GB2437619A

GB2437619A - Doppler measuring device and water current meter

Info

Publication number: GB2437619A
Application number: GB0707502A
Authority: GB
Inventors: Satoshi Kawanami; Kazuya Okimoto; Masahiko Mushiake
Original assignee: Furuno Electric Co Ltd
Current assignee: Furuno Electric Co Ltd
Priority date: 2006-04-26
Filing date: 2007-04-18
Publication date: 2007-10-31
Anticipated expiration: 2027-04-18
Also published as: GB0919681D0; GB2437619B; GB0707502D0; JP2007292668A; JP4828295B2

Abstract

A Doppler measuring device for measuring a Doppler shift of acoustic waves emitted underwater performs cross-correlation between a power spectrum of a transmitting waveform and a power spectrum of a received echo signal, and determines the Doppler shift of the acoustic waves from a provisional Doppler shift which is a frequency at which the cross-correlation output is maximized. The Doppler shift is calculated from barycentric frequencies detected in individual barycentric frequency calculation regions containing peaks of the transmitting waveform power spectrum and Doppler-shifted barycentric frequencies detected in individual barycentric frequency calculation regions containing peaks of the echo signal power spectrum.

Description

TITLE OF THE INVENTION

DOPPLER MEASURING DEVICE AND WATER CURRENT METER

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates to a Doppler measuring device for measuring a Doppler shift of ultrasonic waves (acoustic waves) emitted underwater as well as to a water current meter incorporating a Doppler measuring device.

2. Description of the Related Art

Water current meters for measuring velocities of water currents are conventionally used in fishing operation and marine research. A water current meter includes a transducer unit for transmitting and receiving acoustic waves. As an example, the transducer unit mounted on a ship's hull produces acoustic beams which are aimed in three azimuthal directions equally spaced by 120 degrees at a specific tilt angle 0. The water current meter receives echoes of the transmitted acoustic waves from numeroUs midwater scatterers (e.g., plankton) drifting at a set depth layer and determines the ship's speed through the water (through-water speed) relative to a water mass at the set depth layer based on Doppler shifts of the acoustic waves returning from the midwater scatterers. The water current meter also receives echoes from the sea bottom and determines the ship's speed relative to the sea bottom (speed over ground, or SOG) based on Doppler shifts of the bottom echoes. The water current meter further determines water current velocity from a difference between the ship's speed over ground and through-water speed. Expressing the ship's speed over ground (or through-water speed) by V, underwater sound velocity by c, frequency of transmitted acoustic waves by fO and the amount of Doppler shift by fd, the speed over ground (or through-water speed) V can be approximated by equation (1) below since c is sufficiently larger than V: V = fd c/(2f0 cos) (1) It is essential to precisely measure the amount of Doppler shift fd to determine the ship's speed over ground (or through-water speed) with high accuracy.

Japanese Patent No. 2695989 proposes a Doppler shift measurement technique, in which a current measuring apparatus emits acoustic signals including a first pulse and a second pulse delayed by a specific lag time from the first pulse into the water, performs autocorrelation between a first quadrature-phase sample of an echo signal produced by the first pulse and a second quadrature-phase sample of an echo signal produced by the second pulse, and calculates the amount of Doppler shift based on a phase shift obtained by the autocorrelation. Japanese Patent No. ) 2695989 also proposes that the first and second pulses are generated by using coded-pulse broadband acoustic signals.

Japanese Patent No. 3028376 proposes another Doppler shift measurement technique, in which a current measuring apparatus calculates Fourier transform of a data train representing phase differences between zero-crossing signals (binary signals) generated from a received echo signal and an internally generated reference signal and calculates the amount of Doppler shift using a Fourier spectrum obtained from the Fourier transform.

The technique of Japanese Patent No. 2695989 enables high-accuracy Doppler shift measurement when the received echo signal contains little noise, because this technique utilizes information derived from an entire bandwidth of the received echo signal. However, this technique has a problem that measuring accuracy is degraded when the received echo signal contains a large amount of noise.

Noise causing such measuring accuracy degradation includes surface water waves, propeller noise and acoustic signals emitted from surrounding ships. This technique also has a problem that the amount of Doppler shift measured by a broadband signal is not compensated for frequency dependence of the Doppler shift. This is because the autocorrelatjon does not take into consideration the frequency dependence of the Doppler shift in calculation.

On the other hand, the technique of Japanese Patent No. 3028376 has a problem that it is impossible to achieve a high measuring accuracy because this technique analyzes a received echo signal having a single spectral peak and measures the amount of Doppler shift from the single spectral peak only. Also, the current measuring apparatus of this technique repeatedly transmits acoustic waves and determines the amount of Doppler shift by performing statistical operation, such as averaging, on Doppler shift measurements derived from echo signals obtained with successive transmissions in order to increase the measuring accuracy. This approach requires a long time to measure the Doppler shift, resulting in a low throughput.

SUQ(ARY OF THE INVENTION Intended to overcome the aforementioned problems of the prior art, the invention has as an object the provision of a Doppler measuring device which can measure a Doppler shift with high accuracy and high antinoise capability.

According to a first principal aspect of the invention, a Doppler measuring device for measuring a Doppler shift of acoustic waves emitted underwater determines a spectrum of a received echo signal of the acoustic waves, determines a second barycentric frequency calculation region for the spectrum of the received echo signal, the second barycentric frequency calculation region being related to a first barycentric frequency calculation region which is a frequency range containing a peak of a spectrum of a transmitting waveform of the acoustic waves, determines a barycentric frequency of the received echo signal within the second barycentric frequency calculation region, and determines the Doppler shift of the acoustic waves based on a difference between a barycentric frequency of the transmitting waveform within the first barycentric frequency calculation region and the barycentric frequency of the received echo signal within the second barycentric frequency calculation region corresponding to the first barycentric frequency calculation region.

The first barycentric frequency calculation region which is the frequency range containing the peak of the spectrum of the transmitting waveform and the barycentric frequency in the first barycentric frequency calculation region mentioned above may either be determined during a process of measuring the Doppler shift or be determined in advance and stored in a memory. Additionally, each "peak" referred to in the explanation of the present invention does not mean a point of a maximum level but any "crest" in a spectrum or a cross-correlation output pattern regardless of whether single or multiple crests occur therein.

In the Doppler measuring device configured as mentioned above, a peak of the spectrum of the received echo signal corresponding to the peak contained in the first barycentric frequency calculation region is contained in the second barycentric frequency calculation region which is related to the first barycentric frequency calculation region. In addition, both peaks represent information containing features of a transmitted signal and the received echo signal having almost an identical spectral waveform. It is therefore possible to obtain the Doppler shift with an accuracy high enough from a practical point of view based on the difference between the barycentric frequencies determined within the first and second barycentric frequency calculation regions.

In one feature of the invention, the spectrum of the transmitting waveform contains a plurality of peaks, and the Doppler measuring device determines the Doppler shift of the acoustic waves based on the difference between the barycentric frequency in each of a plurality of first barycentric frequency calculation regions and the barycentric frequency in the corresponding one of a.

plurality of second barycentric frequency calculation regions.

If the spectrum of the transmitting waveform contains a plurality of peaks as mentioned above, the spectra of the transmitting waveform and the received echo signal provide a large amount of information usable for calculating the Doppler shift, thus increasing the accuracy of Doppler shift measurement. A plurality of peaks are produced in the spectrum of the transmitting waveform when the transmitted acoustic signal is formed of a train of multiple broadband signals, such as those coded by maximum-length sequence, or N-sequence, Binary Phase Shift Keying (BPSK), for instance, as will be later described with reference to a preferred embodiment of the invention.

In another feature of the invention, the Doppler measuring device converts the difference between the barycentric frequencies in each pair of the first and second barycentric frequency calculation regions into a value corresponding to the amount of Doppler shift of a specific frequency component (e.g., a center frequency of the spectrum of the transmitting waveform) and determines the Doppler shift of the acoustic waves based on the amount of Doppler shift obtained by such conversion.

Since the Doppler shift is corrected for frequency dependence thereof with this arrangement, it is possible to improve the measuring accuracy of the Doppler shift.

In another feature of the invention, the Doppler measuring device determines the Doppler shift of the acoustic waves by averaging the differences between the barycentric frequencies or the amounts of Doppler shift obtained by the aforementioned conversion.

Since the Doppler shift is calculated in accordance with magnitudes of spectral components of the spectrum of the transmitting waveform or the received echo signal and a distribution thereof and, thus, in accordance with certainty of information on characteristic features of the transmitting waveform or the received echo signal, it is possible to determine the Doppler shift with high measuring accuracy. Various methods are available for averaging the differences between the barycentric frequencies or the amounts of Doppler shift. Examples of averaging methods are weighted averaging operation in which an integral value of the spectrum in each of the barycentric frequency calculation regions is used as a weighting factor, weighted averaging operation in which a maximum value of the spectral components of the spectrum in each of the barycentric frequency calculation regions is used as a weighting factor, and simple averaging operation in which the spectral components of the spectrum in each of the barycentric frequency calculation regions are simply averaged without performing any weighting.

In another feature of the invention, the Doppler measuring device performs cross-correlation between the spectra of the transmitting waveform and the received echo signal and determines the frequency range of the second barycentric frequency calculation region based on a frequency at which an cross-correlation output is maximized.

This frequency corresponds to a provisional Doppler shift which will be later described in the preferred embodiment of the invention. The provisional Doppler shift scarcely fluctuates and has a value generally equal to a true value of the Doppler shift. As the Doppler shift frequency is estimated in this fashion, it is possible to measure the Doppler shift with little influence of noise.

In still another feature of the invention, the Doppler measuring device repeatedly calculates the Doppler shift until calculated values thereof converge, in which the frequency range of the second barycentric frequency calculation region is corrected based on the Doppler shift value calculated in a preceding calculation cycle.

In this case, a center frequency of the second barycentric frequency calculation region approaches the frequency of a peak contained in the second barycentric frequency calculation region each time the frequency range of the second barycentric frequency calculation region is corrected. Thus, this approach serves to further improve the measuring accuracy of the Doppler shift.

In yet another feature of the invention, the frequency range of the second barycentric frequency calculation region is narrowed when corrected based on the Doppler shift value calculated in the preceding calculation cycle.

If the barycentric frequency calculation region is narrowed from one calculation cycle to the next in this way, the frequency range within which the barycentric frequency should be calculated is constricted to the vicinity of a spectral peak. This approach also serves to further improve the measuring accuracy of the Doppler shift while ensuring high antinoise performance.

According to a second principal aspect of the invention, a Doppler measuring device for measuring a Doppler shift of acoustic waves emitted underwater performs cross-correlation between a spectrum of a transmitting waveform of the acoustic waves and a spectrum of a received echo signal of the acoustic waves, and determines the Doppler shift of the acoustic waves based on a frequency at which an cross-correlation output is maximized, wherein the frequency producing the maximized cross-correlation output corresponds to the provisional Doppler shift which will be later described with reference to the preferred embodiment of the invention.

The Doppler measuring device configured as mentioned above extracts information on the location of each peak of the received echo signal in a frequency domain from the spectrum of the received echo signal containing nonperiodic noise by using periodicity of spectral peaks of the transmitting waveform and the received echo signal. When the spectra of the transmitting waveform and the received echo signal each have multiple peaks, the individual peaks exhibit a single common Doppler shift and the Doppler measuring device extracts this Doppler shift. The aforementioned cross-correlation operation provides enhanced antinoise capability and, thus, the frequency producing the maximized cross-correlation output scarcely fluctuates and has a value generally equal to a true value of the Doppler shift (or a value obtained by adding a specific constant to the true value of the Doppler shift).

It is therefore possible to configure a Doppler measuring device featuring both high antinoise performance and high measuring accuracy according to the second principal aspect of the invention.

According to a third principal aspect of the invention, a water current meter for determining velocity of a water current from speed over ground and through-water speed of a ship determined by measuring a Doppler shift of an echo signal of acoustic waves emitted underwater includes one of the aforementioned Doppler measuring devices which measures the Doppler shift of the echo signal based on the transmitting waveform and the received echo signal of the acoustic waves, wherein the water current meter determines the speed over ground and the through-water speed of the ship based on the Doppler shift measured by the Doppler measuring device.

The aforementioned third principal aspect of the invention makes it possible to configure a water current meter capable of determining water current velocities with high antinoise performance and high measuring accuracy.

It will be appreciated from the foregoing and following discussion that the present invention provides a Doppler measuring device and a water current meter capable of determining the Doppler shift and water current velocities with high antinoise performance and high measuring accuracy.

These and other objects, features and advantages of the invention will become more apparent upon a reading of the following detailed description in conjunction with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram showing the configuration of a water current meter incorporating a Doppler measuring device according to a preferred embodiment of the invention; FIG. 2 is a block diagram showing the configuration of the Doppler measuring device of the invention; FIG. 3 is a diagram showing a waveform of a transmit signal; FIG. 4 is a diagram showing a power spectrum of the waveform of the transmit signal; FIG. 5 is a diagram showing a power spectrum of a received echo signal; FIG. 6 is a diagram showing the result of cross-correlation between the power spectra of FIGS. 4 and 5; FIGS. 7A, 7B and 7C are diagrams showing how barycentric frequency calculation regions are successively narrowed; and FIGS. BA and 8B are diagrams showing other types broacthand signals usable as the transmit signal.

DETAILED DESCRIPTION OF THE PREFERRED

EMBODIMENT OF THE INVENTION

Now, a specific embodiment of the present invention is described with reference to the accompanying drawings.

FIG. I is a block diagram showing the configuration of a water current meter 1 incorporating a Doppler measuring device 3 according to the embodiment of the invention, FIG. 2 is a block diagram showing the configuration of the Doppler measuring device 3 of the invention, and FIG. 3 is a diagram showing a waveform of a transmit signal.

The water current meter 1 of this embodiment includes three transducer units 11 which are aimed in three azimuthal directions equally spaced by 120 degrees at a specific tilt angle to transmit acoustic waves underwater and receive signals returning from the sea bottom and numerous midwater scatterers. The transducer units 11 are driven by the transmit signal which is generated by a transmitting waveform generator 14, amplified by respective transmitting amplifiers 13 and fed through respective transmit/receive (T/R) switching circuits 12. The transmitting waveform generator 14 also supplies the waveform of the transmit signal (transmitting waveform) to the Doppler measuring device 3, wherein the transmitting waveform is a train of waveform data which is equivalent to what would be obtained by converting the transmit signal from analog to digital form (A/D conversion).

The transmit signal (transmit pulse) shown in FIG. 3 is a train of multiple broadband signals coded by M-sequence BPSK. The transducer units 11 emit these broadband signals in each successive water current measuring cycle. The transmit signal is made up of four identical elements each including seven BPSK-coded subpulses. In FIG. 3, "Ta" designates time duration of each transmit signal, "Tb" designates time duration of each element, and "Tc" designates time duration of each subpulse.

The time duration Ta is approximately 0.7 ms and the subpulse has a carrier frequency of approximately 250 kHz, for example. If a different code pattern (e.g., +1, +1, +1, -1, +1, -1, -1) in the M-sequence is used, an envelop of a.

power spectrum of the transmitting waveform shown in FIG. 4 The signals received by the three transducer units 11 including echo signals are respectively amplified by receiving amplifiers 15 and A/D-converted by A/D converters 16. Sample data trains output from the A/D converters 16 are stored in a buffer memory 17. The Doppler measuring device 3 reads sample data of the received signals including midwater echo signals for one or more set depth layers and bottom echo signals out of the buffer memory 17 and calculates Doppler shifts fd from the sample data. The Doppler measuring device 3 outputs the amounts of these Doppler shifts fd to a water current calculator 18 which calculates speeds over ground and through-water speeds of own ship by substituting the Doppler shifts fd derived from the echo signals received by the three transducer units 11 in the earlier-mentioned equation (1) . Further, the water current calculator 18 calculates current speeds and directions at the set depth layers using the calculated speeds over ground and through-water speeds of own ship. A display unit 19 shows "current vectors't at specified intervals along a plotted course line of own ship, for instance, the current vectors indicating speeds and directions of water currents at plotted positions and depth layers.

The water current meter 1 is further provided with a controller 20 which includes a central processing unit (CPU) 20a, a digital signal processor (DSP) 20b and a memory 20c (having program and data storage sections). The controller 20 controls mathematical operations performed in the water current meter 1 and the working of individual elements thereof in response to operator inputs entered by pressing keys on an operating panel (not shown), for example. Specifically, if the operator specifies a current measuring depth layer by pressing relevant keys. for instance, the specified depth layer is stored in the memory 20c and the controller 20 controls the water current meter 1 to measure the water current at the set depth layer.

Referring now to FIG. 2, the working of the Doppler measuring device 3 is described with particular attention given on how the Doppler measuring device 3 determines the amount of Doppler shift fd occurring in the echo signal.

The Doppler measuring device 3 includes a discrete Fourier transform (DFT) unit 31 which converts a time-domain transmitting waveform into a frequency-domain amplitude spectrum by performing discrete Fourier transform on the transmitting waveform using a fast Fourier transform algorithm. The DFT unit 31 further generates the power spectrum shown in FIG. 4 by squaring the amplitude spectrum.

Generated with a resolution of 10 Hz, this power spectrum Pt[f ii (where fi is each frequency of power spectrum Pt[fi]) has a left-right symmetric pattern with five peaks including a peak 52 (53) and one small peak 54 distributed on each side of a center peak 51. Since the spectrum Pt[f ii is a power spectrum, and not an amplitude spectrum, the peaks of the spectrum Pt[fiJ have sharp-pointed crests.

Also, due to the nature of the discrete Fourier transform and transmitting waveform, components of the power spectrum Pt[fi] are distributed within a range of 2/Tc (FIG. 3), a center frequency f c, or the frequency fc of the peak 51, is equal to the frequency of the subpulse, a frequency difference between two adjacent peaks is 1/Tb and each peak has a width (zero-crossing width) equal to 2/Ta.

Referring now to FIG. 4, a barycentric frequency calculator 32 of the Doppler measuring device 3 determines barycentric frequency calculation regions Wt[k] (k=l,...n) such that the spectral peaks 51-54 lie at the center of each region Wt{k] . Each of the barycentric frequency calculation regions Wt[k] (also written as Wt[l:n]) is defined in terms of a lower limit and an upper limit of frequency, or a lower limit and a frequency range, where the frequency range is made equal to the frequency difference 1/Tb between two adjacent peaks. The barycentric frequency calculator 32 then calculates barycentric frequencies fwt[kl (k=l, .. .n) for the respective barycentric frequency calculation regions Wt[k] by equation (2) below: fwt[k] = E(Pt[fj] fj)JPt(fj] (2) where Pt[fj] is the power spectrum applied as a weight factor for each barycentric frequency calculation region Wt [k].

The time duration Tb of each element of the transmit signal is determined such that the barycentric frequency calculation region Wt[k] has a width which is at least twice as wide as a supposedly maximum Doppler shift that can occur. This is to ensure that the barycentric frequency calculation regions Wt[k] of the transmit signals can be correlated to later-described barycentric frequency calculation regions Wr[k] of the received signals. Since the power spectrum Pt [fi] , barycentric frequency calculation region Wt[k] and barycentric frequency fwt[k] do not necessarily vary from one transmit cycle to the next, previously measured values of these parameters are stored in the memory 20c and the values retrieved from the memory 20c are used in a process of calculating the amount of Doppler shift fd.

Receiving circuitry of the Doppler measuring device 3 is now described. The Doppler measuring device 3 includes a DFT unit 33 which converts a time-domain received signal into a frequency-domain amplitude spectrum by performing discrete Fourier transform on a sample data train containing an echo signal from the set depth layer corresponding to the time duration Ta that is stored in the buffer memory 17 using a fast Fourier transform algorithm.

The DFT unit 33 further generates a power spectrum shown in FIG. 5 by squaring the amplitude spectrum. This power spectrum Pr[fi] (where fi is each frequency of power spectrum Pr[fi]), also generated with a resolution of 10 Hz, has five peaks including a peak 62 (63) corresponding to the peak 52 (53) of the transmitting waveform power spectrum Pt[fi] distributed on each side of a center peak 61. Peaks corresponding to the peaks 54 of FIG. 4 are buried in noise and indiscernible.

It can be seen from FIG. 5 that a center frequency of the power spectrum Pr[f ii, or the frequency of the peak 61, is offset from the center frequency fc of the power spectrum Pt[fi] of the transmitting waveform by as much as a, which represents the amount of Doppler shift. While a frequency difference between two adjacent peaks of the power spectrum Pr[f ii shown in FIG. 5 is approximately equal to the frequency difference 1/Tb between two adjacent peaks of the transmitting waveform power spectrum Pt[f ii shown in FIG. 4, the former is slightly larger or smaller than 1/Tb due to frequency dependence of the Doppler shift.

The power spectrum Pt[f ii obtained at this stage is stored in the memory 20c for use by a barycentric frequency calculator 42 and a weighting factor calculator 46 which will be later discussed.

A provisional Doppler shift detector 34 including a cross-correlator 35 and a peak detector 36 calculates a provisional Doppler shift from the power spectrum Pt[fi] of the transmitting waveform and the power spectrum Pr[f ii of the received signal and outputs the provisional Doppler shift. The cross-correlator 35 performs cross-correlation between the power spectra Pt[fi] and Pr[f ii. Specifically, the cross-correlator 35 calculates the sum of products of the power spectra Pt[fiJ and Pr[f ii while shifting the power spectrum Pt[f ii with respect to the power spectrum Pr[fi] in steps of the aforementioned resolution of 10 Hz and outputs the result of this cross-correlation operation.

FIG. 6 is a diagram showing the result of the cross-correlation operation output from the cross-correlator 35, in which a highest peak 71 at a central point represents a point where the peak 51 of the power spectrum Pt[f ii coincides with the peak 61 of the power spectrum Pr[f ii, the peak 71 being shifted from the center frequency fc of the peak 51 by as much as zfb. Also, a peak 72 represents a point where the peak 51 of the power spectrum Pt[fi] coincides with the peak 62 of the power spectrum Pr[fi], and a peak 73 represents a point where the peak 51 of the power spectrum Pt[fi] coincides with the peak 63 of the power spectrum Pr[fi] . As an alternative, the cross-correlation operation may involve a process of performing a discrete Fourier transform after obtaining the same number of discrete signal samples from both the power spectra Pt[fi] and Pr(fi] by zero padding, for instance, multiplying the samples of one of the power spectra by conjugate complex numbers of the samples of the other power spectrum, and performing an inverse discrete Fourier transform on the result of multiplication.

The peak detector 36 detects the peak 71 having a maximum value from the result of the cross-correlation output from the cross-correlator 35, calculates the value of the aforementioned frequency shift ifb by subtracting the center frequency fc of the peak 51 from the frequency of the peak 71, and outputs the frequency shift Lfb as the provisional Doppler shift. While the cross-correlation operation is performed on a range centered on the center frequency fc of the peak 51 as discussedabove, the cross-correlation operation may be performed on a range centered on 0 Hz, in which case the frequency of the peak 71 itself becomes the provisional Doppler shift. Also, the provisional Doppler shift may be obtained by a method other C) than the aforementioned method. For example, it is possible to obtain the provisional Doppler shift from a barycentric frequency calculated from an cross-correlation output. In this case, the provisional Doppler shift is expressed by equation (3) below: Provisional Doppler shift = ((L[fi:1 fi)/Pt[fi] -fc (3) where L[f ii is the cross-correlation output.

In this invention, the provisional Doppler shift can be calculated by using the frequency of any desired peak at which the cross-correlation output is maximized. This means that it is not necessarily needed to use the frequency of the peak 71 at the central point of FIG. 6 in calculating the provisional Doppler shift. Therefore, the provisional Doppler shift may be calculated based on the frequency of the peak 72 or the peak 73, for instance.

In the above-described cross-correlation operation, the provisional Doppler shift detector 34 extracts information on the location of each peak of the periodic echo signal in the frequency domain from the power spectrum Pr[f ii containing nonperiodic noise by using periodicity of the peaks of the power spectra Pt[fi] and PrEf ii. When the power spectra Pt[fi] and Pr[fi] each have multiple peaks, the individual peaks exhibit a single common Doppler shift and the provisional Doppler shift detector 34 extracts this Doppler shift. Consequently, the cross-correlation operation of the invention provides enhanced antinoise capability, and the provisional Doppler shift obtained by the aforementioned operation scarcely fluctuates and has a value generally equal to a true, or final, value of the Doppler shift. It is therefore possible to use the provisional Doppler shift as the final value of the Doppler shift. Although the frequency dependence of the Doppler shift is not considered in the foregoing discussion, the provisional Doppler shift calculated by the aforementioned operation has a value which is accurate enough from a practical point of view.

While the provisional Doppler shift can be used as the final value of the Doppler shift as mentioned above, the Doppler measuring device 3 of the present embodiment includes a barycentric frequency processor 40 which calculates the amount of Doppler shift fd with greater measuring accuracy by using the provisional Doppler shift as a starting value. Although the starting value of Doppler shift calculation should preferably be as close as possible to the true value of the Doppler shift (e.g., the provisional Doppler shift), it is possible to use a rough value, such as a value of Doppler shift obtained by conventional autocorrelation of the echo signal, or even 0 Hz as the starting value. This is because the amount of Doppler shift fd is calculated with high accuracy through barycentric frequency processing operation. Also, instead of using the frequency shift fb at the maximum value of the cross-correlation output as the starting value of Doppler shift calculation, the Doppler measuring device 3 may simply use a frequency shift of. a frequency at which the power spectrum Pr[f 1] shows a maximum power level. If a rough value greatly deviating from the true value of the Doppler shift is used as the starting value of Doppler shift calculation, however, the barycentric frequency processing operation will require a larger number of repetitive calculation cycles to be performed, and thus a longer period of time, until calculated values of the Doppler shift fd converge, compared to a case where the aforementioned provisional Doppler shift is used as the starting value. It is to be noted that the calculated values of the Doppler shift fd may not converge under conditions where the received echo signal has a poor signal-to-noise ratio (SNR) and the Doppler shift fd calculated by the autocorrelation, for instance, greatly deviates from the true value due to noise.

Now, the barycentric frequency processor 40 is described in detail. As shown in FIG. 2, the barycentric frequency processor 40 includes a barycentric frequency calculation region deteriminer 41, the aforementioned barycentric frequency calculator 42, a Doppler shift corrector 43, a weighted average processor 44,,, a convergence detector 45 and the aforementioned weighting factor calculator 46. Once the aforementioned cross-correlation operation is performed, the barycentric frequency processor 40 performs repetitive calculation cycles of the barycentric frequency processing operation using the aforementioned provisional Doppler shift as the starting value until values of the Doppler shift fd output from the weighted average processor 44 converge. Here, the expression "converge" (or "convergence") refers to a state in which a difference between two values of the Doppler shift fd obtained in preceding and current calculation cycles becomes equal to or smaller than a specified value (including zero value) . The aforementioned barycentric frequency processing operation is a mathematical operation performed by internal elements of the barycentric frequency processor 40 including the barycentric frequency calculation region deteriminer 41. The cross-correlation operation and the barycentric frequency processing operation are advantageous in that the former is less susceptible to noise and the latter offers high measuring accuracy. Thus, it is possible to substantially exclude the influence of noise by initially performing the cross-correlation operation and calculate the amount of Doppler shift by subsequently performing the barycentric frequency processing operation in a range in the vicinity of a spectral peak within a high SNR frequency region. This makes it possible to simultaneously achieve improved antinoise performance and measuring accuracy.

The barycentric frequency calculation region deteriminer 41 determines the barycentric frequency calculation regions Wr[k] (k=l,...n) (FIG. 5) which are calculation regions of barycentric frequencies of the power spectrum Pr[f i] of the received signal by using the provisional Doppler shift output from the peak detector 36 such that the center of a central barycentric frequency calculation region generally coincides with the peak 61.

Stated exactly, the barycentric frequency calculation regions Wr[kJ are offset rightward (toward a higher-frequency side) from the barycentric frequency calculation regions Wt[k] by as much as zfb as shown in FIG. 6, and not f a as shown in FIG. 5. Although the barycentric frequency calculation regions Wr[k] each have the same width as the barycentric frequency calculation regions Wt[k] in the explanation of this embodiment, the width of each barycentric frequency calculation region Wr[k] may be made slightly narrower or wider than the width of each barycentric frequency calculation region Wt [k] . Also, when the amount of Doppler shift fd is output from the later-described convergence detector 45, the barycentric frequency calculation regions Wr[k) are determined in a similar fashion. This means that the barycentric frequency calculation regions Wr[k] are corrected by the amount of Doppler shift fd.

The barycentric frequency calculator 42 calculates barycentric frequencies fwr[k) (k=l, .. .n) for the respective barycentric frequency calculation regions Wr[kJ by equation (4) below: fwr[k] = (Pr(fj] f1)/Epr[fl] (4) where Pr[fj] is the power spectrum applied as a weight factor for each barycentric frequency calculation region Wr[k] If the peaks of the power spectrum Pr[f 1], such as the peak 62, are located generally at central points of the respective barycentric frequency calculation region Wr[k], noise and side lobe components appearing at points on lower-and higher-frequency sides of each spectral peak are canceled out and each barycentric frequency fwr[k] calculated becomes generally equal to a corresponding true frequency shifted by the Doppler effect. On the other hand, if the peaks of the power spectrum Pr[fi], such as the peak 62, are offset leftward (rightward) from the central points of the respective barycentric frequency calculation region Wr[k] when plotted in a power spectrum diagram like FIG. 5, each barycentric frequency fwr[k] calculated becomes higher (lower) than the corresponding true frequency shifted by the Doppler effect. As the barycentric frequency calculation regions Wr[k] are corrected by the amount of Doppler shift fd output from the convergence detector 45, the central points of the individual barycentric frequency calculation region Wr[k] approach the peaks of the power spectrum Pr[f i], such as the peak 62, so that each barycentric frequency fwr[k] calculated also approaches the corresponding true frequency shifted by the Doppler effect.

The Doppler shift corrector 43 converts a difference (Doppler shift) between the barycentric frequency fwr[k) calculated by the barycentric frequency calculator 42 and the barycentric frequency fwt[k] calculated by the barycentric frequency calculator 32 into a Doppler shift fd[k] which corresponds to the Doppler shift of the center frequency (carrier frequency) fc of the transmitting waveform for each value of k (kl, .. .n) by using equation (5) below: fd[k] = (fwr[k] -fwt[kJ) fc/fwt[k] (5) The Doppler shift fd[k] thus calculated is a frequency-corrected version of a Doppler shift (fwr[kJ-fwt [k]) in which the frequency dependence of the Doppler shift is not compensated for. As the Doppler shift is corrected for the frequency dependence thereof by equation (5) above, it is possible to improve the measuring accuracy of the Doppler shift fd as will be further discussed later.

The weighting factor calculator 46 calculates a sum s[k] (k=l,. . .n) of spectral components of the power spectrum Pr[f i] for each of the barycentric frequency calculation regions Wr(k], in which s[k] is substantially an integral value of the power spectrum Pr[f ii. The weighting factor calculator 46 then calculates weighting factors w[k] (k=l, .. .n) from the sums s(k] of the spectral components of the power spectrum Prifi] for the individual barycentric frequency calculation regions Wr[k] Expressing segmented portions of the power spectrum PrEf ii in the individual barycentric frequency calculation regions Wr[k] by Pr[fj], the sum s[k] of the spectral components of each segmented power spectrum Pr[fjj is expressed by Z Pr[fj] and the weighting factor w[k] is expressed by s[k)/ s[k]. While the weighting factors w[k] are calculated from the power spectrum Pr[f ii of the received signal as discussed above in this embodiment, it is possible to calculate the weighting factors w[k] from the power spectrum Pt [f ii of the transmitting waveform which does not contain any external noise. Experiments conducted by the inventors have proved that the measuring accuracy of the Doppler shift fd. slightly deteriorates if the weighting factors w[kJ are calculated from the power spectrum Pt[fi] of the transmitting waveform.

The weighted average processor 44 calculates the amount of Doppler shift fd by performing weighted averaging operation on the Doppler shift fd[k] while applying the weighting factor w[k] calculated by the weighting factor calculator 46 for each value of k (k=l, . . .n). The weighted averaging operation performed by the weighted average processor 44 is expressed by equation (6) below: fd = (fd[k] w[k])/w[k] (6) As a result of this weighted averaging operation, the amount of Doppler shift fd is determined with high accuracy with the influence of noise suppressed in accordance with magnitudes of the spectral components of the power spectrum Pr[fi] of the received echo signal and a distribution thereof and, thus, in accordance with certainty of information on characteristic features of the received echo signal. As an alternative, it is possible to use an integral value of spectral components of the power spectrum Pt[f ii of the transmitting waveform as a weighting factor for each of the barycentric frequency calculation regions Wt[k] instead of the aforementioned weighting factor w[k].

Also, while the integral value of the power spectrum Pr[f ii in each of the barycentric frequency calculation regions Wr[k] is used as the weighting factor w[kJ in the weighted averaging operation of the present embodiment, it is possible to use a maximum value of the spectral components of the power spectrum Pr[f ii in each of the barycentric frequency calculation regions Wr[k] as the weighting factor.

Moreover, the spectral components of the power spectrum Pr[fiJ in each of the barycentric frequency calculation regions Wr[k] may be simply averaged without performing any weighting.

The convergence detector 45 judges whether the amount of Doppler shift fd output from the weighted average processor 44 has converged. If the amount of Doppler shift fd is judged to have converged, the convergence detector 45 delivers the Doppler shift fd to the water current calculator 18 (FIG. 1). If the amount of Doppler shift fd is judged to have not converged yet, the convergence detector 45 delivers the Doppler shift fd back to the barycentric frequency calculation region deteriminer 41.

In the latter case, the Doppler shift fd is used for correcting the barycentric frequency calculation regions Wr[k} and calculating the barycentric frequencies fwt[k] for the corrected barycentric frequency calculation regions Wr[k] . The Doppler shift fd is repeatedly calculated in this way until the calculated values of the Doppler shift fd converge.

More specifically, the convergence detector 45 judges whether the difference between values of the Doppler shift fd obtained in the preceding and current calculation cycles has become equal to or smaller than a specified value (e.g., 1 Hz) after executing the barycentric frequency processing operation a specific number of times (e.g., twice). If the difference between the two values of the Doppler shift fd is equal to or smaller than the specified value, the convergence detector 45 judges that the calculated Doppler shift fd has converged and outputs the value of the Doppler shift fd obtained in the current calculation cycle as the final value of the Doppler shift fd. If the difference between the calculated values of the Doppler shift fd do not converge even when the barycentric frequency processing operation has been repeated a specific number of times (e.g., 30 times), the convergence detector 45 judged that the barycentric frequency processing operation has failed

and generates an "error" statement. As previously

mentioned, the provisional Doppler shift has a value close to the true value of the Doppler shift. Thus, the aforementioned configuration of the embodiment may be so modified as to output the provisional Doppler shift as the final value of the Doppler shift fd without passing through the barycentric frequency processor 40.

Reasons why the barycentric frequency processor 40 repeatedly performs the barycentric frequency processing operation are as follows. A first reason is that the calculated value of the Doppler shift fd contains a slight error due to the influence of noise. This is because the provisional Doppler shift is calculated without any consideration of the frequency dependence of the Doppler shift, so that center frequencies of individual barycentric frequency calculation regions Wr[k] containing the peaks, such as the peak 62, other than the peak 61 are slightly offset from peak frequencies of the power spectrum Pr[f ii As a consequence, the calculated barycentric frequencies fwt[k] are affected by the influence of noise, causing a slight error in the calculated value of the Doppler shift fd. This error gradually decreases as the barycentric frequency calculation regions Wr[k] are corrected by the calculated Doppler shift fd and the Doppler shift fd is recalculated based on the barycentric frequencies fwr[k] in the corrected barycentric frequency calculation regions Wr[k] as discussed above. Briefly stated, the measuring accuracy of the Doppler shift fd improves through the repetitive calculation cycles of the barycentric frequency processing operation.

A second reason why the barycentric frequency processing operation is repeatedly performed is that a frequency error occurs in the provisional Doppler shift due to the number of discrete signal data sampled from the received echo signal, causing an error in the calculated value of the Doppler shift fd. This error can also be made sufficiently small by repeatedly performing the barycentric frequency processing operation as mentioned above.

Alternatively, the error in the provisional Doppler shift due to the number of sampled discrete signal data may be eliminated by using the barycentric frequency of the cross-correlation output, instead of the detected spectral peak frequency, as the provisional Doppler shift, although this approach will require an additional processing time for calculating the provisional Doppler shift.

As thus far described, the peaks of the power spectrum Pr[fi], such as the peak 62, corresponding to the peaks of the power spectrum Pt[f i], such as the peak 52, contained in the barycentric frequency calculation regions Wt[k] are contained in the barycentric frequency calculation regions Wr[k] related to the respective barycentric frequency calculation regions Wt[k] . Moreover, the corresponding peaks (e.g., the peaks 52 and 62) of the two power spectra Pt[fi] and Pr[fi] represent information containing features of the transmitted signal and the received signal having almost an identical spectral waveform. It is therefore possible to obtain the Doppler shift fd with an accuracy high enough from a practical point of view based on the difference between the barycentric frequency fwr[k] in the barycentric frequency calculation region Wr[k] and the barycentric frequency fwt[kJ in the barycentric frequency calculation region Wt(k] for each value of k (k=1,...n).

Since the power spectra Pt[fi] and Pr[f 1] each have multiple spectral peaks, the power spectra PtEfi] and Pr[f ii provide a large amount of information usable for calculating the Doppler shift fd, thus increasing the measuring accuracy of the Doppler shift fd. Additionally, the repeatedly performed barycentric frequency processing operation, a correction for the frequency dependence of the Doppler shift made by the Doppler shift corrector 43 and the weighted averaging operation performed by the weighted average processor 44 together serve to further increase the measuring accuracy of the Doppler shift fd. While the configuration of the Doppler measuring device 3 of the present embodiment has thus far been described with reference to functional blocks, the Doppler measuring device 3 is configured by software implemented in the CPU 20a and the DSP 20b in actuality.

While the individual barycentric frequency calculation regions Wr[k] have a fixed width (frequency range) even when the number of repetitive calculation cycles of the barycentric frequency processing operation increases in the foregoing embodiment, the width of the barycentric frequency calculation regions Wr[k} may be gradually decreased with an increase in the number of repetitive calculation cycles of the barycentric frequency processing operation. As an example, the width of the barycentric frequency calculation regions Wr[k] may be decreased such that the width in the current calculation cycle becomes 90% that in the preceding calculation cycle inasmuch as the width of each barycentric frequency calculation region Wr[k] does not become equal to or smaller than the frequency range of a spectral peak (2/Ta).

FIGS. 7A, 7B and 7C are diagrams showing how the barycentric frequency calculation regions Wr[k) are gradually narrowed as stated above provided that there are three barycentric frequency calculation regions. The barycentric frequency calculation regions for the first calculation cycle of the received signal power spectrum Pr[f ii shown in FIG. 7B have the same width as the barycentric frequency calculation regions of the transmitting waveform power spectrum Pt[fi] shown in FIG. 7A and are offset rightward by as much as the amount of provisional Doppler shift DA. Upward-pointing arrows shown in FIG. 7B indicate barycentric frequencies calculated in the first calculation cycle. The barycentric frequency calculation regions for the second calculation cycle shown in FIG. 7C are made narrower than those for the first calculation cycle and offset rightward by as much as the amount of provisional Doppler shift DB obtained based on ) the barycentric frequency calculation regions for the first calculation cycle shown in FIG. 7B, and not the amount of provisional Doppler shift obtained based on the barycentric frequency calculation regions of the transmitting waveform power spectrum Pt[fi] shown in FIG. 7A.

If the barycentric frequency calculation regions Wr[k] are narrowed from one calculation cycle to the next as described above, a frequency range within which each barycentric frequency should be calculated is constricted to the vicinity of a spectral peak. This serves to reduce the influence of noise whose frequency components are separated from the peaks of the power spectrum Pr[f ii, thereby contributing to a further increase in the measuring accuracy of the Doppler shift fd. Since the Doppler effect, by its nature, causes a contraction or expansion of a spectral pattern along a frequency axis with a 0 Hz point remaining immovable, it is preferable to vary the widths of the individual barycentric frequency calculation regions Wr[k] according to center frequencies thereof in such a manner that the barycentric frequency calculation regions wr[k} have wider widths on a higher-frequency side and narrower widths on a lower-frequency side. Instead of shifting all of the barycentric frequency calculation regions for the second calculation cycle by the amount of the provisional Doppler shift DB as described above, it would be possible to determine a central point of each barycentric frequency calculation region for the second calculation cycle such that the barycentric frequencies calculated in the first calculation cycle are located at the middle of the respective barycentric frequency calculation regions for the second calculation cycle. This approach however might make the result of the barycentric frequency processing operation susceptible to the influence of noise in a case where the noise appears in a specific barycentric frequency calculation region alone.

While a train of multiple M-sequence BPSK signals, a type of broadband signals, which produces a plurality of peaks in the power spectrum Pt[fi] is used as the transmit signal in the foregoing embodiment, the invention can also be implemented by using a transmit signal formed of a train of multiple BPSK signals other than the M-sequence or other types of broadband signals. For example, it is possible to use a signal formed of a train of linear frequency-modulated (FM) signals of which frequency continuously varies from a maximum value fmax to a minimum value frnin (or from fmin to fmax) with the lapse of time within a specific period of time T as shown in FIG. BA, or a signal formed of a train of sine-wave signals having two or more different frequencies (e.g., fl-f3) over specific periods of time in succession as shown in FIG. 8B, as the transmit signal. In short, it is possible to use a broadband signal which produces a plurality of discrete peaks in a power spectrum of a transmitting waveform obtained by discrete Fourier transform as the transmit signal.

Also, while a train of multiple M-sequence BPSK signals which produces a plurality of peaks in the power spectrum Pt[fi] is used as the transmit signal in the foregoing embodiment, the invention can also be implemented by using a narrowband signal, such as a sine-wave signal, which produces a single peak in the power spectrum as a transmit signal. Although there is produced one each barycentric frequency calculation region Wt[k] and Wr[kl only in this case, it is possible to calculate the amount of Doppler shift fd by performing operation similar to that performed in the case of the broadband signals. In this case, however, the power spectra Pt[fi] and Pr[f ii each contain a smaller amount of information on frequencies of the transmitting waveform and the received echo signal usable for frequency detection, so that the measured (calculated) value of the Doppler shift fd is more or less inaccurate compared to the case where the broadband signals are used for measuring the Doppler shift fd.

Although the Doppler shift fd is determined by using the power spectra Pt[f ii and Pr[fi] in the foregoing embodiment, it is possible to calculate the amount of Doppler shift fd by using an amplitude spectrum or three or more power-law spectra. If an amplitude spectrum is used, each spectral peak has a wider width (between 3 dB-down points on both sides of a maximum level point of a main lobe, for example) and, therefore, the amount of the Doppler shift fd is calculated with a slightly poorer accuracy. If three or more power-law spectra are used, on the other hand, the width of each spectral peak of the received echo signal narrows but noise components also have high-level peaks, so that the calculated barycentric frequencies fwt[k) tend to be affected by noise, resulting in some deterioration in antinoise performance.

Furthermore, while the foregoing discussion of the preferred embodiment has used an example in which the discrete Fourier transform is used for calculating a spectrum, various other means are usable for calculating the spectrum. For example, a nonparametric method like the Welch's method, a parametric method like the Yule-Walker autoregressive (AR) method and a subspace method like the multiple signal classification (MUSIC) method are usable for calculating the spectrum. It is possible to achieve approximately the same level of accuracy by these methods as achieved by the discrete Fourier transform.

Moreover, while the foregoing discussion of the preferred embodiment has illustrated the configuration of the water current meter 1 incorporating the Doppler measuring device 3, the present invention may be implemented in an underwater communications system including a transmitter and a receiver. The transmitter of this type of underwater communications system transmits an information signal having a waveform of which frequency varies as a result of combination of a known reference signal and information to be sent. The receiver incorporates a Doppler measuring device 3 having essentially the same configuration as that of the aforementioned embodiment. The Doppler measuring device 3 of the underwater communications system calculates the amount of Doppler shift in the above-described fashion based on a transmitting waveform power spectrum of the reference signal stored in an internal memory, barycentric frequency calculation regions and barycentric frequencies, as well as the received reference signal. Then, the Doppler measuring device 3 corrects the frequency of the received information signal by using the calculated Doppler shift and obtains the transmitted information signal from the frequency-corrected received information signal. An example of this kind of underwater communications system is a net depth monitoring system including an acoustic transmitter with a built-in depth sensor attached to a fishing net for transmitting an information signal (sine-wave signal) and a receiver (hydrophone) installed on a ship's hull for receiving the information signal. The net depth monitoring system measures the depth of the fishing net using the information signal whose frequency varies in accordance with water pressure measured by the depth sensor, for instance.

Claims

WHAT IS CLAIMED IS: 1. A Doppler measuring device for measuring a

Doppler shift of acoustic waves emitted underwater comprising: a spectrum calculator for calculating a spectrum of a received echo signal of the acoustic waves; a barycentric frequency calculation region determiner for determining a second barycentric frequency calculation region for the spectrum of the received echo signal, the second barycentric frequency calculation region being related to a first barycentric frequency calculation region which is a frequency range containing a peak of a spectrum of a transmitting waveform of the acoustic waves; a barycentric frequency determiner for determining a barycentric frequency of the received echo signal within the second barycentric frequency calculation region; and a Doppler shift determiner for determining the Doppler shift of the acoustic waves based on a difference between a barycentric frequency of the transmitting waveform within the first barycentric frequency calculation region and the barycentric frequency of the received echo signal within the second barycentric frequency calculation region corresponding to the first barycentric frequency calculation region.

2. The Doppler measuring device according to claim 1, wherein the spectrum of the transmitting waveform contains a plurality of peaks, and said Doppler measuring device determines the Doppler shift of the acoustic waves based on the difference between the barycentric frequency in each of a plurality of first barycentric frequency calculation regions and the barycentric frequency in the corresponding one of a plurality of second barycentric frequency calculation regions.

3. The Doppler measuring device according to claim 2, wherein said Doppler measuring device converts the difference between the barycentric frequencies in each pair of the first and second barycentric frequency calculation regions into a value corresponding to the amount of Doppler shift of a specific frequency component and determines the Doppler shift of the acoustic waves based on the amount of Doppler shift obtained by such conversion.

4. The Doppler measuring device according to claim 2 or 3, wherein said Doppler measuring device determines the Doppler shift of the acoustic waves by averaging the differences between the barycentric frequencies or the amounts of Doppler shift obtained by said conversion.

5. The Doppler measuring device according to one of claims 1 to 4, wherein said Doppler measuring device performs cross-correlation between the spectra of transmitting waveform and the received echo signal and determines the frequency range of the second barycentric frequency calculation region based on a frequency at which an cross-correlation output is maximized.

6. The Doppler measuring device according to one of claims 1 to 5, wherein said Doppler measuring device repeatedly calculates the Doppler shift until calculated values thereof converge, in which the frequency range of the second barycentric frequency calculation region is corrected based on the Doppler shift value calculated in a preceding calculation cycle.

7. The Doppler measuring device according to claim 6, wherein the frequency range of the second barycentric frequency calculation region is narrowed when corrected based on the Doppler shift value calculated in the preceding calculation cycle.

B. A Doppler measuring device for measuring a Doppler shift of acoustic waves emitted underwater comprising: a cross-correlator for performing cross-correlation between a spectrum of a transmitting waveform of the acoustic waves and a spectrum of a received echo signal of the acoustic waves; and a Doppler shift determiner for determining the Doppler shift of the acoustic waves based on a frequency at which an cross-correlation output is maximized.

9. water current meter for determining velocity of a water current from speed over ground and through-water speed of a ship determined by measuring a Doppler shift of an echo signal of acoustic waves emitted underwater, said water current meter comprising the Doppler measuring device as defined in one of claims 1 to B for measuring the Doppler shift of the echo signal based on the transmitting waveform and the received echo signal of the acoustic waves, wherein said water current meter determines the speed over ground and the through-water speed of the ship based on the Doppler shift measured by said Doppler measuring device.

10. A Doppler measuring device substantially as hereinbefore described with reference to the accompanying drawings.

11. A water current meter substantially as hereinbefore described with reference to the accompanying drawings.