Вы находитесь на странице: 1из 4

Proceedings of the 7th European Radar Conference

UWB Radar Signal Processing for


Through Wall Tracking of Multiple Moving Targets
Jana Rovna kova, Dusan Kocur
Technical University of Kosice
Letna 9, 041 20 Kosice, Slovak Republic
{Jana.Rovnakova, Dusan.Kocur}@tuke.sk

Abstract Through wall tracking can be very helpful in the


situations where the entering of a room or a building is considered
hazardous and it is desired to inspect its interior from outside
through the walls. In majority of such cases, the tracking of
multiple moving targets is needed. The radar signal processing
for this application has to deal with several supplementary tasks
in comparison with a single target tracking. Their solution is
included in the complex signal processing procedure introduced
in this paper. The experimental results obtained by the real radar
signal processing confirm good performance properties of the
proposed procedure.

I. I NTRODUCTION
It is well-known, that the bandwidth of ultrawideband
(UWB) radars, e.g. up to several GHz, results in their high
spatial resolution, typically a few cm. This unique feature
together with facts that UWB radar devices may be built small
and light weight whereby they employ low-power harmless
electromagnetic waves which in the lower GHz range penetrate
through most common building materials, is the reason, why
UWB radars can be advantageously used for through wall
tracking of moving targets [1]. Such tracking systems can be
then applied in rescue, security or surveillance operations.
UWB radars for through wall detection and tracking can
utilize various stimulus signals. In the literature, impulse [2],
[3], [4], stepped frequency [5], or noise [6] waveforms have
been reported. An alternate development represents a through
wall radar employing a continuous-wave Doppler signals [7].
A common feature of the radars described in [2]-[7] is that
they process raw radar data in such a way as to produce a
radar image of the scanned area, where moving targets are
seen as radar blobs (so-called radar imaging methods). There,
the experimental results for multitarget tracking have been
demonstrated in [3], [4] and [7]. From a signal processing
view, the aproach of [3] is the most simple (subtracting of the
received signals from the empty room response followed by
a back projection algorithm). The others approaches employ
more advanced signal processing methods, such as collecting
of integrated samples over multiple segments for each antenna
pair, assembling of waveform set as functions of time and
time history, motion filtering, range filtering, image generation
and image envelope detection in [4] or combining of Doppler
processing and spatial beamforming with the CLEAN and
RELAX algorithm implementation in [7].
The intention of this paper is to introduce a signal processing procedure for multiple moving targets tracked by UWB

978-2-87487-019-4 2010 EuMA

pseudo-noise radar [8]. As the result of this processing, the


positions of tracked targets moving behind obstacle are not
seen as radar blobs localized in the scanned area, but the
target coordinates are analytically computed by the localization
and tracking methods. For that purpose, the complex radar
signal processing procedure for a single target scenario has
been proposed in [9]. In this paper, the procedure will be
extended for multiple target scenarios. Based on preliminary
experimantal results achieved by processing of the real UWB
signals, the practical analysis of the scenario with two moving
targets will be performed.
II. S IGNAL P ROCESSING P ROCEDURE
The radar signal processing described in this section was
designed for signals provided by the UWB pseudo-noise radar
system using maximum-length-binary-sequence (M-sequence)
as the stimulus signal [10]. The experimental radar system
of that kind equipped with one transmitting (T x) and two
receiving horn antennas (Rx1 , Rx2 ) is shown later in Fig.
1(a). The system clock frequency of the radar device is about
4.5 GHz, which results in the operational bandwidth of about
DC 2.25 GHz. The M-sequence order emitted by radar is 9,
i.e. the impulse response covers 511 samples regularly spread
over 114 ns. This corresponds to an observation window of
114 ns leading to an unambiguous range of about 16 m.
The proposed signal processing procedure for through wall
tracking of multiple moving targets consists of seven phases,
namely background subtraction, weak signal enhancement,
detection, time of arrival estimation, wall effect compensation,
localization and target tracking. In the next paragraphs, a
significance of the particular phases together with the specific
methods providing stable, good and robust performance for
the considered application are outlined.
A. Background Subtraction
Raw radar signals can be interpreted as a set of impulse
responses of surrounding through which the signals emitted
by the radar were propagated. The first task of radar signal
processing is to improve a signal to noise ratio. It is done
by the background subtraction which rejects especially the
stationary and correlated clutter such as antenna coupling,
impedance mismatch response and ambient static clutter, and
allows the response of moving targets to be detected. The
exponential averaging was chosen from variety of background

372

30 September - 1 October 2010, Paris, France

subtraction methods because of its robust performance and low


complexity [11].
B. Weak Signal Enhancement
It follows from the basic radar equation that the level of signal components scattered by a target and received by the radar
depends among others on the distance between transmitting
antenna-target-receiving antenna. Then, the target located close
to radar antenna system is able to produce strong reflections
whereas another target located far from antenna system will
reflect only weak signals. This effect strengthened by the
multipath reflections can result in mimic target signatures or
target disappearing. These theoretical considerations have been
confirmed by our experiences with real radar signal processing.
Whereas in single moving target scenarios, the target has been
detectable almost in all observation time instants, in the case
of multiple target scenarios the reflections only from target
situated the most closely to a receiving antenna can be usually
seen. In order to solve this problem, the enhancement of week
signals scattered by the target has been included into proposed
signal processing procedure. It has been shown in [12], that
the method based on hierarchical searching of maxima of the
impulse response with subtracted background referred to as
the advance normalization, can be used with advantage for
that purpose.
C. Detection
Detection represents a class of methods that on base of a
statistical decision theory determine whether a target is absent
or present in examined radar signals. Between detectors able
to provide good and robust results in the case of multitarget
through wall detection by UWB radar, a constant false alarm
rate (CFAR) detector can be assigned. It is based on NeymannPerson optimum criterion providing the maximum probability
of detection for a given false alarm rate. In considered radar
signal processing, the CFAR detector that assumed a Gaussian
clutter model has been applied [13].
D. Time of Arrival Estimation
Binary data representing the detector output form a noticeable trace of the moving targets. It represents time of arrival
(TOA) of the electromagnetic waves reflected by the target for
the particular instants of the observation time. As the range
resolution of UWB radars is considerably high, with regard
to the physical dimensions of the targets to be detected, the
targets are usually represented by more TOA values in the
detector output. In order to simplify the target localization,
such distributed targets are replaced by simple targets i.e. the
target position in every observation time instant is given by the
only one TOA. This phase of radar signal processing is reffered
to as the TOA estimation. For its realization, a novel algorithm
entitled TOA association has been proposed in [14]. It enables
furthermore to combine TOA estimated from both receiving
antennas to such couples from which only the positions of
the potential true targets can be computed during localization
phase. This part of algorithm represents a data-association
phase and is responsible for deghosting task solution.

E. Wall Effect Compensation


The propagation of electromagnetic waves through a wall
results in the delay time of signals reflected by targets moving
behind the wall. It means, the TOA estimated by the previous
phase of radar signal processing are time shifted because of
the wall presence. Their correction can be achieved by the
subtraction of the mentioned delay time, whereby its estimation is the task of the wall effect compensation phase. The
method referred to as the target trace correction of the 2nd kind
[15] provides promising results in this area. For its utilization
the wall parameters, such as permittivity, permeability and
thickness of the wall, have to be known in advance or they can
be estimated very effectively by the same M-sequence UWB
radar by using the method described in [16].
F. Localization
The aim of the localization phase is to determine target
coordinates in defined coordinate systems whereby the target
locations estimated in consecutive time instants create the
target trajectory. As the input of this radar signal processing
phase, the estimated and corrected TOA couples are used.
Because the considered radar system consists of one transmitting and two receiving antennas, only the non-iterative direct
method of localization can be employed. In that case, the
target coordinates are simply calculated by the tri-lateration
methods as intersections of two ellipses formed on the basis
of the estimated TOA couples and known coordinates of the
transmitting and receiving antennas [17].
G. Tracking
The particular locations of the targets are estimated with
certain random error usually described by its probability
distribution function. Taking into account this model of the
target position estimation, the target trajectory can be further processed by tracking algorithms. They provide a new
estimation of target location based on foregoing positions of
the target. Usually, the tracking results in the target trajectory
error decreasing including trajectory smoothing. In the case
of the multiple targets, track filtering must also deal with
a track maintenance and with the problem of determining
which measurements to associate with which targets being
tracked. From a wide spectrum of tracking algorithms, the
multiple target tracking (MTT) system utilizing linear Kalman
filtering has been chosen as the method enclosing the complex
procedure of the UWB radar signal processing applied for
through wall tracking of the multiple targets [18].
III. E XPERIMENTAL R ESULTS
The performance of the described signal processing procedure is demonstrated by processing of the real radar signals acquired by the experimental M-sequence UWB radar according
to the scenario outlined in Fig. 1. In this scenario, two persons
were moving in a gymnasium behind 24 cm thick wooden
wall covered by tile (Fig. 1(c)). The first person (labelled as
A) was walking from the position P1.1 to the position P1.2
and in the same time the second person (B) was walking from

373

P1.2

11

16

P2.1

P2.2
12

P1.1
7.5
5

16.77
0.24
1.3

Rx1

(a)
Fig. 1.

4.17

Tx

Rx2

0.2

(b)

(c)

(d)

Measuremt scenario: (a) radar device and antennas, (b) building scheme with the antenna layout, (c) building exterior, (d) building interior.

the position P2.1 to the position P2.2 (Fig. 1(b)). The antenna
radar system was located on the bench visible in Fig. 1(c)
according to the layout outlined in the building scheme in
Fig. 1(b), i.e. all antennas were placed along a line with T x
in the middle of Rx1 and Rx2 with distances between adjacent
antennas setting to 1.3 m. There was no separation between
the antennas and the wall.
The raw radar signals corresponding to the described scenario and obtained by the first receiving channel Rx1 are
depicted in Fig. 2(a) in the form of radargram (a two dimensional picture, in which the vertical axis is related to the
time propagation of the impulse response and the horizontal
axis is related to the observation time). In this radargram,
only the cross-talk signal and the reflections of the emitted
electromagnetic wave from the wall can be viewed, forasmuch
as they are very strong in comparison with weak signals
scattered by the moving targets.
This situation is changed after the phase of the background
subtraction when the primary traces of both moving targets
have arisen in the radargrams (Fig. 2(b)). Here, untypical
occurrence in the multitarget scenarios can be observed. It
is related to the strong reflections from target moving further
from the radar system. These reflections are represented by the
steeper trace belonging to person A in Fig. 2(b). This event
can be explained by the fact that the person A was all the
time moving directly in front of the antennas. By contrast,
the trace of the person B is gradually getting weak, what
is due to the flaring angle of the radar antennas. However
globally, the strength of the signals reflected by the both targets
was sufficient and therefore the phase of the weak signal
enhancement has been able to be omitted from the signal
processing procedure for this scenario.
The detector output for the first channel is shown in Fig.
2(c). Here, not only the target traces, but also some harmful
artifacts have been highlighted. The shadowing effect, which
appears as a time-shifted copy of the trace belonging to person
A, is the most massive. Moreover, the part of this copy exceeds
the observation window of 114 ns what results in its presence
at the beginning of the following impulse responses (right
bottom corner of the radargram in Fig. 2(c)). With the high
probability, the fitness centre full of metal equipments located
behind the right wall was the reason of the described effect.
Other visible artifacts are so called cable reflections. They
copy the original target traces, too, but are much more weaker.

In larger distances from the radar system they disappear


completely. The last negative effect is a mutual shadowing of
person A to person B at the beginning of their trajectories. It
results in disappearing of the target trace belonging to person
B approximately in the first 4 seconds of the observation time.
A similarity of radargrams obtained by both receiving
channels can be seen in Fig. 2(d), where the TOA couples
belonging to the same target are depicted by white colour.
Primary estimated and artificially widened TOA from Rx1 and
Rx2 are outlined by red and yellow colour, respectively. Their
conjuction implies that the both receiving antennas captured
the relevant reflections. Not associated TOA are considered to
be the false alarms. In such a way it is avoided to the ghost
generation.
Seeing that the wall is thin and with small relative permittivity, the consequences of the wall effect can be consider negligible in that case. Therefore the wall effect compensation phase
could be omitted from processing, too. The target locations
computed on the basis of estimated and not corrected TOA
couples are depicted in Fig. 2(e). Following artifacts confirmed
by both channels, the redundant amount of estimated positions
appeared in the scanned area. However, the applied MTT
system correctly identified and preserved tracks of two targets.
As can be seen from Fig. 2(f), the estimated tracks correspond
very well with the true target trajectories.
IV. C ONCLUSIONS
In this contribution, the novel complex radar signal processing procedure for through wall tracking of multiple targets
by the M-sequence UWB radar has been introduced. The
proposed method consists of the application of the sequence
of conveniently selected signal processing phases enabling to
compute analytically the target coordinates by the localization
and tracking methods. This approach allows solve also a
number of problems following from multiple target tracking
(e.g. deghosting problem, target positioning improvement by
MTT, etc.) which have not been yet solved satisfactory by
the imaging methods. The preliminary experimental results
obtained by the real radar signal processing confirm good
performance properties of the proposed procedure. Because
the novel radar signal processing procedure employs 1D signal
processing, its computational complexity is less than that of
imaging methods based on 2D signal processing, what is the
additional benefit of the novel approach presented in this paper.

374

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 2. The outputs of the signal processing phases: (a) radargram from Rx1 depictive preprocessed raw radar signals, (b) radargram from Rx1 depictive
signals with subtracted backround, (c) radargram from Rx1 depictive detector outputs, (d) joint radargram from Rx1 and Rx2 depictive estimated TOA
couples (outlined by white colour; primary estimated and artificialy wided TOA from Rx1 and Rx2 are depicted by red and yellow colour, respectively), (e)
scanned area with estimated target locations, (f) scanned area with estimated target tracks.

ACKNOWLEDGMENT
This work was supported by the Slovak Cultural and Educational Grant Agency (KEGA) under contract No. 3/7523/09
and by the Slovak Research and Development Agency under
contract No. LPP-0080-09. This work is also the result of
the project implementation of the Center of Information and
Communication Technologies for Knowledge Systems (project
number: 26220120020) supported by the Research & Development Operational Program funded by the ERDF.
R EFERENCES
[1] I. J. Immoreev, Ultrawideband systems - features and ways of development, in Ultrawideband and Ultrashort Impulse Signals, Second
International Workshop, Sevastopol, Ukraine, Sept. 2004, pp. 3741.
[2] E. Engin, B. Ciftciolu, M. Ozcan, and I. Tekin, High resolution ultrawideband wall penetrating radar, Microwave and Optical Technology
Letters, vol. 49, no. 2, pp. 320325, Feb. 2007.
[3] S. S. Gauthier and W. Chamma, Surveillance through concrete walls,
in Proc. of SPIE - Sensors, and Command, Control, Communications,
and Intelligence (C3I) Technologies for Homeland Security and Homeland Defense III, E. M. Carapezza, Ed., Sept. 2004, pp. 597608.
[4] S. Nag, M. A. Barnes, T. Payment, and G. Holladay, Ultrawideband
through-wall radar for detecting the motion of people in real time, Proc.
SPIE-Radar Sensor Technology and Data Visualization, vol. 4744, pp.
4857, July 2002.
[5] A. R. Hunt, A wideband imaging radar for through-the-wall surveillance, in Proc. SPIESensors, and Command, Control, Communications, and Intelligence Technologies, vol. 5403, pp. 590596, Sept. 2004.
[6] R. Narayanan, Through wall radar imaging using UWB noise waveforms, in IEEE International Conference on Acoustics, Speech and
Signal Processing (ICASSP 2008), Apr. 2008, pp. 51855188.

[7] S. S. Ram and H. Ling, Through-Wall Tracking of Human Movers


Using Joint Doppler and Array Processing, IEEE Geoscience and
Remote Sensing Letters, vol. 5, no. 3, pp. 537541, July 2008.
[8] J. Sachs et al., Detection and Tracking of Moving or Trapped People Hidden by Obstacles using Ultra-Wideband Pseudo-Noise Radar,
in 5th European Radar Conference (EuRAD 2008), Amsterdam, The
Netherlands, Oct. 2008, pp. 408411.

[9] D. Kocur, J. Rovnakova, and M. Svecov


a, Through Wall Tracking of
Moving Targets by M-Sequence UWB Radar, in Towards Intelligent
Engineering and Information Technology, I. J. Rudas, J. Fodor, and
J. Kacprzyk, Eds. Springer, Aug. 2009, pp. 349363.
[10] D. Daniels, Ground Penetrating Radar, 2nd ed. London, U.K.: The
Institution of Electrical Engineers, July 2004, ch. M-sequence radar.
[11] R. Zetik et al., Detection and localization of persons behind obstacles
using M-sequence through-the-wall radar, in Proc. of SPIE - Defense
& Security Symposium, vol. 6201, Orlando, Florida USA, May 2006.
[12] J. Rovnakova, Complete signal processing for through wall target tracking by M-sequence UWB radar system, Ph.D. dissertation, Technical
University of Kosice, Letna 9, 041 20 Kosice, Slovakia, Aug. 2009.
[13] P. K. Dutta, A. K. Arora, and S. B. Bibyk, Towards radar-enabled sensor
networks, in Proc. of the Fifth International Conference on Information
Processing in Sensor Networks, Apr. 2006, pp. 467 474.
[14] J. Rovnakova and D. Kocur, TOA Association for Handheld UWB
Radar, in International Radar Symposium, accepted for publication.
[15] , Compensation of Wall Effect for Through Wall Tracking of
Moving Targets, Radioengineering: Special Issue on Workshop of the
COST Action IC0803, vol. 18, no. 2, pp. 189195, June 2009.
[16] M. Aftanas, J. Sachs, M. Drutarovsky, and D. Kocur, Efficient and Fast
Method of Wall Parameter Estimation by Using UWB Radar System,
Frequenz, vol. 63, no. 11-12, pp. 231235, Nov. 2009.
[17] E. Paolini et al., Localization Capability of Cooperative Anti-Intruder
Radar Systems, EURASIP Journal on Advances in Signal Processing,
vol. 2008, p. 14, March 2008.
[18] J. Khan et al., Multiple Target Tracking System Design for Driver
Assistance Application, in Conference on Design and Architectures for
Signal and Image Processing (DASIP 08), Belgium, Nov. 2008.

375

Вам также может понравиться