See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/321066557

Beamforming echo-localization system using multitone excitation signals

Conference Paper · August 2017

DOI: 10.1109/INSCIT.2017.8103522

CITATION READS

1 57

3 authors, including:

Josue Pareja Efrain Zenteno

Universidad Católica "San Pablo" (Perú) Universidad Católica "San Pablo" (Perú)
2 PUBLICATIONS   1 CITATION    33 PUBLICATIONS   89 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Radio Frequency Measurement Technology View project

All content following this page was uploaded by Josue Pareja on 12 January 2018.

The user has requested enhancement of the downloaded file.

 Beamforming Echo-Localization System
Using Multitone Excitation Signals
Josue Pareja-Contreras Manuel Sotomayor-Polar Efrain Zenteno-Bolanos˜
School of Electronics and School of Electronics and School of Electronics and
Communications Engineering Communications Engineering Communications Engineering
Universidad Catolica San pablo Universidad Catolica San pablo Universidad Catolica San pablo
Arequipa, Peru´ Arequipa, Peru´ Arequipa, Peru´
Email: [email protected] Email: [email protected] Email: [email protected]

Abstract—Determining the position of a target is the central

issue in several engineering problems, from computer vision to obtain a constant-directivity spatial response [10]. Another
sonar or radar systems. Echo-localization systems use the key parameter in the design of the sound-based echo-
reflected sound waves to extract information from a specific target
localization system is the excitation signal. In this sense,
taking advantages of the sound or ultrasound frequency range. In
contrast to traditional echo-localization systems, which use single- several candidates are proposed in the literature and the
tone excitation, this paper excite the enviorment using multitone “optimal” signal depends heavily on the desired properties
signals. The utilization of multitone signals gives better time of the resulting signal after applying match filter. That is,
resolution in the ambiguity function for detection in a noisy achieving a 2-D Dirac Delta as ambiguity function [7].
environment. This paper shows an eigth receiver beamforming Basically, the ambiguity function is a measure of the
echo-localization system based on a 32 bit micro controller that is
responsible for analog acquisition and multitone signal excitation. uncertainty of the time-delay and doppler-shift plane.
In contrast to many other works in sound echo-localization
Index Terms—Beamforming, echo-localization, Multitone, which use single tone pulses, this paper explores the use of
Sin-gletone multitone signals as excitations. There are several conse-
quences produced by this design choice: i) The ambiguity
I. INTRODUCTION
function becomes closer to ideal. ii) The excitation signal is
Intelligence and awareness are two desired characteristics in now modulated. This increases the robustness to clutter
today’s modern systems. These characteristics refer to the abil-ity effectively as it is shown with similar modulated excitation
of the systems to detect, predict and take actions according to its signals [11]. iii) The excitation signal is not contained in a
surroundings. That is, basically knowing the position and velocity single frequency point (as a sine wave), then the spatial
of persons or things in a specific scenario. Echo-localization constant-directivity of linear arrays is lost [10]. iv) Compared to
systems are in charge of this task which theory has been largely a single sine wave, the multitone signal requires more
developed in signal processing communities [1], [2] and have hardware or digital processing to be generated. Thus, incurring
found uses in military, biomedical [3], commercial [4], home [5] in extra costs of hardware or computational complexity.
and industrial applications [6].
Depending on the technology used, echo-localization can be
The advantages described before (i and ii) offer
sound-based, radio frequency-based or optically-based. Several
enhanced spatial resolution and clutter immunity,
compromises are made when choosing a specific technology, for
respectively. On the other hand, the drawbacks point out to
instance, accuracy, cost, ease of deployment and others
non-constant directivity (iii) and higher system complexity
(iv). The experiments performed in this paper shown that
[1]. Using mechanical waves (sound-based) is cost-effective in
although the directivity is non-constant, it worsens slightly
many applications as they yield reasonable accuracy and use
and the system retains enough accuracy to be used in
low frequency signals which can be captured and processed
several applications. Furthermore, the extra costs for signal
with cheaper hardware compared to the other solutions. This
generation are mitigated using low-cost modern
paper deals with a sound-based echo-localization system.
instrumentation combined with simpler digital processing.
In this paper, a speaker excites the environment while a
linear array of microphones, captures the echoes produced by The paper layout is as follows: Section II reviews the basic
such excitation (commonly referred to as active sensing [7]). idea behind beamforming, Section III shows the advantages of
This captured echoes are processed rendering spatial using multitoned excitation, Section IV describes the system
information (a process historically referred to as beamforming design considerations including the pulse excitation, receiver
[8]). The relative position among the sensors is a key param- array and data acquisition. Section V describes briefly the dig-
eter in beamforming and gives different array properties. For ital processing performed. The results from the manufactured
instance, it may help to classify multiple echo surfaces [9] or to system are presented and discussed in Section VI. Finally, in
Section VII the conclusions are drawn.
 II. BEAMFORMING

This technique relies on a sensor array configuration, which

sensors are organized in a specific geometry positions
such as lines, planes, cubic or 3D structure. The disposition
of the sensors allows to gather information of the a signal
traveling through space. This spatial information can be
used to detect the direction of arrival of the signal which
resolution is given by the number of sensors (N) in the
array. For instance, this property is exploited in speaker
localization. However it may be used in active scenarios
such as sonar and radar systems. In Fig 1 the radiation
pattern of an 8 and 3 element array is shown

Fig. 2. Ambiguity Function of a Singletone (1) and a Multitone (2) signal

A. Pulse Generation
Fig. 1. Radiation pattern diagram for a linear array with N elements Sine waves were created using digital square waves that
are low pass filtered by the speaker natural frequency
One of the most notorious advantages of the beamforming response saving costs for the analog generation. For a single
technique is the increased gain observed in the processed tone excitation, one timer was used to create a square wave
data. The multiple sensor data can be processed aiming to
signal oscillating at a central frequency fc of 10 KHz. For the
reduce the impact of noise, increase the signal strength and
multi tone excitation, multiple timers where used to generate
focus the receiving beam on a specific direction. The price of
fourteen square wave signals that oscillates at frequencies
these advantages is the fact that multiple sensors require a
from 9300 Hz to 10600 Hz, with a frequency spacing of 100
multiple acquisition system that increases the complexity and
Hz, these square waves are combined using an operational
cost.
amplifier whose output is connected to the speaker. Amplitude
III. SINGLETONE AND MULTITONE SIGNAL and the phase for all 14 tones were considered constants.
The usage of different types of signals defines the resolution Equation (1) shows the mathematical representation of a multi
of detection of an echo-localization system. Although a single tone signal, where f = [9300,9400,9500 ... 10600] Hz.
tone signal is much easier to be generate it shows less time
14
and doppler resolution than a multitone signal, these can be
seen in the ambiguity functions shown in Fig 2. S(t )= ∑ cos(2 πf i t +θi )
i= 1 (1)
The ambiguity function is the representation of the response
to the adapted filter of the excitation signal, it shows its time
and doppler resolution of detection. Since the objectives are Where fi is the i-th frequency (i-th element of f ) and θi
static, the doppler influence is not considered. is the phase of the i-th square wave signal.
The pulses were transmitted with a duration T = 14.3 ms,
IV. SYSTEM DESIGN CONSIDERATIONS and a time of flight TF = 244.1 ms . Fig. 3 shows the time
The excitation signal and analog data acquisition are han- domain excitation wave for the single-tone (a) and multi-
dled by the TM4C129 ENDPT 32 bit MCU from Texas tone (b). As observed, the multitone excitation produces an
Instruments [12]. This device was chosen because of its amplitude and frequency modulated signal (a) in contrast to
performance (120 MHz processor). In addition, direct memory the signal generated by single tone excitation (b).
access channels (DMA) allows the system to achieve sampling
rates of 125 Ksps per channel. In the designed system 8 out of B. Microphone Array
the 20 ADC input channels of the MCU where used. The array is composed by eight electret microphones
The USB 2.0 peripheral is the bridge of communication mounted on a PCB. To maintain a simple design, the distance
between the acquisition (MCU) and the processing stage (PC). between each element is constant (1.5cm) but since a multi-
 Fig. 5. System blocks diagram

Fig. 3. Time domain of the generated pulses (a) Multitone (b) Singletone

tone signal with a bandwidth that goes from 9300 Hz to

10.6 KHz is going to be used, it is necessary to ensure
that the radiation pattern of the array is not going to
have considerable modifications in relation to the single
tone signal that works at fc.

Fig. 6. Mounted system (1) Microphone Array (2) MCU (3) Speaker (4) PC

V. DIGITAL ARRAY PROCESSING

A. Spatial filtering
In order to obtain the information of the signals that
come from ahead with different directions φi is
Fig. 4. Variation of the radiation pattern in a fixed array for different
frequencies necessary to apply a spatial filtering technique. Whence,
Sum Delay Beamforming was selected to be used. As
the name implies, time delays n that are calculated with
Fig. 4 shows the simulated radiation patterns for the
eight-element array with different frequencies. Despite of equation (2) must be applied to the incoming signal xn(t)
from the microphones, where n represents the element
the variations in the radiation pattern, the deviations do number in the array, that in this case goes from 1 to 8.
not cause a great variation of the total pattern due to
frequency change (compared to a central frequency of
10 KHz). This is later corroborated with our experimental n.d. cos(ϕ ' )
τ (ϕ ' )=
results on localization of a specific target. c (2)

The distances between each element are represented by d,

C. Data Acquisition c is the speed of sound through air and φi is the desired angle.
The signals are delayed in the frequency domain as depicted
in equation (3), then they are all summed in order to obtain the
The signals from the microphones are amplified by the
information from only one angle φi. This process must be
AD822 [13] operational amplifiers from Analog Devices. This applied for every single angle, from 0 to 180°.
output is captured by the microcontroller’s analog to digital
converter that has a resolution of 12 bits. In this case the N
1
sampling frequency was set to 125 Ksps with maximum depth Y(ϕ ' , f )= ∑
N n= 1
Xn( f ) e
j 2π τ n (ϕ ' )

resolution. A direct memory access channel (DMA) fills the (3)

USB buffer with the captured data from the eigth ADC inputs.
Once the buffer is completely filled, the data is sended to the The total number of elements in the array are
PC for digital processing. Fig 5 depicts the block diagram of represented by N, Xn(f) represents the signal obtained
the system and Fig 6 shows a photograph of its from the n-th microphone in the frequency domain.
implementation.
B. Echo-localization

A target’s echo is a small and noise contaminated version of

the excitation signal. In order to obtain the precise moment in
which the echo arrives to the microphone array, it is necessary
to perform a cross correlation operation between the data that
was recorded at time T (pulse) and the data from the time of
flight TF. This operation is performed in every angle 0 that was
obtained from the spatial filtering processing. The complete
processing procedure is shown in Fig. 7.
Fig. 8. Signal to noise relation (1) Multi Tone on a reverberant environment
(2) Single Tone on a reverberant environment (3) Multi Tone on a non
reverberant environment (4) Single Tone on a non reverberant environment

A. Target localization
The tests were conducted using a target at 1.85 meters in
front of the array in a reverberant environment, and at 2.5
meters in a direction of 69 ° in a non reverberant environment.
In a reverberant environment the relative position of
the target was found at 1.8632 meters in a direction of
91° using a multitone pulse, see (1) in Fig 9 ,
nevertheless the position found using a single tone was
3.1954 meters in a direction of 90°, this error is due to
the massive clutter and noise, see (2) in Fig 9.
In a non reverberant environment the target was found at
Fig. 7. Procesing Block Diagram (1) Spatial Filtering (2) Inverse Fast 2.4957 meters in a direction of 69° using a multitone pulse,
Fourier Transform (3) Cross-correlation (4) Result visualization
see (3) in Fig 9, while the position using a single tone was
found at 2.4984 meters in a direction of 68 °, (4) in Fig 9 .

VI. EXPERIMENTATION AND RESULTS

To analyze the behavior of the system the tests were

made in a reverberant and non reverberant environment.
Multitone and singletone excitation signals where emitted,
the T F in each experiment was contaminated with additive
white gaussian noise at different levels of SNR.
The energy around the known position of the target (Ps),
was taken, the rest is considered the energy of the clutter and Fig. 9. (1-3)Multi Tone and (2-4)Single Tone echo localization
the added noise (Pc), a relation between this variables (Ps/Pc)
was taken with different the levels of contamination as
VII. CONCLUSIONS
mentioned before, in Fig 8 the x axis represents the SNR of
the contaminated data of the T F and the y axis represents A beamforming echo-localization system using multitone
(Ps/Pc) in a logarithmical scale for a better appreciation. excitation was designed and implemented. A target was
Although the influence of noise is not representative, the placed in different directions in front of the microphone array in
enhancement of clutter immunity can be seen in Fig 8 order to analyze the effects of clutter and noise on detection
comparing (1) and (2), where the improvement is of magnitude using single and multitone exitation for comparison purposes.
order of 100 approximately, and comparing (3) and (4) , where The detection and noise and clutter immunity where enhanced
improvement is of order one approximately. with the multitone excitation that was generated with the same
processor that acquire the analog data from the microphones.
With the accuracy obtained in this experiments, this system
can be used for different applications in echo-localization.
REFERENCES
[1] A. Ens, L. M. Reindl, J. Bordoy, J. Wendeberg, and C.
Schindelhauer, “Unsynchronized ultrasound system for tdoa
localization,” in Int. Conf. on Indoor Positioning and Indoor
Navigation (IPIN),. IEEE, 2014, pp. 601–610.
[2] D. Caicedo and A. Pandharipande, “Ultrasonic arrays for
localized presence sensing,” IEEE Sensors journal, vol. 12, no. 5,
pp. 849–8558, 2012.
[3] M. Fink, G. Montaldo, and M. Tanter, “Time-reversal acoustics in
biomedical engineering,” Annual review of biomedical
engineering, vol. 5, no. 1, pp. 465–497, 2003.
[4] P. Peltola and T. Moore, “Towards seamless navigation,” in Multi-
Technology Positioning. Springer, 2017, pp. 125–147.
[5] X. Bian, G. D. Abowd, and J. M. Rehg, “Using sound source
local-ization in a home environment,” in Int. Conf. on Pervasive
Computing. Springer, 2005, pp. 19–36.
[6] R. F. Armansyah, F. D. Ikram, S. S. Nolika, and T. Adiono,
“Efficient sound-source localization system using low cost tdoa
computation,” in Electronics and Smart Devices (ISESD),
International Symposium on. IEEE, 2016, pp. 315–319.
[7] H. He, J. Li, and P. Stoica, Waveform design for active sensing systems: a
computational approach. Cambridge University Press, 2012.
[8] B. D. Van Veen and K. M. Buckley, “Beamforming: A versatile approach to
spatial filtering,” IEEE assp magazine, vol. 5, no. 2, pp. 4–24, 1988.
[9] L. Kleeman and R. Kuc, “An optimal sonar array for target
localiza-tion and classification,” in Proc. of IEEE Int. Conf. on
Robotics and Automation. IEEE, 1994, pp. 3130–3135.
[10] M. Brandstein and D. Ward, Microphone arrays: signal processing
techniques and applications. Springer Science & Business Media, 2013.
[11] P. H. Leong, T. D. Abhayapala, and T. A. Lamahewa, “Multiple target
localization using wideband echo chirp signals,” IEEE Trans on Signal
Processing, vol. 61, no. 16, pp. 4077–4089, 2013.
[12] http://www.ti.com/lit/ds/symlink/tm4c129encpdt.pdf.
[13] http://www.analog.com/media/en/technical-
documentation/data-sheets/AD822.pdf.

View publication stats