, a simple algorithm is proposed which
takes into account presumptive redundancy error to
improve the ability against measurement errors. Its fast
implementation relies on the established look-up table. The
look-up table should be modified when the blind area can't
be overlooked, to assure the completeness of the table.

Furthermore, the performance of the suggested method


about the frequency filter bandwidth unity is compared
with that of the clustering algorithm, each has its
advantages.
The proposed method is so simple that it can be easily
processed in real time processing.

2. Range and Frequency Ambiguity


Resolution
2.1 The Principle of the Residue Look-Up Table
Algorithm
This method makes use of the differences of the
residues on different PRF to resolve ambiguities. First, it
selects the residue of one PRF as the reference. Then it
makes the differences of the residues on other PRFs and
the reference PRF into a look-up table.
Take an example of m different PRFs in time. The
principle of the method is the same on the frequency, as
shown in figure 1 , where
represents different pulse
repetition time (PRT).
When the target. locates on T , its residue q on
different PRF can be described as:
5 = T - N f c = mod(T,c) i = 1,2,..-,m
(1)
Where N i is the ambiguous order, mod(A,B)
represents the module of A to B.
If taking the residue of m B PRF as the reference, the
values stored in the look-up table can be described as:
ei,k = ri - r m
i=1,2,...,m-l
(2)

Where

ei,k represents the differences between the

residues of the range bin on different PRF and the


reference, k represents which group table value it is. In
the table, the sect value Bk, B, = NmTm,is also stored
corresponding to the k" group values. Here only one Bk
corresponds to some PRF residue set.
So each group in the table has m values. The first m-I
values represent the differences between the measured
ambiguous range and the reference. The last value is the
range sect value of the reference.
The process of de-ambiguity can carry through as
follows:

560

0-7803-5776-O/OO/$lO.OO
0 (2000IEEE)

IEEE INTERNATIONAL RADAR CONFERENCE

Stepl: a set of measured ambiguous range values is


obtained.
Step2: the measured ambiguous range value of the
mth PRF is taken as the reference.
Step3: the real range sect value is derived from the
matched values in the look-up table, according to the
differences between the measured ambiguous range of
other PRF and the reference. The range sect value is
obtained as N,T, of the mth PRF.
Step4: the result will be
T = N,T, + r,,,
(3)
If error exists in the measured ambiguous range values,
the search will leave some room for admitting the error.
The difference e,' between measured ambiguous range on
other PRF and reference PRF can be described as:
ei'=r, +Mi-r, -M,
(4)
Where M i presents the measured error on different PRF,

dashed-dotted position. So T locates at the current


repetition of TI, while at the former repetition of T,, Under
this condition, the result can't be found in the look-up table.
While modified look-up table algorithm takes into account
this status, joining the difference of the residues, which has
redundant error, and its corresponding range sect value into
the look-up table.
2.3 Examplefor application
Example 1: three PRFs are adopted to resolve range
ambiguity.
LetT, =103,T2 =119,T3 =137,R,
=1500(the

width of the range bin is taken as 1 IJ s, the maximum


detection range R,, is 225km), [hi= N I 1 .

Take T, = 137 as the reference PRT, B is the range


sect value of the reference PRT, A (T,

6 exists between

ei,, and ei'

T3) represent the

differences between the residue on the i" PRT and the


residue on the third PRT.
Then the values in the look-up table are obtained as
table 1.
Let the true range of the target be R = 574, then the
residues
on
different
PRT
should
be
r, =59,r2=98,r, = 2 6 .
If the measured ranges on different PRT are

r, + Mi represents the measured ambiguous range value.


So the process of resolving the range ambiguity is to
find ei,, which meet the equation ( 5 ) and (6), and some
error redundancy

r, =60,r, =99,r3 =25,

then

Sl=r, -r3 = 3 5

According to the e,+ and its corresponding range

6 2 = r, -r, = 74. Let the redundant error be 2. From the

sect value B , , the resolved ambiguous range is

table, the 14Ih group values (33,72,548) meet the


requirement of inequality ( 5 ) and equation (6), and the
corresponding range sect value B=548. The real range

B, + r , + M , .

value can be got from R' = B +r3 = 573 with range error
of 1.

2.2 The Complete Residue Look-up TableAlgorithm


If the resolving power of radar decreases to some
degree, i.e., the size of range resolution AR is large
enough which meets inequality (7), where t, represents
the width of blind range area. And if these PRFs meet
equality (8):
2 A R > t,
(7)

3. Algorithm Performance Analysis


As discussed in the previous section, the Chinese
Remainder Theorem is easy to use, but anti-error
capability is not so good.
The error performance of residue look-up table
algorithm is identical to clustering algorithm by using only
two PRFs. By using more than three PRFs, clustering
algorithm completely utilizes the relationship between
PRFs, calculating all the mean squared error of possible
range. While residue look-up table algorithm only makes
use of part of the relationship, i.e. only calculating the
squared error between these measured ambiguous range
value and the reference. But because of the number of PRF
utilized increasing, i.e. the number of values which can be
used to search the table are increased, the possibility of

k,T, =kzTz + S , SS26,-

(8)
Where rb represents the width of blind range area,

S
,

represents the maximum measured error.


Then the former look-up table method can't find the
answer in the table, and can7 resolve the ambiguity. Figure
2 shows an example of some range bin T, which will fall
into the blind range area of TI and T, simultaneously. The
dashed line indicates the location of T.
Because the maximum measured range error normally
be f l bin, redundancy 6 can be taken as 2. If there is +1
measured error of T on TI, -1 measured error of T on T,,
and the target can be seen on both PRF, and locate at the

561

iEEE INTERNATIONAL RADAR CONFERENCE

error is largely cut down.


Let the number of PRF be m, and the sum of all the
number of the repetition on m PRF be N. By Clustering
algorithm, which needs at first rank all the possible
solutions, it will cost N computational steps. Then in order
to get the minimum mean squared error solution, it still
needs about N circular computation, each circular
computation includes m-1 addition, 1 division, m
subtraction, m squares. So it needs at least (3m+l)N
computation. The residue look-up table algorithm also
needs about N circular computation, each circulation
includes m-1 compare, but it usually not needs equation (6)
So the total computation is approximately (m-1)N. Then it
seems that the computational throughout of residue lookup table algorithm is less than the clustering algorithm.

F;=F+int[U

(9)

clustering algorithm.
F ~ F=+int[M, + 0.5]+int[Afz + 0.51
(10)
4 represents the rounded-unified frequency by the
residue look-up table algorithm.
F,=F+int[AA +Uz
+0.5]+4f4
(11)
Fp represents the real rounded-unified frequency.

F =int[n,f,]+int[N",N,f,l
(12)
F represents the integer of which is unified
frequency.
Where f , represents repetition frequency, n,
represents the number of filter of the measured ambiguous
frequency, Nu, represents the number of FFT, NI
represents frequency ambiguous order.
The variables in these formations can be described as:
4fi=(n,A lfw)-intInlf, lfm/ll 0<4f,<I (13)

4. Special Issues on Frequency Bandwidth


Unitary

M z = ~ ~ " ~ ~ l f , ~ f ~ , ~ - ~ ~ <~I ~ ~ " , ~ l f

The prerequisite in solving frequency ambiguity is to


settle the fundamental problem of different filter
bandwidth at different PRFs by requantizing the
ambiguous Doppler measurements to a unique reference
PRF. In the process of re-quantizing, it needs to round the
unified frequency value.
On the base of requantization, the error performance
of the residue look-up table algorithm and the clustering
algorithm can be evaluated as follows:
To explain what happens during requantizing, let us
consider the instance of two PRFs.
The errors of the rounded-unified frequency are shown
in figure 3. Where MI and n separately represent the
decimal fraction and the integer of the rounded-unified
measured ambiguous frequency. Af, and NI separately
represent the decimal fraction and the integer of the
rounded-unified measured ambiguous order frequency.
represents the error of the rounded-unified real
frequency, -0.5 < q3
< 0.5. Af4 represents the integer
of Af3, i.e., when -0.5<4f3 < O , 4f4=-I; when
4 f 3 = 0 , 4f4=O;when O < 4 f 3 < 0 . 5 , 4 f 4 = l .
In the process of de-ambiguity, the first step for
clustering algorithm is to add ambiguity order frequencies
with measured ambiguous frequencies, then to unify the
sum. And the first step for residue look-up table algorithm
is to unify ambiguous order frequencies and measured
ambiguous frequencies respectively, then to add these two
values. So the results represented by the two algorithms
can be described as follows, int[] represents the rounding
operation, i.e., int[f(x)l equals to the nearest integer
toward zero of f ( x ) :

(14)
represents reference frequency.
The errors produced by the two algorithms is given as
follows:
fmf

M I S F,-F;IS q4
I
AFzsF,,-F,I =
I int[4fl +4fz+0.5]+4f4-int[Afl +0.5] -int[&f,

(15)

+0.5] I
(16)
A F -MI +Vi+0.5
(17)
Where AF, represents the absolute error produced by
clustering algorithm, Mz represents the absolute error
produced by residue look-up table algorithm.
Then the relation between hFl and M, can be got
as follows:
(1). If MI and U
, meet:
O<Afl <0.5,0<4f, <0.5, and 0 . 5 < A f l + ~ z < l .
Then
int[AF]=int[M, +OS]+int[Af,
0.5]+1=1
(18)
AFzsl+4f4 I
(19)
When Af4 3 0, hFI<AF,. The ability of deambiguity for clustering algorithm is better than that for
residue look-up table algorithm. The principle is shown in
figure 4.
Where the real location of the rounded-unified
frequency should be N , + n + l ; if the residue look-up
table algorithm is used, the rounded-unified ambiguous
order frequency is NI, the rounded-unified measured

w3

ambiguous frequency is n, so the result will be N , + n ,

562
IEEE INTERNATIONAL RADAR CONFERENCE

+4f2 +0.5]

F; represents the rounded-unified frequency by the

there exists error 1. If the clustering algorithm is adopted,


then the result will be consistent with the real result.
When 44=
-1, AFl>Lv;,. The ability of deambiguity for residue look-up table algorithm is better than
that for clustering algorithm. The principle is similar as in
figure 4.

4,and 42
meet:
0.5<4, 4,0.5<Af2 4,and 1 < 4 , + 4 2 < l . 5 .
Then
int[hl;]=int[AA + 0.5]+int[Af2 + 0.5]-1=1
(20)
AE2sL\f,-l I
(21)
When 44< 0, AF,<Al?,. The ability of deambiguity for clustering algorithm is better than that for
residue look-up table algorithm.
When 44=
1, M,>M,. The ability of deambiguity for residue look-up table algorithm is better than
that for clustering algorithm.
(3) 4,
and 4, are under other conditions.
Then
int[AF]=int[Ly; + 0.5]+int[Af2 + O S ]
(22)
A4=IL\f41
(23)
i.e. M,= M,,the ability of de-ambiguity for residue
look-up table algorithm equals to that for clustering
algorithm.
From the upper analysis, under different conditions,
de-ambiguity performance of residue look-up table
algorithm and clustering algorithm has its strong point.

(2)If

Doppler Ambiguities in Medium PRF Radars in


Multiple-Target Environment, signal-proceeding,
Vol. 11, pp.223-226.
4. Trunk, G. and Brockett, S.(1993), Range and Velocity
Ambiguity Resolution. the Record of 1993 IEEE,
International Radar Conference, pp. 146-149.
5. Albano G., Cacopardi S.,Fedele G. (1990), Resolution
of Velocity Ambiguities For MPRF Frequency agile
Radars in Multiple Target Environment, IEEE
International Radar Conference, pp.595-599.
6. FERRARI, A., BERENGUER C., and ALENGRIN, G.
(1997), Doppler Ambiguity Resolution Using Multiple
PRF, IEEE Transactions on Aerospace and Electronic
Systems, Vo1.33, No.3, pp.73 8-75 1.

T1

r1

.!

N,+1
T1

NZ+1
.............
..

..............., . ...............
s

Figure 1.

Residue Look-up Table Method

5. Summary
A new de-ambiguity algorithm is presented here to
resolve both range and fiequency ambiguity by residue
look-up table. This algorithm provides good anti-error
ability by taking into account redundant error. Also, the
performances of computation and frequency unitary are
analyzed and compared of this residue look-up table
method and clustering method. From above argument, the
proposed algorithm possesses good anti-error ability,
simple and convenient computation. It is a fine
compromise scheme to be adopted to resolve both range
and Doppler ambiguities for a MPRF radar.

kz

+t,

T
I

Figure 2.

Lost target of the Look-up Table

Af;

n
*I
Figure 3a. Fraction of the Rounded-Unified Measured
Ambiguous Frequency

References

Afa

1. Hovanessian

S.A. (1982), An algorithm for


calculation of range in a multiple PRF radar, lEEE
Trans. AES-18, No.3, pp.288-296.
2. Ringel M.B., Mooney D.H., Long W.H.(1983), F-16
Pulse Doppler Radar(AN/APG-66) Performance,
IEEE TMS. AES, Vol. 19, NO.1, pp. 147-158.
3. Sirdhar N, REDDY.(1986), Resolution of Range and

Ni

Ni+1

Figure 3b. Fraction of the Rounded-Unified Frequency


Ni

563
IEEE INTERNATIONAL RADAR CONFERENCE

k
:

Ni
Ni+n
N?n+ 1
Figure 3c. Fraction of the Rounded-Unified Measured
Frequency

Ni

Figure 4b. Unified Location of Ambiguous Order


Frequency (not rounded)

Ni
Figure 4a.

Ni+n

clr
n

Ni+n+ 1

n+l

Figure 4c. Unified Location of Measured Ambiguous


Frequency (not rounded)

Unified Location of Real Frequency


(not rounded)

V,=A(T,-T3)

-103

-103

34

-69

-69

68

-35

v,=ACr,-Ts)

-119

18

18

-101

36

36

564

IEEE INTERNATIONAL RADAR CONFERENCE

N,+1

...
...

-1

33

...

-72

-65

72

...

-58

Вам также может понравиться