Algorithm of Restore Shape of the Object from

the Reflected UWB Signal

S. L. Chernyshev
Bauman Moscow State Technical University
Moscow, Russia
e-mail: [email protected]

AbstractAlgorithm of reconstructing the shape of the object inhomogeneities which completely determine the shape of the
from the reflected UWB signal based on the mentioned reflections object.
developed. This algorithm uses a discrete model of the object and
allows to determine the wave reflection coefficients of local II. MECHANISM OF REFLECTED WAVE FORMATION
inhomogeneities which completely determine the shape of the An arbitrary configuration of the longitudinal section of the
object. metallic object shown in Fig. 1. Rungs show heterogeneities of
the object, which create obstacles to the propagation of the
Keywordsalgorithm; inhomogeneities; long-range portraits; transverse electromagnetic wave. Each rung creates wave
shape; UWB signals
reflection coefficient S i , where i - the rung number (i=1,,N).
I. INTRODUCTION
UWB signals are highly informative. This feature allows the
detection of subtle objects and their recognition [1,2]. As is
known UWB signals are also used for high-speed wireless data
transmission [3]. For object recognition, you must have a priori
a database of their long-range portraits. These portraits can be x
obtained experimentally by making the models of objects and S1 S
irradiate them in an anechoic chamber. Another way to obtain 2

these portraits is the decision of the electrodynamic problem of Si SN

determining the reflected waves from the object. The solution of
this problem has a great theoretical and computational
difficulties. Simplified method is based on applying the Fig. 1. Object inhomogeneities.
principles of physical optics. In this case, the scalar impulse
g (t ) of the object is determined by the second The transverse wave is incident on the object along an axis x.
characteristic
The length 'x of each link is equal to the discretization step of
derivative of the so-called profile function
A(x) of the object surface of the object, which is determined by the expression
[4,5]: 'x cW / 2 , where W - the duration of the probe UWB pulse.

q d 2 A(t ) When one is filed to input ( a1 (k ) k 0 1 ), reflected signal

g (t )  , t 2 x / c, coincides with pulse response and is a sequence of pulses
 S 2 dt 2   reflected from the local inhomogeneities supplemented pulses
resulting from reflections of the waves between irregularities. If
where c - the speed of light, and coefficient q introduced us you apply the Z-transform, as shown in [6], the image of the
to account for the intensity of reflection from the object. reflected signal b1 ( z ) can be written as
However, this method leads to the approximate shape of the
impulse response, not taking into account the mutual reflection b1 ( z ) G( z ) G0  G1 z 2  G2 z 4  G3 z 6  ...
of electromagnetic waves between the local inhomogeneities of 
the object. This article is devoted to the development of the
algorithm of reconstructing the shape of the object from the where z exp( pW / 2), p - Laplace operator. Coefficients
reflected UWB signal based on the mentioned reflections. This
algorithm uses a discrete model of the object and allows you to Gi are discrete values of the impulse response in the time
determine the wave reflection coefficients of local domain.

978-1-5386-3592-6/17/$31.00 2017 IEEE

30
 The analysis showed that the coefficients Gi of this series This expression is the basis of the developed algorithm. It
must be supplemented by the calculation equations for
depends on the reflection coefficients and are defined by
expressions bn (n  1) and an (n  1) - reflected and incident waves on the
heterogeneity of a number n.
G0 S1 , G1 (1  S12 ) S 2 , G2 (1  S12 )(1  S 22 )S3 
For example, for the second inhomogeneity
(1  S12 ) S1S22 , G3 (1  S12 )(1  S22 )(1  S32 ) S4 
(1  S12 )(1  S22 ) S1S2 S3  (1  S12 )(1  S 22 ) S 2 S32  b2 (1) b1 (2) / 1  S12  a2 (1) a1 (0) 1  S12 
(1  S12 ) S12 S23 , ..., G i ... (3)
i 1 Substituting these expressions in (5), we obtain S 2 ,
The number of terms in these expressions is 2 for an providing the second equation in (3).
arbitrary value of the index i. This feature does not allow to
build the appropriate algorithm using these expressions for Waves falling on other irregularities are determined from the
Gi , and especially to solve the following obtained recurrent equations:
determining the coefficients
S
inverse problem of determining unknown coefficients n from
a n 1 (k ) a n (k  1) 1  S n2  bn 1 (k  2) S n ,

the given values

Gi . bn (k ) a n (k )  bn 1 (k  1) 1  S n2 .
III. CREATION OF ALGORITHM Thus, consideration of the object using the discrete model
allowed to determine the object shape recovery algorithm,
The construction of the algorithm is possible, if we consider because the reflection coefficients are directly related to
the impulse responses of the individual parts of the object long-range portraits.
formed by successive elimination of irregularities, beginning
with the first. IV. SCHEME OF ALGORITHM
The image of reflected wave from the complete object is The expressions obtained above make it possible to build a
given by (2) and from (3) easily to conclude that scheme of algorithm. The simplified scheme is presented in
Fig. 2.
S1 G0 b1 (0).
Fig. 2 conditionally shown cycles by n number of rung and
If we exclude the first heterogeneity, the image of the k number of temporary values of waves. In fact, cycle by k
impulse response G( 2) ( z ) of the remaining part is described must be embedded to cycle by n. Program based on this
algorithm results in a coefficients values to determine object
by a power series of the form (2) in which the coefficients are shape.
calculated as follows:

G20 S2 , G21 (1  S 22 ) S3 , G22 (1  S 22 )(1  S32 ) S 4  b1 (0), b1 (1),...; n k 0

(1  S ) S S , G23
2
2
2
2 3 (1  S )(1  S )(1  S ) S5 
2
2
2
3
2
4

(1  S )(1  S ) S2 S3 S4  (1  S 22 )(1  S32 ) S3 S 42 

2
2
2
3
n : n  1,
(1  S ) S S , ..., G
2
2
2 3
21 3 2i ... (4) k : k 1
The wave reflected from the second inhomogeneity is given by
Calculatio n of a n (k ), bn (k ),
b2 ( z ) G( 2) ( z )a2 ( z ) . a n (n  1), bn (n  1)
Using (3), we can write
b2 (1) S 2 a2 (1) , bn (n  1)
Sn
and we find, that an (n  1)

 S2 b2 (1) / a2 (1)  
Yes S n  S n 1  H
We can show that we can similarly determine the reflection
coefficient for any inhomogeneity: No
End
Sn bn (n  1) / an (n  1)
 
Fig. 2. Simplified sheme of algorithm.

31
 This expression describes a rectangular video pulse. If we
consider a discrete model of a spherical surface, breaking it into 0,2
10 stages and putting, for example, q 0, 2 , the reflection 0,18 S
0,16
coefficients of irregularities are as
0,14
0,12


Sn 0,1; n 1,...,10   0,1
0,08
0,06
If we substitute in the developed algorithm the data
corresponding to the solid line, we fully restore the reflection 0,04
coefficients of given above in (6). If we substitute in the 0,02
developed algorithm the data corresponding to the dotted line, 0
1 6 11 16 n
we can not restore these the reflection coefficients.

g 0,12
Fig. 4. The dependence of the reflection coefficients for the impulse response
0,1
(dotted line in Figure 3).
0,08
0,06 V. CONCLUSIONS
0,04 This algorithm makes it possible to determine the shape of
0,02 k the object from the reflected UWB signal, taking into account
internal reflections between local inhomogeneities, which make
0
the analysis more accurate than the method of physical optics.
-2
-0,02 3 8 13 18

-0,04
-0,06 REFERENCES
[1] Introduction to Ultra-Wideband radar systems. Edited by James D.Taylor,
CRC Press, 1995.
Fig. 3. Normalized pulse response of sphere (solid line - taking into account [2] H.L. Bertoni, L. Carin, and L.B. Felsen (Editors), Ultra-widebandshort
internal reflections, the dotted line - according to the method of physical pulse electromagnetics, Kluwer Publications, New York, 1992.
optics). [3] J. Foerster et al., Ultra-wideband technology for short- or
mediumrangewireless communications, Intel Technology J Q2 (2001).
Fig. 4 shows the relationship obtained in the latter case, the [4] E.M.Kennaugh, D.L.Moffatt, Transient and impuls response
reflection coefficients. It can be seen that the graph differs approximations // Proc. IEEE, 8, 1965, PP. 893-901
significantly from previously obtained data and do not end with [5] D.L.Moffatt, J.D. Young, A.A. Ksinski, H.C. Lin, C.M. Roads,
ten points, continuing much further. It is physically impossible "Trancient Characteristics in Identification and Imaging", Special Issue
and does not correspond to a sphere. This indicates that the on Inverse Methods in Electromagnetics // Antennas and Propagation, v.
shape of the object of the characteristics obtained by the method AP-29, 2, PP. 192-205.
of physical optics can not be restored. [6] S.L.Chernyshev, Analysis and Synthesis of UWB Filters and Shapers in
Time Domain // 2013 IEEE International Conference on Microwave
Technology & Computational Electromagnetics, Proceedings. 6812452.
Qingdao, China, 2013. PP.127-130.

32