JPH11134002A - Control method for adaptive array and adaptive array device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a adaptive array which has small breathing noise and generates an output signal of high quality. SOLUTION: From a signal having its phase adjusted so as to match the phase of a component of a target signal between two sensor output signals, the result p(k) obtained by comparing 1st correlation calculated with a 1st time constant with a 1st threshold value and the result q(k) obtained by comparing 2nd correlation calculated with a 2nd time constant with a 2nd threshold value are inputted to a state circuit 60. The output of this state circuit 60 is regarded as the step size of the adaptive algorithm of an adaptive filter. The internal state of the state circuit 60 represents which of state transition due to a target signal stop or state transition due to target signal generation the next generated state transition is.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an adaptive array that uses a plurality of sensors to spatially receive signals.

[0002]

2. Description of the Related Art In the field of audio signal acquisition, sonar, wireless communication, and the like, in order to receive only a specific signal from a plurality of signal sources, adaptive microphone array is used as an application of adaptive microphone array technology. 2. Description of the Related Art A wireless transmission / reception device using an enhancement device or an adaptive antenna array is known.

As a sensor, a microphone, an ultrasonic sensor, a sonar receiver, an antenna, or the like can be used. Here, a case where a microphone is used as a sensor will be described. For the sake of simplicity, consider a case where microphones are arranged at equal intervals on a line segment. Also, consider a situation in which the target sound source is sufficiently distant from the line segment on which the microphone is located, and the target sound source exists in a direction substantially orthogonal to the line segment.

The microphone array forms a spatial filter by filtering signals collected by a plurality of microphones and then adding the filtered signals. With this spatial filter, ambient noise is attenuated, and only a signal arriving from a predetermined direction, that is, a target signal is received.
An adaptive microphone array is a microphone array that changes the spatial filter characteristics adaptively. The configuration of the adaptive microphone array is described in the literature, “Generalized Sidelobe Canceller” (IEE, Transactions on Antennas and Propagation (I
EEE, Transactions on Anten
(Nas and Propagation), Vol. 30, No. 1, 1982, pp. 27-34: hereinafter referred to as "Reference 1"), (IEE, Transactions on Antennas and Propagation (I
EEE, Transactions on Anten
nas and Propagation), Vol. 40, No. 9, 1992, pp. 1093-1096: hereinafter referred to as "Reference 2", IEICE Transactions A, Vol. 79, No. 9, 1996, 1516 ~ 15
Page 24: (hereinafter referred to as “Reference 3”), “Frost beamformer” (IEE,
Proceedings of IEE (IEEE
E, Proceedings of IEEE), No. 6
Vol. 0, No. 8, 1972, pp. 926-935: hereinafter referred to as "Document 4"), (IEEE, Proceedings of International Conference on Acoustic Speech and Signal Processing 94 (IEEE, Proceedi)
ngs of International Conf
erenceon Acoustics, Speech
and Signal Processing 9
4), 1994, pages IV-269-272: hereinafter referred to as “Reference 5”, (IEEE, Proceedings of International).
Conference on acoustics speech
And Signal Processing 97 (IEEE,
Proceedings ofInternational
al Conference Acoustic
s, Speech and Signal Process
ssing 97), 1997, I-367-370.
Page: hereinafter referred to as “Document 6”).

Here, as a representative, the operation of the configuration of Document 3 will be described with reference to the drawings. FIG.
FIG. 11 is a block diagram in a case where a signal processing unit of an adaptive array shown in Reference 3 as a first conventional method is configured using M microphones. Microphone group 1m (m = 0, 1,
. . . , M-1) are each converted from an analog signal to a digital signal. These digital signals (hereinafter referred to as “microphone signals”) are subjected to signal processing to extract target signals.

The first conventional method comprises a fixed beamformer 2, a blocking matrix 4, and a multi-input canceller 3. Hereinafter, each of the fixed beamformer 2, the blocking matrix 4, and the multi-input canceller 3 will be described in detail. First, the fixed beam former 2 will be described.
In FIG. 6, a fixed beam former 2 is a microphone 1
Receive the signal group from. From these microphone signal groups, a signal is output in which the target signal is emphasized and signals other than the target signal are attenuated.

As the fixed beamformer, a delay-and-sum beamformer that delays and adds signals received from microphone groups and a filter-and-sum beamformer that filters and adds signals received from microphone groups respectively are used. It is possible.
These fixed beamformers are described in the literature, D.A.
H. Johnson, D.M. E. FIG. Dudgeon's "Ar
ray Signal Processing ”(Pre
nice Hall, EnglewoodCliff
s, 1993) in Chapter 4 (hereinafter referred to as "Reference 7"). Here, the operation will be described using a delay and thumb beamformer.

The processing in the delay and sum beamformer is represented by the following equation (1).

[0009]

(Equation 1)

Here, k indicates the sample number on the time axis, and rm indicates the number of delayed samples of each microphone signal xm (k). g (k) is an output signal of the fixed beamformer 2, xm (k) is an output signal of the microphone lm, and fm is a coefficient corresponding to the microphone signal in the fixed beamformer. The delay and sum beamformer converts each microphone signal xm (k) into rm
The sum of signals obtained by multiplying the value delayed by the number of samples by the coefficient fm is calculated and output.

The number rm of each delayed sample is represented by a signal xm (k-rm) obtained by delaying the output of each microphone lm.
Set so that the phase of the target signal is the same. As a result, xm (k-rm) [m = 0, 1,. . . , M-1]
Are added, the target signal is emphasized. On the other hand, since the disturbing signals arrive from directions other than the target signal and have significantly different phases, when the signals xm (k-rm) obtained by delaying the outputs of the microphones are added, they cancel each other out and attenuate. Therefore, at the output of the fixed beamformer, the target signal is enhanced and the interfering signal is attenuated.

Next, the blocking matrix 4 will be described with reference to the drawings. In FIG. 6, the blocking matrix 4 includes a microphone group 1m (m = 0, 1,..., M
-1), the target signal is attenuated using adaptive signal processing, and a signal other than the target signal, that is, a signal group in which an interference signal is enhanced is transmitted to the multi-input canceller 3. The amount of coefficient updating in adaptive signal processing is determined by the step size α (k) at the step size input terminal 8. In the first conventional method shown in FIG.
(K) is always 1 and is time-invariant.

Next, the detailed configuration and operation of the blocking matrix 4 will be described with reference to FIG. In FIG. 7, the blocking matrix 4 is the fixed beamformer 1
03, delay unit group 104m (m = 0, 1,... M−
1), an adaptive filter group 105m (m = 0, 1,...,
M-1), a group of subtractors 106m (m = 0, 1,..., M
−1), a multiplier group 49m (m = 0, 1,..., M−
1) It comprises a multiplier 79.

Adaptive filter group 10 of blocking matrix 4
As the filter structure of 5 m, it is possible to use an FIR (finite impulse response) filter, an IIR (infinite impulse response) filter, a lattice filter, or the like. Here, a case where an FIR filter is used will be described. The blocking matrix 4 is a microphone group 1 m
A signal group from (m = 0, 1,..., M-1) is received through an input terminal 101m (m = 0, 1,..., M-1), and a fixed beamformer 103 and a corresponding delay are received. Group 1
04m. The fixed beamformer 103 receives the signal group from the microphone group 1m, and outputs a signal in which the target signal is enhanced and the interference signal is attenuated by the same signal processing as that of the fixed beamformer 2 in FIG. Fixed beam former 103 and fixed beam former 2
Can also be shared. Fixed beam former 1
The output signal 03 is an input signal common to the adaptive filter group 105m.

Each delay unit group 104m has an input terminal 101m.
From the corresponding subtractor 106m
To communicate. Each adaptive filter 105m receives the output signal of the fixed beamformer 103, and transmits an output signal obtained by convolving the tap coefficient of each adaptive filter 105m to the corresponding subtractor 106m. Each subtractor 106
m subtracts the output signal of the corresponding adaptive filter 105m from the output signal of the corresponding delay unit 104m. The subtraction result of the subtractor 106m is transmitted as an output signal of the blocking matrix 104 to the multi-input canceller 3 of FIG. 6 through the output terminal 102m, and at the same time, is transmitted to the corresponding multiplier 49m for updating the tap coefficients. . Each multiplier 4
9m is the subtraction result of the corresponding subtractor 106m and the multiplier 7
The multiplication is performed in response to the output of No. 9 and the multiplication result is transmitted to the corresponding adaptive filter 105m for updating the tap coefficients. The multiplier 79 multiplies the step size received through the step size input terminal 8 by the step gain α0, and transmits the result to the multiplier group 49m. Delay device 104
The number m of delay samples is set so that the phase of the target signal component at the output of the delay unit 104m and the phase of the target signal component at the output of the adaptive filter 105m can match.

The processing in the blocking matrix is represented by the following equation (2).

[0017]

(Equation 2)

Here, ym (k) is the output signal of the group of subtractors 106m, xm (kP) is the output signal of the delay unit 104m, and P is the number of delay samples of the delay unit 4m. Also,
Hm (k) is the tap coefficient vector of the adaptive filter 105m, ｛·｝ T is the transpose, and D (k) is the fixed beamformer 1.
03 is a signal vector composed of a plurality of signals obtained by delaying the output signal d (k), and is defined by the following equations (3) and (4), where N is the number of taps.

[0019]

(Equation 3)

[0020]

(Equation 4)

Next, the adaptive filter 105m will be described. Various adaptive filters, such as a tap coefficient constrained adaptive filter and a leak adaptive filter, can be used as the adaptive filter 105m. Here, a case in which a leak adaptive filter is used will be described. The adaptive filter 105m updates tap coefficients based on the signal from the corresponding subtractor 106m so that the output signal power of the subtractor 106m is minimized.

Hereinafter, updating of coefficients in the adaptive filter 105m will be described. NL is used as an adaptive algorithm for updating tap coefficients in adaptive filter 105m.
MS algorithm (also called "learning identification method"), R
There are an LS algorithm (also called a "sequential least squares method"), a projection algorithm, a gradient method, a least squares method, a block adaptive algorithm, an adaptive algorithm using an orthogonal transform, and the like. Here, NL is used as the adaptive algorithm.
The case where the MS algorithm is used will be described.

The tap coefficient Hm (k) of the leak adaptive filter using the NLMS algorithm (m = 0, 1,
. . . , M-1) are updated by the following equations (5) and (6).

[0024]

(Equation 5)

[0025]

(Equation 6)

Where ‖ · ‖ is the Euclidean norm, α
0 is a step gain and δ is a leak coefficient. In the first conventional method shown in FIG. 6, since the step size α (k) is always 1, the step gain α0 is directly multiplied. Updating the tap coefficients of each adaptive filter 105m minimizes the power of the signal at the output of the corresponding subtractor 106m. Since the output signal of the fixed beamformer 103, which is the input signal of the adaptive filter 105m, mainly includes the target signal, as a result of minimizing the output signal power of each subtractor 106m, the output of the subtractor 106m greatly attenuates the target signal. Will be done.

Next, the multi-input canceller 3 will be described with reference to the drawings. In FIG. 6, the multi-input canceller 3 receives the output signal of the fixed beamformer 2 and the output signal group of the blocking matrix 4 and, from the signal obtained by delaying the output of the fixed beamformer 2 by adaptive signal processing,
A component having a correlation with the output signal group of the blocking matrix 4 is removed. The amount of coefficient updating in adaptive signal processing is determined by the step size β at the step size input terminal 9.
(K). In the first conventional method, β
(K) is always 1 and is time-invariant.

Next, the configuration and operation of the multi-input canceller 3 will be described with reference to FIG. The multi-input canceller 3 includes an adaptive filter group 207m (m = 0, 1, 1).
. . . , M-1), the adder 208, and the delay unit 211.
, A subtractor 209, a multiplier 212, and a multiplier 214
It is composed of Each adaptive filter 207m outputs the output signal of the corresponding subtractor 106m from the blocking matrix 4 to an input terminal 201m (m = 0, 1,..., M−
1), and transmits the result of convolution with the filter coefficient to the adder 208 as an output signal.

As the filter structure of the adaptive filter 207m, an FIR filter, an IIR filter, a lattice filter, or the like can be used. Here, F
The case where an IR filter is used will be described. Adder 2
08 receives all output signal groups of the adaptive filter group 207m, calculates the sum of the received signals, and transmits the calculation result as an output signal to the subtractor 209.

The delay unit 211 receives the output signal of the fixed beam former 2 through the input terminal 202, delays the output signal, and transmits the delayed signal to the subtractor 209. The subtracter 209 subtracts the output signal of the adder 208 from the output signal of the delay unit 211. The output signal of the subtracter 209 is the output of the multi-input canceller 3, that is, the output of the entire adaptive array device.
Output from the output terminal 10.

The output signal of the subtracter 209 is transmitted to the multiplier 212 for updating the coefficient. Multiplier 212
Multiplies the signal received from the subtractor 209 by the signal received from the multiplier 214 to generate an adaptive filter group 207m (m = m
0, 1,. . . , M-1). The multiplier 214
The step size β (k) is received through the step size input terminal 9 and is multiplied by a step gain β0.
Send to 2. Each adaptive filter 207m includes a multiplier 212
The tap coefficient is updated based on the signal received from.

The number of delay samples Q of the delay unit 211 is such that the phase of a signal arriving from an arbitrary direction is
The setting is made so that the output of m and the output of the delay unit 211 can match. For example, the time is set to about ４ to ４ of the number of taps of the adaptive filter 207m. The processing in the multi-input canceller 3 is represented by the following equation (7).

[0033]

(Equation 7)

Here, z (k) is the output signal of the subtractor 209, that is, the output signal of the entire adaptive array. G (k−Q) is a signal obtained by delaying the output signal of the delay unit 211, that is, the output signal of the fixed beamformer 2. Wm (k) is a tap coefficient vector of the adaptive filter 207m, and Ym (k) is an adaptive filter 20m.
This is a signal vector composed of a signal obtained by delaying the 7 m input signal, that is, the output signal ym (k) of the blocking matrix 4. Wm (k) and Ym (k) indicate the number of taps.
L is defined as in the following equations (8) and (9).

[0035]

(Equation 8)

[0036]

(Equation 9)

In each adaptive filter 207m, the subtracter 20
The tap coefficient Wm (k) is updated so that the power of the output signal z (k) of No. 9 is minimized. Hereinafter, the adaptive filter group 207m will be described. Adaptive filter 207m
It is possible to use a tap coefficient constrained adaptive filter or a leak adaptive filter, but the case where a leak adaptive filter is used will be described here. Further, as adaptive algorithms for updating tap coefficients in the adaptive filter 207m, NLMS algorithm, RLS algorithm, projection algorithm, gradient method, least square method,
There are a block adaptation algorithm and an adaptation algorithm using an orthogonal transform. Here, the case where the NLMS algorithm is used as the adaptive algorithm will be described. Further, when the NLMS algorithm is applied as an adaptive algorithm, there are a method of applying the NLMS algorithm for each adaptive filter and a method of applying the NLMS algorithm collectively to all the adaptive filter groups. Here, a method of applying the NLMS algorithm to all adaptive filters will be described.

The updating of the tap coefficient Wm (k) in the adaptive filter 207m is represented by the following equation (10), where β0 is the step gain.

[0039]

(Equation 10)

Here, γ is a leak coefficient. In the first conventional method, since the step size β (k) is always 1, the step gain β0 is directly multiplied.
As a result of updating the tap coefficients of the adaptive filter 207m, the output of the signal at the output of the subtractor 209 is minimized. Subtractor 106m which is an input signal to each adaptive filter 207m
Since the output signal of the subtractor 209 mainly includes an interference signal, as a result of minimizing the output signal power of the subtractor 209, the interference signal is greatly attenuated in the output signal of the subtractor 209, that is, the signal at the output terminal 10. Become. The signal at the output terminal 10 becomes the output signal of the entire adaptive array.

According to the first conventional method described above, a target signal can be extracted in a situation where an interfering signal exists. In the first conventional method, when only a target signal is present and no interfering signal is present, the tap coefficient Wm in the adaptive filter group 207m of the multi-input canceller 3 is set.
Since (k) is disturbed by the target signal, the target signal deteriorates at the final output.

When the interference signal is regenerated after the tap coefficient Wm (k) is disturbed, the interference signal is not removed,
Breathing noise occurs at the final output. In order to prevent such deterioration and noise, it is necessary to reduce the step gain β0 of the adaptive algorithm so that the adaptive filter group 207m of the multi-input canceller 3 is not disturbed. However, when the step gain β0 is small, the following speed of the adaptive filter group 207m in the multi-input canceller 3 with respect to the movement of the disturbing signal source becomes slow, and the period during which the disturbing signal removal in the final output becomes insufficient becomes long.

When only the interfering signal is present and the target signal is not present, the tap coefficient Hm (k) in the adaptive filter group 105m of the blocking matrix 4 is disturbed, and a breathing noise is generated in the final output. To prevent this, make sure that the adaptive filter 105m is not disturbed.
It is necessary to reduce the step gain α0 of the adaptive algorithm. However, when the step gain α0 is small, the blocking matrix 4
, The following speed of the adaptive filter group 105m becomes slow, and the quality of the target signal at the final output is deteriorated.

To deal with such a problem, an adaptive array apparatus in which the step size of an adaptive filter is controlled with the aim of causing the movement of an interfering signal source to follow at a high speed while maintaining the quality of an output signal high, Publications, Journal of Acoustic Society of America (Journal of Acoustic)
Society of America), Vol. 91, No. 3
No. 1992, pp. 1662-1676 (hereinafter referred to as "Reference 8").

A method for controlling the step size disclosed in Reference 8 will be described below with reference to the drawings. FIG.
FIG. 11 is a block diagram in a case where a step size control circuit shown in Reference 8 is applied to an adaptive array device shown in Reference 3 as a second conventional method. The difference between the adaptive array device shown in FIG. 6 and the adaptive array device shown in FIG. 9 is that both the step size α (k) and the step size β (k) are time-invariant in the adaptive array device shown in FIG. In contrast to the constant 1, the adaptive array device shown in FIG. 9 is controlled by the step size control circuit 50 in a time-varying manner. Hereinafter, the step size control circuit 50 will be described.

The step size control circuit 50 comprises code cross-correlation calculation circuits 20 and 40, comparison circuits 17 and 37, threshold value generation circuits 19 and 39, and an inversion circuit 69.
The code cross-correlation calculation circuit 20 is connected to the corresponding microphone 1
The signal from _j and 1 _{j + 1} is received, the code cross-correlation is calculated, and the calculation result is transmitted to the comparison circuit 17. Comparison circuit 17
Receives the output value of the code cross-correlation calculation circuit 20 and the threshold value θb from the threshold value generation circuit 19 and compares them. When the output value of the code cross-correlation calculation circuit 20 is larger than the threshold value θb, 1 is output, and when it is smaller, 0 is output.
The output of the comparison circuit 17 is transmitted to the blocking matrix 4 as a step size α (k).

Similarly, the code cross-correlation calculating circuit 40 receives signals from the corresponding microphones 1 _j and 1 _{j + 1} and
The code cross-correlation is calculated, and the calculation result is transmitted to the comparison circuit 37. The comparison circuit 37 includes a code cross-correlation calculation circuit 40
And the threshold value θc from the threshold value generation circuit 39, and compares them. If the output value of the code cross-correlation calculation circuit 40 is larger than the threshold value θc, 1 is output, and if it is smaller, 0 is output.

The output of comparison circuit 37 is transmitted to inversion circuit 69. The inverting circuit 69 receives the output of the comparing circuit 37 as an input, and outputs 0 when the input is 1 and outputs 1 when the input is 0. The output of the inverting circuit 69 is transmitted to the multi-input canceller 3 as a step size β (k). The code cross-correlation calculation circuit 20 includes band-pass circuits 11 and 12,
It comprises sign determination circuits 13 and 14, a multiplication circuit 15, and a low-pass circuit 16. The bandpass circuits 11 and 12 have the same frequency characteristics, receive signals from the corresponding microphones 1 _j and 1 _{j + 1} , respectively, pass necessary frequency components, and attenuate unnecessary components. Depending on the environment,
In some cases, all frequency components are passed. In this case, the band pass circuits 11 and 12 can be omitted.

The output signals of the bandpass circuits 11 and 12 are sent to the corresponding code determination circuits 13 and 14, respectively.
The sign determination circuits 13 and 14 receive the output signals of the corresponding bandpass circuits 11 and 12, respectively, and output 1 when the sign of the signal is positive, and output 1 when the sign of the signal is negative. Outputs -1. Output signals of the sign determination circuits 13 and 14 are sent to a multiplication circuit 15.

In the multiplication circuit 15, the sign determination circuits 13 and 1
4, the product is calculated, and the calculation result is sent to the low-pass circuit 16. The low-pass circuit 16 receives the output signal of the multiplication circuit 15 and attenuates high-frequency components.
Outputs a signal that has passed low-frequency components. The output signal of the low-pass circuit 16, that is, the code cross-correlation calculating circuit 20
Is the code cross-correlation of the signals from the microphones 1 _j and 1 _{j + 1} .

Here, the nature of the output signal of the code cross-correlation calculation circuit 20, that is, the output signal of the low-pass circuit 16 will be described. When the microphone arrangement straight line is substantially orthogonal to the target signal source direction, the phases of the components of the target signal are almost the same in the signals of the microphones 1 _j and 1 _{j + 1} . Therefore, when the target signal exists, the output signal of the low-pass circuit 16 increases. Conversely, when the target signal does not exist, the output signal of the low-pass circuit 16 becomes small.
Therefore, the output signal of the low-pass circuit 16, that is, the output signal of the code cross-correlation calculation circuit 20, is an index regarding the presence or absence of the target signal. The output signal of the code cross-correlation calculation circuit 20 is compared with a threshold value θb in the comparison circuit 17 to detect the presence or absence of the target signal, and the step size α (k) is controlled based on the detection result. .

The configuration of the code cross-correlation calculation circuit 40 is exactly the same as that of the code cross-correlation calculation circuit 20, so that the description thereof will be omitted. Similarly to the output signal of the code cross-correlation calculation circuit 20, the output signal of the code cross-correlation calculation circuit 40 is an index regarding the presence or absence of the target signal. This code cross-correlation calculation circuit 4
By comparing the output signal of 0 with the threshold value θc in the comparison circuit 37, the presence or absence of the target signal is detected, and the step size β (k) is controlled based on the output obtained by inverting the detection result by the inversion circuit 69. .

However, each parameter in the code cross-correlation calculation circuit 40 may be different from each parameter in the code cross-correlation calculation circuit 20. In particular, when the time constant of the low-pass circuit 36 is smaller than that of the low-pass circuit 16, the noise removal performance is higher. As described above, by using the step size control circuit 50 to control the blocking matrix and the step size of the multiple input canceller, it is possible to quickly follow the movement of the disturbing signal source while maintaining the quality of the output signal high. be able to.

[0054]

However, the low-pass circuit 1 of the code cross-correlation calculation circuit 20 according to the second conventional method.
In the time constant setting in No. 6, there was a trade-off between the deterioration of the output signal quality and the breathing noise due to the accuracy of the target signal detection. Breathing noise is generated when a target signal is erroneously detected when there is no target signal (hereinafter, referred to as “false detection”). This erroneous detection occurs when the variance of the code cross-correlation is large. Therefore, in order to reduce erroneous detection, the time constant of the low-pass circuit 16 may be increased.

On the other hand, the deterioration of the target signal occurs when it is determined that the target signal does not exist when the target signal exists (hereinafter, referred to as “detection failure”). This detection failure occurs when the change in the code cross-correlation cannot follow the presence or absence of the target signal. Therefore, in order to reduce detection failure, the time constant of the low-pass circuit 16 may be reduced. Thus, there is a trade-off in setting the time constant of the low-pass circuit 16. A similar trade-off exists in the setting of the time constant in the low-pass circuit 36 of the code cross-correlation calculation circuit 40.

The threshold value θ generated by the threshold value generation circuit 19
Also for b, there is a trade-off between the degradation of the output signal quality and the breathing noise due to the accuracy of the target signal detection. In order to reduce breathing noise, the probability of false detection must be reduced. Further, in order to reduce erroneous detection, the threshold value θb must be set large.

However, if the threshold value θb is large, the tracking of the index to the presence or absence of the target signal is delayed, so that the quality of the output signal is deteriorated especially when the target signal is generated. A similar trade-off exists for the threshold θc generated by the threshold generation circuit 39. In order to increase the quality of the output signal, the probability of detection failure must be reduced. In order to reduce the probability of detection failure, the threshold θc must be set small. However, if the threshold θc is set small, the probability of erroneous detection increases, so that the convergence speed decreases,
An increase in breathing noise occurs.

The present invention has been made in view of the above-mentioned problems in the conventional adaptive array control method, and has a small breathing noise and a high output signal quality control method for an adaptive array. , An adaptive array device.

[0059]

According to a first aspect of the present invention, a specific signal source is selected from a plurality of signal sources using a plurality of sensors and an adaptive filter. In the adaptive array control method for receiving only the target signal source, a first mutual constant calculated by a first time constant from a signal whose phase is adjusted so that the phases of the components of the target signal in the two sensor output signals coincide with each other. The output of the state circuit, which receives the result of comparing the correlation with the first threshold value and the result of comparing the second cross-correlation calculated with the second time constant and the second threshold value, with the adaptive filter And a step size of the adaptive algorithm.

According to a second aspect of the present invention to solve the above-mentioned problem, a plurality of sensors and an adaptive filter are used to
In the adaptive array control method for receiving only a specific signal source as a target signal source from among a plurality of signal sources, a method of controlling an adaptive array includes the following.
A result of comparing a first code cross-correlation calculated with a first time constant from a signal whose phase has been adjusted so as to match the phase of a target signal component in two sensor output signals with a first threshold, and Wherein the output of the state circuit, which receives as input the result of comparison between the second code cross-correlation calculated with the time constant and the second threshold, is used as the step size of the adaptive algorithm in the adaptive filter. A method for controlling an adaptive array is presented.

According to a third aspect of the present invention to solve the above-mentioned problems, a plurality of sensors and an adaptive filter are used.
In the adaptive array control method for receiving only a specific signal source as a target signal source from among a plurality of signal sources, a method of controlling an adaptive array includes the following.
A result obtained by comparing a first threshold value with a first cross-correlation coefficient calculated by a first time constant from a signal whose phase has been adjusted to match the phase of a component of a target signal in the two sensor output signals; Wherein the output of the state circuit, which has as input the second cross-correlation coefficient calculated with the time constant of 2 and the result of comparison with the second threshold, is used as the step size of the adaptive algorithm in the adaptive filter. A control method of an adaptive array is presented.

Also, the internal state of the state circuit may be such that the next state transition is a state transition due to a stop of the target signal, or
A method of controlling an adaptive array according to the present invention is also provided, which indicates whether a state transition is caused by generation of a target signal, and wherein control of a step size by the state circuit differs depending on the internal state. Further, the internal state of the state circuit is
The first cross-correlation or the first code cross-correlation or the first cross-correlation coefficient is greater than the first threshold, and the second cross-correlation or the second code cross-correlation or the When the second cross-correlation coefficient is larger than the second threshold, the state transition is performed so as to indicate that the next state transition is a state transition due to the stop of the target signal, and the first cross-correlation or the first cross-correlation is performed. Code cross-correlation or the first cross-correlation coefficient is smaller than the first threshold value, and the second cross-correlation or the second code cross-correlation or the second cross-correlation coefficient is The above-mentioned adaptive array control method, wherein the state transition is performed so as to indicate that the next state transition is a state transition due to the generation of a target signal when the threshold value is smaller than the threshold value of 2, is also presented.

According to a fourth aspect of the present invention for solving the above-mentioned problems, a plurality of sensors and an adaptive filter are provided,
In an adaptive array that receives only a specific signal source as a target signal source from among a plurality of signal sources, a first signal is adjusted from a signal whose phase has been adjusted to match the phase of a target signal component in two sensor output signals. The result obtained by comparing the first cross-correlation calculated with the time constant of the first time with the first threshold value and the result of comparing the second cross-correlation calculated with the second time constant with the second threshold value are input to The output of the state circuit is set to a step size of an adaptive algorithm in the adaptive filter.

According to a fifth aspect of the present invention for solving the above-mentioned problems, a plurality of sensors and an adaptive filter are provided.
From among a plurality of signal sources, in an adaptive array receiving only a specific signal source as a target signal source, a signal whose phase has been adjusted to match the phase of the component of the target signal in two sensor output signals, The result of comparing the first code cross-correlation calculated with the first time constant with the first threshold value, the second code cross-correlation calculated with the second time constant and the second code cross-correlation
An adaptive array device, characterized in that an output of a state circuit, which receives a result of comparison with the threshold value of the adaptive filter, is set as a step size of an adaptive algorithm in the adaptive filter.

According to a sixth aspect of the present invention to solve the above-mentioned problems, a plurality of sensors and an adaptive filter are provided.
In an adaptive array that receives only a specific signal source as a target signal source from among a plurality of signal sources, a first signal is adjusted from a signal whose phase has been adjusted to match the phase of a target signal component in two sensor output signals. The result of comparing the first cross-correlation coefficient calculated with the time constant of the first time with the first threshold value, and the result of comparing the second cross-correlation coefficient calculated with the second time constant with the second threshold value And an output of the state circuit having the input as the step size of the adaptive algorithm in the adaptive filter is provided.

The internal state of the state circuit indicates whether the next state transition is a state transition due to the stop of the target signal or a state transition due to the generation of the target signal. The adaptive array device is provided, wherein the adaptive array device is different depending on the internal state. And the internal state of the state circuit is such that the first cross-correlation or the first code cross-correlation or the first cross-correlation coefficient is greater than the first threshold and the second cross-correlation Or the second code cross-correlation or the second
When the cross-correlation coefficient is larger than the second threshold value, the state transition is performed so as to indicate that the next state transition is a state transition due to the stop of the target signal, and the first cross-correlation or the first code is performed. The cross-correlation or the first cross-correlation coefficient is smaller than the first threshold, and the second cross-correlation or the second code cross-correlation or the second cross-correlation coefficient is the second The above adaptive array device is also provided, wherein a state transition is performed so as to indicate that the next state transition is a state transition due to generation of a target signal when the value is smaller than the threshold value.

According to the adaptive array control method and the adaptive array apparatus of the present invention, the detection of the target signal becomes accurate, so that the step size control can be performed more accurately, the breathing noise is reduced, and the output is reduced. Signal quality degradation is reduced.

[0068]

Next, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of an adaptive array device according to the first embodiment of the present invention. In the present embodiment, the step size control circuit 50 is replaced with a step size control circuit 51 in the second conventional method shown in FIG. The difference between the step size control circuit 50 and the step size control circuit 51 is that the inversion circuit 69 is removed and the state circuit 60 is added. Referring to FIG. 1, comparison circuits 17 and 3
7 is transmitted to the state circuit 60. The state circuit 60 receives the output signals of the comparison circuits 17 and 37,
Step sizes α (k) and β (k) are output. Less than,
The state circuit 60 will be described.

The state circuit 60 has one internal state i (k).
And the internal state i (k) and the comparison circuits 17 and 37
Output the internal state i (k + 1), the step size α (k), and the step size β (k) in the next sample (k + 1). Table 1 shows the state circuit 60
Is an example of the state transition and changes in the step sizes α (k) and β (k).

[0070]

[Table 1]

In Table 1, 0/1 may be either 0 or 1, that is, so-called Don't care (Don't care).
care) term. The internal state i (k) indicates whether the next state transition occurs due to the generation of the target signal or the stop of the target signal. Here, the internal state i
(K) is set to 1 when it means that the next state change is due to generation of the target signal, and is set to 0 when it means that the next state change is due to stoppage of the target signal. The internal state i (k) changes according to the comparison result of the comparison circuits 17 and 37 which is the input signal. When both outputs of the comparison circuits 17 and 37 are 0, the internal state i (k + 1) at the next sample (k + 1) is 1
Becomes

When both the outputs of the comparison circuits 17 and 37 are 1, the internal state i at the next sample (k + 1)
(K + 1) becomes 0. Apart from these two cases, the internal state i (k + 1) at the next sample (k + 1) is
Maintain the internal state i (k) at the current sample k. Further, as shown in Table 1, the step size α (k)
Is 1 when the output of the comparison circuit 17 is 1 and the output of the comparison circuit 37 is 1
This is the case where (k) is 1. Otherwise, the step size α (k) is zero. Step size β
(K) becomes 1 when the output of the comparison circuit 17 is 0 and the output of the comparison circuit 37 is 0. Otherwise, the step size β (k) is zero.

FIG. 2 is a time chart schematically showing the internal signal, threshold value, step size, and internal state of the step size control circuit 51. Next, the operation of the step size control circuit 51 will be described with reference to FIG. FIG. 2 shows an internal signal p common to the step size control circuit 50 of the second conventional method and the step size control circuit 51 of the present invention.
(K), q (k) waveforms, threshold values θb, θc, and step size α output by the step size control circuit 50
The waveforms of (k) and β (k), the internal state i (k) of the step size control circuit 51, and the waveforms of the output step sizes α (k) and β (k) are schematically shown. I have. Ideally, the step size α (k) in the blocking matrix 4 should be 1 and the step size β (k) in the multi-input canceller 3 should be 0 at the moment of generation of the target signal. Conversely, at the moment when the target signal stops, the step size α (k) should be 0 and the step size β (k) should be 1.

In FIG. 2, the waveform of the step size α (k) in the step size control circuit 50 according to the second conventional method is 1 after being delayed from the time when the target signal is generated. The waveform of the step size α (k) in the step size control circuit 51 becomes 1 immediately after the generation of the target signal. That is, the result of the target signal detection by the step size control circuit 51 of the present invention has a small delay from generation and stop of the target signal. That is,
When the step size control circuit 51 of the present invention is used,
The target signal can be detected with higher accuracy than when the step size control circuit 50 of the second conventional method is used. As a result, breathing noise in the output signal and deterioration of the target signal quality are reduced.

Although Table 1 has been used as a table for determining the state transition of the state circuit 60 and the step sizes α (k) and β (k) in the present invention, a table different from Table 1 may be used. It is. Table 2 shows state transitions and step sizes α (k) and β (k) in the state circuit 60.
9 shows another example of a table for determining the number.

[0076]

[Table 2]

The difference between Tables 1 and 2 is only the method of determining α (k). When α (k) is 1 in Table 2, in addition to the case of Table 1, when the internal state i (k) is 0 and the output of the comparator 17 is 1, or when the internal state i
This is the case where (k) is 0 and the output of the comparator 37 is 1. FIG. 3 is another time chart schematically showing the internal signal, the threshold value, the step size, and the internal state of the step size control circuit 51.

FIG. 3 shows a case where the state transition table of Table 2 is used in the step size control circuit 51 of the present invention.
The waveforms of the internal signals p (k) and q (k) and the threshold values θb and θ
c, and the waveforms of the step sizes α (k) and β (k) are schematically shown. 3 is different from FIG. 2 only in the waveform of α (k) in the present invention. As can be seen from FIG. 3, when the state transition according to Table 2 is performed in the step size control circuit 51 of the present invention, the target signal can be detected with higher accuracy than when the step size control circuit 50 of the second conventional method is used. You.

In the case where Table 2 is used, similarly to the case where Table 1 is used, the internal state i () indicating whether the next state transition is due to the generation of the target signal or the stop of the target signal. By using k), highly accurate target signal detection is possible, and the same effect as in the case of using Table 1 is produced. Table 3 shows another example of a table for determining state transitions and step sizes α (k) and β (k) in the state circuit 60.

[0080]

[Table 3]

The difference between Tables 1 and 3 lies only in the method of determining β (k). When β (k) becomes 1 in Table 3, in addition to the case of Table 1, the internal state i (k) is 0 and the output of the comparator 17 is 1, or the internal state i
This is the case where (k) is 0 and the output of the comparator 37 is 1. FIG. 4 is another time chart schematically showing the internal signal, threshold value, step size, and internal state of the step size control circuit 51.

FIG. 4 shows a case where the state transition table of Table 3 is used in the step size control circuit 51 of the present invention.
The waveforms of the internal signals p (k) and q (k) and the threshold values θb and θ
c, and the waveforms of the step sizes α (k) and β (k) are schematically shown. The difference between FIG. 4 and FIG.
Only the waveform of β (k) in the present invention. As can be seen from FIG. 4, when the state transition according to Table 3 is performed in the step size control circuit 51 of the present invention, the target signal can be detected with higher accuracy than when the step size control circuit 50 of the third conventional method is used. You.

In the case where Table 3 is used, similarly to the case where Table 1 is used, the internal state i () indicating whether the next state transition is due to the generation of the target signal or the stop of the target signal. By using k), highly accurate target signal detection is possible, and the same effect as in the case of using Table 1 is produced. As described above, the index regarding the presence or absence of the target signal is 2
We have used the code cross-correlation of the microphone signals,
Different indices can be used. FIG. 5 is a block diagram of an adaptive array device according to the second embodiment of the present invention.

The only difference between the first embodiment of the present invention shown in FIG. 1 and the present embodiment is that the step size control circuit 51 is replaced by a step size control circuit 53 in FIG. The difference between the step size control circuit 51 and the step size control circuit 53 is that the code cross correlation calculation circuit 20 is replaced by the cross correlation calculation circuit 21 and the code cross correlation calculation circuit 40 is replaced by the cross correlation calculation circuit 41. It is. Hereinafter, only this difference will be described.

The difference between the code cross-correlation calculation circuit 20 and the cross-correlation calculation circuit 21 is that the code determination circuits 13 and 14 in the code cross-correlation calculation circuit 20 are removed, and the outputs of the band-pass circuits 11 and 12 are sent to the multiplication circuit 15. It is only directly connected. The output of the low-pass circuit 16, that is, the output of the cross-correlation calculation circuit 21, is the cross-correlation of the signals from the two microphones _1j and _{1j + 1} . Therefore, the cross-correlation is an index for detecting the target signal. The difference between the code cross-correlation calculation circuit 40 and the cross-correlation calculation circuit 41 is also that the code determination circuits 33 and 34 have been removed, so that the description is omitted.

The cross-correlation increases when the target signal is present, and decreases when the target signal stops.
This property is similar to code cross-correlation. Therefore,
The step size control circuit 53 according to the second embodiment of the present invention has the same properties as the step size control circuit 51 according to the first embodiment of the present invention. As in the first embodiment,
This has the effect of reducing breathing noise in the output signal and increasing the quality of the output signal.

There are various definitions of the cross-correlation.
According to the definition, the cross-correlation calculation circuit can have various configurations other than the configuration shown in FIG. In the above description, a code cross-correlation is used in addition to a cross-correlation as an index for detecting a target signal. However, a cross-correlation coefficient can be used. Alternatively, an estimated value of the ratio between the target signal power and the noise power can be used as another index for detecting the target signal.

In the above, both the step size α (k) of the blocking matrix 4 and the step size β (k) of the multi-input canceller 3 are controlled by the method of the present invention. (K) only or step size β of multi-input canceller 3
It is also possible to control only (k) by the method of the present invention. In this case, the control of the step size not controlled by the method of the present invention can be controlled by a conventional method.

The case where the NLMS algorithm is used as the adaptive algorithm for updating the tap coefficients of each adaptive filter 207m of the multi-input canceller has been described above.
The same method can be applied to the case where another adaptive algorithm is used, such as an algorithm, a projection algorithm, a gradient method, a least squares method, a block adaptive algorithm, and an adaptive algorithm using an orthogonal transform.

When the NLMS algorithm is used,
Although only the step size is controlled, when other adaptive algorithms are used, parameters other than the step size can also be controlled. For example, when the RLS algorithm is used, not only the step size but also a constant called a forgetting constant can be controlled by the same method.

The case where a leak adaptive filter is used as each adaptive filter 207m of the multi-input canceller has been described above. In addition to the leak adaptive filter, a tap coefficient constrained adaptive filter, a norm constrained adaptive filter, or the like can be used. is there. As described above, the case where the FIR filter is used as the filter structure of each adaptive filter 207m of the multi-input canceller has been described.
It is possible to use an IIR filter, a lattice filter, or the like.

Also, the case has been described where the NLMS algorithm is used as an adaptive algorithm for updating the tap coefficients of each adaptive filter 105m of the blocking matrix.
Algorithm, RLS algorithm, projection algorithm, gradient method, least squares method, block adaptation algorithm,
The same method can be applied to a case where another adaptive algorithm such as an adaptive algorithm using an orthogonal transform is used.

When the NLMS algorithm is used,
Although only the step size is controlled, when other adaptive algorithms are used, parameters other than the step size can also be controlled. For example, when the RLS algorithm is used, not only the step size but also a constant called a forgetting constant can be controlled by the same method.

The case where a leak adaptive filter is used as each adaptive filter 105m of the blocking matrix has been described above. In addition to the leak adaptive filter, a tap coefficient constrained adaptive filter, a norm constrained adaptive filter, or the like can be used. . The case where the FIR filter is used as the filter structure of each adaptive filter 105m of the blocking matrix has been described above. However, other than the FIR, an IIR filter, a lattice filter, or the like can be used.

As described above, the configuration of the fixed beam former 2 and the fixed beam former 103 has been described using the delay-and-sum beam former, but various other configurations such as a filter-and-sum beam former can be used. As described above, as the configuration of the adaptive array,
Has been described as an example. In addition, as an adaptive array, the configuration of Reference 1, the configuration of Reference 2, the configuration of Reference 4,
The same method can be applied to the case where the configuration of Reference 5 and the configuration of Reference 6 are used.

The adaptive array control method and the adaptive array device of the present invention can be applied to the frost beamformer of Reference 4. In Literature 4, an adaptive algorithm with linear constraint is used for updating the tap coefficients. The step size and forgetting constant in this adaptive algorithm are
What is necessary is just to control by the output of the step size control circuit which has the property similar to what was demonstrated so far, for example, the step size control circuit 51 in FIG.

In the above description, the case where the target signal source exists in the direction near the straight line orthogonal to the line segment where the microphone array is arranged has been described, but the straight line orthogonal to the line segment where the target signal source is arranged with the microphone array is described. The present invention can also be applied to a case where it does not exist in the vicinity direction. Bandpass circuit 11
A means for adjusting the phase may be inserted into the output of each microphone 1m so that the phases of the target signal components substantially coincide with each other in the input signals of and.

The case where the microphone arrays are arranged at equal intervals on the line segment has been described above. However, the microphone arrays are not arranged at equal intervals, are arranged on a curved line, or are arranged on a plane or a curved surface. , Or even in a three-dimensional arrangement, a configuration based on the same principle is possible. Although the case where the sensor array is used as the fixed beam former has been described, the same method can be applied to a case where the microphone itself has directivity. Furthermore, although a microphone has been described as a sensor, it is possible to use an ultrasonic sensor, a sonar receiver, an antenna, or another sensor other than the microphone.

[0099]

As described above, according to the adaptive array control method and the adaptive array apparatus of the present invention, the detection of the target signal becomes accurate, and the control of the step size becomes more accurate. For this reason, the effect of reducing breathing noise in the final output and the effect of increasing the quality of the target signal occur.

[Brief description of the drawings]

FIG. 1 is a block diagram of an adaptive array device according to a first embodiment of the present invention.

FIG. 2 is a time chart schematically showing an internal signal, a threshold, a step size, and an internal state of a step size control circuit.

FIG. 3 is another time chart schematically showing an internal signal, a threshold, a step size, and an internal state of the step size control circuit.

FIG. 4 is another time chart schematically showing an internal signal, a threshold, a step size, and an internal state of the step size control circuit.

FIG. 5 is a block diagram of an adaptive array device according to a second embodiment of the present invention.

FIG. 6 is a diagram showing a configuration of a first conventional method.

FIG. 7 is a block diagram illustrating an example of a configuration of a blocking matrix.

FIG. 8 is a block diagram illustrating an example of a configuration of a multi-input canceller.

FIG. 9 is a block diagram showing a configuration of a second conventional method.

[Explanation of symbols]

 10-1M-1 Microphone 2 Fixed beamformer 3 Multiple input canceller 4 Blocking matrix 8 Step size input terminal 9 Step size input terminal 10 Output terminal 11, 12, 31, 32 Bandpass circuit 13, 14, 33, 34 Code determination circuit 15, 35 Multiplier 16, 36 Low-pass circuit 17, 37 Comparator 19, 39 Threshold generation circuit 20, 40 Code cross-correlation calculation circuit 21, 41 Cross-correlation calculation circuit 50, 51, 53 Step size control circuit 60 State circuit 490-49M-1 Multiplier 69 Inverting circuit 79 Multiplier 1010-101M-1 Input terminal 102-102M-1 Output terminal 103 Fixed beamformer 1040-104M-1 Delay unit 1050-105M-1 Adaptive filter 1060-106M-1 Subtractor 2010-201M- Reference Signs List 1 input terminal 202 input terminal 2070-207M-1 adaptive filter 208 adder 209 subtractor 211 delay unit 212 multiplier 214 multiplier

Claims (10)

[Claims] 1. Using a plurality of sensors and an adaptive filter,
An adaptive array control method for receiving only a specific signal source from a plurality of signal sources as a target signal source, wherein a signal whose phase is adjusted to match the phase of a target signal component in two sensor output signals And the result of comparing the first cross-correlation calculated with the first time constant with the first threshold value, and the result of comparing the second cross-correlation calculated with the second time constant with the second threshold value An adaptive array control method, characterized in that an output of the state circuit which receives the input is a step size of an adaptive algorithm in the adaptive filter. 2. Using a plurality of sensors and an adaptive filter,
An adaptive array control method for receiving only a specific signal source from a plurality of signal sources as a target signal source, wherein a signal whose phase is adjusted to match the phase of a target signal component in two sensor output signals From the result of comparing the first code cross-correlation calculated with the first time constant with the first threshold value, and comparing the second code cross-correlation calculated with the second time constant with the second threshold value A control method of an adaptive array, wherein an output of the state circuit which receives the result as an input is a step size of an adaptive algorithm in the adaptive filter. 3. Using a plurality of sensors and an adaptive filter,
An adaptive array control method for receiving only a specific signal source from a plurality of signal sources as a target signal source, wherein a signal whose phase is adjusted to match the phase of a target signal component in two sensor output signals From the result of comparing the first cross-correlation coefficient calculated with the first time constant with the first threshold value, and the second cross-correlation coefficient calculated with the second time constant and the second threshold value Wherein the output of the state circuit, which receives the result of the comparison, is the step size of the adaptive algorithm in the adaptive filter. 4. An internal state of the state circuit, wherein the next state transition is a state transition due to a target signal stop,
4. The control method of an adaptive array according to claim 1, wherein the state circuit indicates whether the state transition is caused by generation of a target signal, and the control of the step size by the state circuit differs depending on the internal state. 5. An internal state of the state circuit, wherein the first cross-correlation, the first code cross-correlation, or the first cross-correlation coefficient is larger than the first threshold, and When the cross-correlation or the second code cross-correlation or the second cross-correlation coefficient is larger than the second threshold, the state is changed to indicate that the next state transition is a state transition due to a target signal stop. Making a transition, wherein the first cross-correlation or the first code cross-correlation or the first cross-correlation coefficient is smaller than the first threshold, and the second cross-correlation or the second code cross-correlation When the correlation or the second cross-correlation coefficient is smaller than the second threshold, a state transition is performed so as to indicate that the next state transition is a state transition due to generation of a target signal. 5. The method for controlling an adaptive array according to claim 4. 6. A method comprising: a plurality of sensors and an adaptive filter.
In an adaptive array for receiving only a specific signal source from among a plurality of signal sources as a target signal source, a signal whose phase has been adjusted so that the phases of components of the target signal in the two sensor output signals are matched with each other is first. And a result obtained by comparing the first cross-correlation calculated with the time constant of the first time with the first threshold value, and a result obtained by comparing the second cross-correlation calculated with the second time constant with the second threshold value with the input The output of the state circuit to be performed is set as a step size of an adaptive algorithm in the adaptive filter. 7. A method comprising: a plurality of sensors and an adaptive filter;
In an adaptive array for receiving only a specific signal source from among a plurality of signal sources as a target signal source, a signal whose phase has been adjusted so that the phases of components of the target signal in the two sensor output signals are matched with each other is first. The result of comparing the first code cross-correlation calculated with the time constant of the first time with the first threshold value, and the result of comparing the second code cross-correlation calculated with the second time constant with the second threshold value The output of the state circuit as input is
A step size of an adaptive algorithm in the adaptive filter. 8. A system comprising a plurality of sensors and an adaptive filter,
In an adaptive array for receiving only a specific signal source from among a plurality of signal sources as a target signal source, a signal whose phase has been adjusted so that the phases of components of the target signal in the two sensor output signals are matched with each other is first. The result of comparing the first cross-correlation coefficient calculated with the time constant of the first time with the first threshold value, and the result of comparing the second cross-correlation coefficient calculated with the second time constant with the second threshold value The output of the state circuit which receives the input and the output is set as the step size of the adaptive algorithm in the adaptive filter. 9. The internal state of the state circuit indicates whether the next state transition is a state transition due to a stop of a target signal or a state transition due to generation of a target signal, and control of a step size by the state circuit. 9. The adaptive array device according to claim 6, wherein the value varies depending on the internal state. 10. An internal state of the state circuit, wherein the first cross-correlation, the first code cross-correlation, or the first cross-correlation coefficient is larger than the first threshold, and When the cross-correlation or the second code cross-correlation or the second cross-correlation coefficient is larger than the second threshold, the state is changed to indicate that the next state transition is a state transition due to a target signal stop. Making a transition, wherein the first cross-correlation or the first code cross-correlation or the first cross-correlation coefficient is smaller than the first threshold, and the second cross-correlation or the second code cross-correlation When the correlation or the second cross-correlation coefficient is smaller than the second threshold, a state transition is performed so as to indicate that the next state transition is a state transition due to generation of a target signal. 10. The adaptive array device according to 9.