ES2597878B2 - Procedure of transmission and estimation of arrival time in acoustic location systems based on DFT-S-DMT modulation - Google Patents

Abstract

This invention proposes the use of a discrete multitone modulation, spread by discrete Fourier transform (DFT-S-DMT, Discrete Fourier Transform-Spread-Discrete Multitone Modulation) in conjunction with a relative velocity estimation technique transmitter / receiver, for application in acoustic location systems. This allows an efficient adjustment of the available bandwidth for broadcasting as well as compensating the Doppler effect from the estimated speed. With this invention, the problems of multi-carrier modulations such as orthogonal frequency division multiplexing (OFDM, Orthogonal Frequency-Division Multiplexing) are also mitigated, although although they allow adapting the bandwidth of the signals to be transmitted efficiently, they require recovery clock on the receiver and amplify generated signals that suffer from a high peak-to-average power ratio (PAPR).

Description

PROCEDURE FOR TRANSMISSION AND ESTIMATION OF THE ARRIVAL TIME IN SYSTEMS

ACOUSTIC LOCATION BASED ON MODULATION DFT-S-DMT

5

SECTOR OF THE TECHNIQUE

The invention belongs to the technical area of electronic technology and

communications Within this area, and according to its application, it is framed in the

field of acoustic sensory systems.

10

STATE OF THE TECHNIQUE

Current acoustic location systems based on instant detection of

On arrival, they usually employ direct sequence spread spectrum techniques

(DSSS, Direct-Sequence Spread Spectrum), which allow mitigating problems

1S

that presented the first distance measurement systems based on the

envelope detection of the received signal [J. Borenstein and Y. Koren. Obstacle

avoidanee wilh u / lrasonie sensors. IEEE Journal 01 Robotics and Automation. 4 (2): 213

218, 1988]. These problems are high sensitivity to noise, poor accuracy in

estimates of instants of arrival and inability to simultaneously access the

twenty

same channel for more than one transmitter.

The first acoustic location system based on DSSS techniques was proposed in [M.

Hazas and A.Ward. A high performance privacy-orienled location system. In Proc. of the 1st

IEEE Inlernalional Conference on Pervasive Computing and Communications (PerCom

25

2003). pages 216-223. Dahlias (United States). March 2003.1. which was based on the

use of pseudorandom sequences to modulate in displacement modulation

of binary phase (BPSK, Binary Phase Shift-Keying) a sinusoidal carrier and estimate the

Arrival time by correlation. Since then, this type of scheme

transmission / detection in acoustic location systems has become popular in

30

In recent years, there are numerous jobs that use this type of technique.

However, these schemes continue to present certain drawbacks such as the

difficulty limiting emissions within the transducer pass band and the

low robustness against Doppler shift, depending on the type of sequence

pseudorandom used [J. A. Paredes, T. Aguilera, F. J. Alvarez, J. Lozano and J.

Mulberry. Ana / ysis of Doppler Effecl on the Pulse Compression of Oifferent Codes Emitted

by an Ultrasonie LPS, Sensors, vol. 11, no. 11, pp. 10765-10784, November 2011], [D. F.

Albuquerque, J. M. N. Vieira, S. 1. Lopes, e. A. e. Bastos, P. J. S. G. Ferreira, Indoor

S

acoustic simulator for ultrasonic broadband signals with Doppler effecl. Journal of Applied

Acoustics, Volume 97, October 2015, Pages140-151].

This is especially important in location systems in which use is made.

of the smartphone microphone as a receiver, in which the bandwidth

10

Employee is quite limited and the user can be on the move. In these

systems, the modulation used is usually spectrum spread by Chirp (CSS, Chirp

Spread Spectrum), in which a high communication protocol must be used

level to determine the moment of arrival.

fifteen

Finally there is a limited number of jobs in which use is made of

Multi-carrier modulations in acoustic location systems. They use the

frequency hopping spread spectrum technique (FHSS, Frequency-Hoppíng

Spread Spectrum) [M. M. Saad, C. J. Bleakley, S. Dobson. Robusl High-Accuracy

Ultrasonic Range Measurement System. IEEE Trans. Instrumentation and Measurement,

twenty

60 (10): 3334-3341] or orthogonal frequency division multiplexing (OFDM,

Orlhagonal Frequency-Division Multiplexing) with only two carriers where time

of arrival is estimated by phase differences [A. Ens, L M. Reindl, J. Bordoy and J.

Wendeberg Unsynchronizafed Ultrasaund System for TOOA Localization. In Proc. of the

International Conference on Indoor Positioning and Indoor Navigation, pp. 1-9, 2014]. In

25

they assume the static receptor and the adverse effects of the effect are not considered

Doppler in this type of modulations except for [e. Bleakley, M. Taylor. System

and melhod for tracking a range af moving object. WO 2014131894 A2] where it is used

FHSS for object tracking.

30

The use of OFDM in acoustic location systems allows to adjust efficiently

the bandwidth of the modulated signal to the frequency response of the transducer, as well

as an efficient implementation through inverse / direct and fast transformation of

Fourier (IFFT / FFT, Inverse Fast Fourier TransformlFast Fauner Transfarm). However,

OFDM has certain limitations for acoustic location systems. In first

instead, the need to use a quadrature modulator at the output of the IFFT to transmit the complex output in the desired frequency band. It is therefore essential to recover the clock (estimation of the phase and frequency of the local oscillator) to effectively reconstruct the received signal, which introduces some complexity in the receiver. In the case of non-coherent demodulation by correlation with the carriers in phase and quadrature, the estimation of the arrival time is degraded, therefore its advantages are less evident compared to positioning systems based on DSSS.

Second, OFDM can have a high peak-to-average power ratio (PAPR) at the output of the modulator, which reduces the signal to quantization noise ratio (SQNR, Signal-to- Quantization-Noise Ratio) of the digital-analog converter as well as the efficiency of the amplifier used at the input of the transducer. All this entails a lower signal-to-noise ratio in the receiver, resulting in a higher rate of errors in the estimation of the time of arrival, despite the efforts of some investigations (DF Albuquerque, JMN Vieira, S. 1. Lopes, C .

A. C. Bastos, P. J. S. G. Ferreira, OFDM Pulse Des; gn w; th Low PAPR far Ultrasan; c Location and Positioning Systems. In Proc. 2013 International Conference on Indoor Positioning and Indoor Navigation (IPIN 2013)].

The present invention presents a method of transmission and detection of the arrival time for acoustic location systems based on a discrete multitone modulation, spread by discrete Fourier transform (OFT-S-DMT, Oiserete Fourier Transform-Spread -Discrete Multitone Modulation ) non-coherent, which retains the advantages of OFOM while reducing the PAPR of the signal to be transmitted and avoids the use of quadrature modulators. Likewise, the proposed procedure allows the detection of the arrival time to be improved by estimating the speed of the receiver and compensating the Ooppler effect.

EXPLANATION OF THE INVENTION This invention presents a procedure for estimating the instant of arrival of acoustic signals for location systems, based on a non-coherent OFT-S-OMT modulation capable of compensating for the Doppler effect. The contemplated location system is formed by a number N, (N?:. 1) of positioning structures (lPS, Local Positioning Syslem) composed in turn by a set of at least 3 emitting beacons. These transmit simultaneously or multiplexed in time, sequences modulated in OFT-S-DMT to position a mobile element within the coverage area of the LPS by hyperbolic multilateration (without synchronization between beacons and mobile element) or spherical (with synchronism between beacons and mobile element). Also in the location system of the invention, synchronism between beacons of the same LPS is necessary, said synchronism between beacons of different LPS being unnecessary.

Discrete Multitone Modulation (DMT) is defined as:

2M-1

7:, [l J == ¿fq [m] · Jh ",,, / Ud: n = O, ..., 2M -1

m = O

Where fq {m] is the symbol transmitted in the subband my and meets the property of hermetic symmetry, f, [2M-m [= f, [ml "; ', [mI" is the conjugate complex of f, [m ]; M is the total number of channels of identical bandwidth into which the spectrum is divided between O and FJ2; Fs the emitter's sampling frequency and Tq {n] is the DMT modulator output in baseband of 2M samples duration. This type of modulation can be implemented efficiently using a 2M point IFFT.

In this invention Zadoff-Chu sequences are used to estimate the arrival time; however, any other type of sequence with good correlation properties could be used. Zadoff-Chu sequences are defined as:

where Llzc is an odd number equal to the length of the sequence and q is an integer known as root. Therefore, Sq [~ is the Zadoff-Chu sequence generated with the root q and assigned to the DMT modulator input for a given emitter. To accommodate the sequence in the transducer's passband, each bit of the Zadoff-Chu sequence must be entered in the IFFT channels that corresponds to the frequencies of interest, while the rest of the channels must be set to the O value. both the symbol of

DMT modulator input, after applying hermetic symmetry is equal to the following 2M length vector:

I, [m [= [O, .. .. O, s; [O [, ..., s; [L, zc -1], s, [L, zc -1], ..., s , [0 [, 0, ..., 0 [

It is possible to generate a family of Zadoff-Chu sequences with low cross correlations if the difference in roots q for each pair of sequences is a prime number relative to the length L1zc. This can be achieved easily if Llze is an odd prime number, thus allowing to generate up to L1zc -1 sequences for the roots q = {1, "', L'lO -1}.

In order to minimize the PAPR of the DMT modulated signal, this invention employs OFT-S-DMT, which can be expressed as:

2M-l

T, / [n] = 2 ..: Fq [m]. cj27fm ../ 2lo.l; 11 = O,. . . , 2M

7 ,, = 0

Where FII [m] is equal to:

F, [m) =] 0 ..... 0, 5; [0 [, ..., 5; [LlZc -1], 5 ,,] LIZC -1 ['..., 5, [0], 0, ..., O]

and Sq [k), OS k S L, 10 -1, represents the discrete Fourier transform of the sequence Zadoff-Chu Sq [n, defined as:

I: .lzc-t

Sq [k] = L 8'1 [IJ. , :: - j27 <1; 1/1: .1 ""

",or

Where Llze <M. Therefore, each of the beacons of the acoustic location system transmits the Zadoff-Chu sequences in the frequency domain through a DMT modulator, after performing a FFT of Llze points and rearranging the channels at the entrance of the modulator These transformations together constitute DFT-SDMT modulation, which introduces similarly to single carrier DSSS techniques, a process gain equal to MILlze. However, DFT-S-DMT modulation retains the advantages of multi-carrier modulations, being able to adjust the bandwidth of the

signal transmitted to the bandwidth of the transducer, as well as perform the signal equalization in the frequency domain efficiently. Note that in the case where L1zc = M, the OFT-S-DMT modulation is reduced to the DMT modulation.

S 10

Given the ideal periodic self-correlation of the Zadoff-Chu sequences of length Llzc an odd prime number, this type of sequences has a constant power spectral density, thereby reducing the PAPR of the transmitted signal. Therefore, the OFT-S-DMT modulation maintains the advantages of OFDM for acoustic location systems, while minimizing the PAPR (thus minimizing the consumption of the beacons) and allows the detection of the arrival time in a non-coherent way .

fifteen

The estimate at the receiver of the moment of arrival of the sequences modulated in DFT -S-DMT and transmitted simultaneously by the beacons of each of the LPS is calculated following the following metric for each of the transmitted signals:

,

['11' (; - 1

P (d) ~

I: n [H m [. S; [me

",,,,,or

where R [m] is the demodulated received signal, obtained after performing the following operations:

20 1. FFT of a sliding window of 2M points of the received signal, affected by the frequency response of the transducer and the acoustic channel. The beginning of the sliding window applied for the detection of the transmitted sequence is defined by the index d.

2. Selection at the exit of the FFT of the channels used during the transmission.

The moment of arrival of the estimated acoustic signal (ti) can be expressed as:

d = argmax P (d)

30 The metric P (d) can be carried out in an equivalent manner in the time domain by means of a bank of orereladores coupled to each of the signals transmitted by the beacons, Tq [n], and calculating the envelope of the result by the Hilbert transform .

Once the estimation of the time of arrival of each of the signals transmitted by the LPS (d) has been made, it is possible to estimate the linear speed of the receiver due to the expansion / compression of the acoustic signal produced by Doppler effect. Since T5 is the sampling period of the signals transmitted by each of the beacons, Tq (n), the relative transmitter-receiver movement has the effect equivalent to resampling the signal with a period Td expressed as:

Where v / e, referenced as a hereafter, corresponds to the expansion / compression factor of the received signal due to the Doppler effect.

Assuming that the relative emitter-receiver velocity is constant during the time in which the emission of two periods of Tq [n) elapses, the compression factor can be expressed as:

Where Le is the period of emission in samples of the signals transmitted by the beacons and Lr is the duration in samples of a period of the received signal, affected by the Doppler effect. In this invention, the estimate Lr is made from the calculation of the number of samples between correlations of two consecutive periods of the received signal with each Tq [n]. Therefore, for each correlator coupled to each signal Tq (n), a linear velocity is estimated from the factor A. In order to reduce the errors in the estimation, it is possible to calculate the average of the estimated linear velocity for each of the runners coupled to each beacon.

Finally, the estimated Doppler offset can be compensated by resampling the input signal or by filters of minimum variance.

DESCRIPTION OF THE DRAWINGS Figure 1 shows the block diagram of the acoustic location system contemplated in the invention.

Figure 2 shows the block diagram of the processing required for the generation of the signals to be emitted by each beacon as well as the transformations suffered due to the filtering of the transducer (2.5) and the acoustic channel (2.6), together with the algorithm of detection of the instant of arrival

Figure 3 shows the block diagram of the preferred embodiment of the correlation module (2.9) without considering the filtering effect of the transducer or the acoustic channel.

Figure 4 shows a schematic of the procedure for estimating the expansion / compression factor of the received signal due to the Doppler effect.

MODE OF EMBODIMENT As shown in the block diagram of the acoustic location system of the invention, Figure 1, this is formed by a number N of location structures (1 .1), each with coverage areas that can overlap each other (1 .2) (1 .3) (1 .4). In this system, a mobile element (1.5), regardless of others that can be positioned simultaneously, is capable of acquiring the acoustic signals of the LPS, positioning and following navigation routes (1 .6) within the coverage area of the global system

The Zadoff-Chu sequences are transformed to the frequency domain by means of an FFT of L, ze punlos (2.1), subsequently the output of it is reordered so that the property of hermetic symmetry is fulfilled and transmitted at the frequencies of transducer passage (2.2). The sequences are transformed back to the time domain by an IFFT of M points (2.3) and passed through a parallel serial converter (2.4) which introduces zero-padding to separate consecutive emissions. The signal to be emitted by each beacon, Tq [n] is filtered by the transducer (2.5) and affected by the impulse response of the channel (2.6a) and noise (2.6b). In the receiver, the received and digitized signal passes through a serial-parallel converter to perform the FFT of M samples of the input signal; at the exit of the same, the channels in which the Zadoff-Chu sequence was introduced in the transmitter are used to calculate the metric P (d) (2.9). The output of said block is used by a peak detector (2.10) to estimate the moment of arrival at the receiver, d, of the emitted signal Tq (n).

The block diagram of the preferred embodiment of the correlation module (2.9) is shown in Figure 3 without considering the filtering effect of the transducer or the acoustic channel. In this situation, the received signal contains an unknown duration delay.

(3.1) and the transmitted signal Tq [n] with duration 2M samples (3.2). For each Tq [n] and d-value, the scalar product of the transmitted signal Tq [n] is made with a sliding window of 2M samples of the input signal (3.3) starting at the sample d of the received signal. The maximum value of the scalar product will occur when d = d (3.4).

Meanwhile, in Figure 4, a schematic of the procedure for estimating the expansion / compression factor of the received signal due to the Doppler effect is shown. The number of samples between the first transmission of Tq [n] (4.1) and the second (4.2) is known and equal to Le. On reception, the separation in samples between the first period (4.3) and the second (4.4) is calculated by estimating the number of samples between the correlation peaks detected. This separation in static situation of the receiver is equal to Le samples, while being affected by Doppler this separation is equal to Lr. Figure 4 shows the particular case in which Lr is less than Le and therefore the receiver approaches the beacons, causing compression of the received signal (Td <

Ts). There may also be the case of expansion of the received signal (Td> Ts).

In order to minimize the number of operations required, the preferred embodiment of the Fourier transform modules of Figure 2 are based on the FFT or the IFFT. On the other hand, the size M of the IFFT and FFT transforms of the DMT modulator and the sampling frequency of the transmitter, Fs, must be chosen taking into account that said parameters set the bandwidth of each sub-band to an equal value Fsl2M. Therefore, each of the beacons that make up the N LPS must transmit a ZadoffChu sequence through the subbands located within the band of the transducer and generated with roots q, whose difference between those used by pairs of beacons is equal to a prime number relative to Llze; so that their cross correlations are low.

The system can make use of the preferred embodiment of Figure 3 or by correlation in the time domain of the received signal with the emitted signal and the subsequent calculation of its envelope with a Hilbert transform.

The peak detector that follows the roll can detect the maximum value of it

or it may be based on a static or dynamic thresholding. Likewise, the peak detector module can provide absolute moments of its detection, in which case a synchronism signal is necessary, or it can provide the

5 detection moments related to the occurrence of a reference peak. Finally, the process of the invention is preferably implemented in devices.

programmable type of programmable logic block arrays (FPGA, Field Programmable Gale-AlTay) or digital signal processors (DSP, Digi / al Signal Processors).

10 INDUSTRIAL APPLICATION Acoustic location systems for people and / or objects, Ultrasonic obstacle detection systems (air and marine sonar). Systems of non-destructive evaluation of materials, Acoustic communication systems.

Claims (2)

1. A method of transmission and detection of the arrival time for acoustic location systems characterized by:

•

The use of a DFT · S · DMT modulation to be used between a receiver and a transmitter that transmits a Zadoff · Chu sequence modulated in OFT · S · DMT, with ideal properties of periodic self · correlation.

•

Being able to detect the moment of arrival of the emitted signal in a non-coherent way, thus avoiding the use of quadrature modulators and the recovery of phase and frequency of the oscillators in the receiver.

•

Minimize the PAPR of the transmitted signal due to the expansion factor L¡zdM used in the modulator and the constant envelope of the Zadoff-Chu sequences, allowing battery savings in the acoustic signal emission system.

•

Allow to efficiently adjust the bandwidth of the transmitted signal without increasing the duration of the emission and adapt to the frequencies of the acoustic transducer.

•

Allow the estimation of the relative transmitter-receiver speed by calculating the compression factor in samples between two correlation peaks of consecutive emission periods, and compensate for the Doppler effect in the calculation of the moment of arrival of the emitted signal.

2. The method according to claim 1, characterized by:

•

The use of the aforementioned DFT -S-DMT modulation to be used between a receiver in the node to be positioned and at least three emitters that constitute each of the N (N) 1) positioning structures (LPSs) where each emitting beacon transmits a Zadoff-Chu sequence modulated in DFT · S · DMT, simultaneously or multiplexed in time, with ideal auto-correlation properties and cross correlations limited in amplitude, minimizing interference by multiple access; while any mobile receiver can be positioned and navigate within the coverage area of the N LPSs by hyperbolic or spherical multilateration, the synchronism in the emission of the signals between different LPSs not being necessary.

•

Being able to detect the moment of arrival of the signals transmitted by each beacon acustiea in a non-coherent way, thus avoiding the use of quadrature modulators and the recovery of phase and frequency of the oscillators in the receiver.

5 • Minimize the PAPR of the signal transmitted by each beacon due to the LlzcfM expansion factor used in the modulator and the constant envelope of the Zadoff-Chu sequences, allowing battery savings in the acoustic signal emission system. • Allow to efficiently adjust the bandwidth of the transmitted signal 10 without the need to increase the duration of the emission and adapt to the frequencies of the acoustic transducer. • Allow the estimation of the relative emitter-receiver speed by calculating the compression factor in samples between two correlation peaks of consecutive emission periods, and compensate for the Doppler effect in the 15 calculation of the moment of arrival of the signal emitted by each beacon. The linear speed is calculated as the average of the estimates of the relative speed between each beacon and the mobile receiver.