JP3183128B2 - Signal to noise power ratio improving apparatus and signal to noise power ratio improving method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus and a method for improving a signal-to-noise power ratio of a periodic signal including noise whose amplitude distribution is Gaussian (Gaussian noise).

[0002]

2. Description of the Related Art FIG.
5091890, ■ Method of Extracting Target Range And
Doppler Information From a Doppler Spread Signal ■
FIG. 2 is a configuration diagram of a signal-to-noise power ratio improvement device using a fourth-order cumulant described in FIG. In FIG. 10, reference numeral 1 denotes a shift register, 2 denotes a fourth-order cumulant calculation circuit, 3 denotes a window function circuit, and 4 denotes a discrete Fourier transform.
Here, a fast Fourier transform circuit is used. Numeral 5 is for performing envelope detection. Here, a square detection circuit is used. Reference numeral 6 denotes a signal-to-noise power ratio calculation circuit.

A fourth-order cumulant is a high-order statistic (High
er-Order Statistics), whose definition and characteristics are described, for example, in CLNikias and JMMende
l, ■ Signal Processing with Higher-Order Spectra
■ 、 IEEE Signal ProcessingMagazine, pp10-37, July 19
93. Therefore, only the definition and characteristics of the quaternary cumulant used in the present invention will be briefly described below.

A fourth-order cumulant c ₄ ^x for a signal having an average of zero is defined by a function of three time lags (t1, t2, t3) as shown in equation (1).

[0005]

(Equation 1)

Here, m ₄ ^x and m ₂ ^x represent a fourth moment function and a second moment function (autocorrelation function), respectively, and are represented by the following equations.

[0007]

(Equation 2)

In Equations (1) to (8), the conjugate complex number (*) is obtained because the values of four input signals (x (n), x (n + t)
1), x (n + t2), x (n + t3)) as long as there are two locations. When the input is a real signal, it is not necessary to take a conjugate complex number (*).

[0009] The high-order cumulant has the following features. (1) Cumulants of third or higher order for a signal whose amplitude distribution is Gaussian distribution are theoretically zero. (2) If the discrete signal x (n) and the discrete signal y (n) are independent, there is a linear relationship represented by the following equation.

[0010]

(Equation 3)

(3) An input signal is a complex sine wave signal as shown in equation (10) (A is amplitude, φ is initial phase, ω is angular frequency)
Is a complex sine wave of a function of t1, t2, and t3 having the same frequency component ω as shown in Expression (11).

[0012]

(Equation 4)

Equation (11) is, for example, t1 = t, t2 = t3 = 0
By doing so, the variables can be made one-dimensional as shown in Expression (12).

[0014]

(Equation 5)

Due to these characteristics, a signal obtained by adding a Gaussian noise independent of a desired signal to a certain desired signal,
When the fourth-order cumulant is obtained, the Gaussian noise component becomes zero theoretically, only the fourth-order cumulant for the desired signal remains, and a high signal-to-noise power ratio improvement effect (theoretical infinity) is obtained. In particular, when the desired signal is a periodic signal, information on its frequency component is also stored in the fourth cumulant.

Next, the operation of the conventional signal-to-noise power ratio improving apparatus will be described with reference to FIG. For the sake of simplicity, here, the input signal x (n) is converted into a complex sine wave signal having an amplitude A, an angular frequency ω, and an initial phase φ as shown in Expression (13). With zero mean and variance σ ²
The signal to which the Gaussian noise g (n) is added is a discrete signal sampled at a certain time interval.

[0017]

(Equation 6)

The signal-to-noise power ratio SNRin of the input signal, which is represented by the equation (13), is represented by the equation (14).

[0019]

(Equation 7)

The input signal x (n) is sequentially input to the shift register 1. The shift register 1 holds the input signals of N samples and simultaneously inputs them to the fourth cumulant calculation circuit 2. In the fourth-order cumulant calculation circuit 2, in the fourth-order cumulant function of the three variables t1, t2, and t3 shown in the equation (1), for example, t1 = t, t2 = t3 = 0 (the aforementioned United States Patent 5091890) Then, t1 = 0, t2 = t3 =
and t. ), And outputs a maximum of 2N-1 fourth-order cumulants obtained by changing t from-(N-1) to (N-1) to the window function circuit 3. However, as shown in equation (11),
Since the amplitude of the fourth-order cumulant is the fourth power of the amplitude of the input signal, a fourth root of the amplitude is given to the window function circuit 3.
Output the next cumulant. The window function circuit 3 multiplies the weight of the window function in order to reduce the effect of the finite number of fourth-order cumulants obtained by the fourth-order cumulant calculation circuit 2 in the fast Fourier transform circuit 4, and multiplies the result by , To the fast Fourier transform circuit 4. The fast Fourier transform circuit 4 separates the output signal of the window function circuit 3 into components for each band determined by the number of fourth-order cumulants calculated by the fourth-order cumulant calculation circuit 2 and the sampling time of the input signal x (n), Output to the square detection circuit 5. Square detection circuit 5
Then, the power of the signal in each band is obtained and output to the signal-to-noise power ratio calculation circuit 6. In the signal-to-noise power ratio calculation circuit 6, when the frequency of the complex sine wave signal of the input signal x (n) is known or predicted, the band in which the frequency exists is calculated from the power value of the band in which the frequency exists. Then, a ratio of a value obtained by subtracting the average value of the powers other than the frequency and the average value of the power other than the frequency band is calculated. If the frequency of the complex sine wave signal of the input signal x (n) is not known, a value obtained by subtracting the average value of the power of the other bands from the largest power value and a band other than the band having the largest power value And the ratio of the average power to the average value.

[0021]

In the signal-to-noise power ratio improving apparatus as described above, since the number of samples of the input signal is finite, the fourth-order cumulant for the Gaussian noise is reduced in the fourth-order cumulant calculation circuit 2. It does not become completely zero, and a remaining signal is generated. As a result, there is a problem that the signal-to-noise power ratio improvement effect is reduced.

The present invention has been made to solve such a problem, and the fourth-order cumulant for Gaussian noise is not completely zero, and even if there is a signal that remains undisturbed,
A signal-to-noise power ratio improvement device and a signal-to-noise power ratio improvement method capable of obtaining a higher signal-to-noise power ratio improvement effect than a conventional device by further reducing the unerased signal are proposed.

[0023]

According to a first aspect of the present invention, there is provided a signal-to-noise power ratio improving apparatus for suppressing a Gaussian noise component of an input signal by sampling an input signal. For the input signal capturing means for generating a set of input signals, and for the same set of input signals from the input signal capturing means, one of the three variables of the fourth-order cumulant function is a variable and the other two are variables. The above parameters are varied for each of a plurality of different two parameter combinations in which the initial parameters of the fourth order cumulant function for the desired signal in the input signal are all the same. A fourth cumulant calculation processing means for obtaining a set of fourth cumulants by means of
An addition processing unit is provided for obtaining an average of the next cumulant set by adding the same variables.

According to a second aspect of the present invention, there is provided a signal-to-noise power ratio improving method for suppressing a Gaussian noise component of an input signal. For the same set of input signals to be generated, one of the three variables of the fourth-order cumulant function is a variable, and the other two are parameters, respectively. For each of a plurality of different combinations of two parameters in which the initial phase of the second order cumulant function is all the same, the above variables are changed to obtain a set of fourth order cumulants. Are added to each of the same variables, and the average is obtained.

According to a third aspect of the present invention, there is provided a signal to noise power ratio improving apparatus for suppressing a Gaussian noise component of an input signal. For the input signal capturing means to be generated and the same set of input signals from the input signal capturing means, one of the three variables of the fourth-order cumulant function is used as a variable and the other two are used as parameters, respectively. For each of the combinations of the two different parameters, the above variables are changed to obtain a fourth order cumulant set, and the fourth order cumulant set from the fourth order cumulant calculation processing means Function shift processing means for giving each fourth order cumulant a function shift of an amount obtained based on each set of the above two parameters. When, in which an adding processing means for obtaining an average by adding processing the output for each same variable from the function shift processing means.

According to a fourth aspect of the present invention, there is provided a signal-to-noise power ratio improving method for suppressing a Gaussian noise component of an input signal. For the same set of input signals to be generated, one of the three variables of the fourth-order cumulant function is a variable, and the other two are parameters, respectively. For each of a plurality of different combinations of two parameters, The above variables are changed to obtain a set of fourth-order cumulants, and for each of the sets of fourth-order cumulants, each fourth-order cumulant is given a function shift of an amount obtained based on the two parameters of each set, and The average is obtained by performing addition processing for each variable.

According to a fifth aspect of the present invention, there is provided a signal to noise power ratio improving apparatus for suppressing a Gaussian noise component of an input signal. For the input signal capturing means to be generated and the same set of input signals from the input signal capturing means, one of the three variables of the fourth-order cumulant function is used as a variable and the other two are used as parameters, respectively. For each of the combinations of the two different parameters, the above variables are changed to obtain a fourth order cumulant set, and the fourth order cumulant set from the fourth order cumulant calculation processing means Phase correction for each fourth-order cumulant based on the two parameters of each set and the frequency information of the desired signal in the input signal A phase compensation processing means for, in which an adding processing means for obtaining an average by adding processing the output from said phase processing means for each same variable.

According to a sixth aspect of the present invention, there is provided a signal-to-noise power ratio improving method for suppressing a Gaussian noise component of an input signal. For the same set of input signals to be generated, one of the three variables of the fourth-order cumulant function is a variable, and the other two are parameters, respectively. For each of a plurality of different combinations of two parameters, By changing the variables, a set of fourth-order cumulants is obtained, and for each set of the fourth-order cumulants, each fourth-order cumulant is based on the two parameters of each set and frequency information of a desired signal in the input signal. In this case, phase correction is performed, and addition processing is performed for each of the same variables to obtain an average.

According to a seventh aspect of the present invention, there is provided the signal-to-noise power ratio improving apparatus according to the first, third or fifth aspect, wherein the addition processing means is replaced with the addition processing means. Fourier transform processing means for converting each set of inputs into a frequency component and outputting the same, and frequency domain addition processing means for adding the output from the Fourier transform processing means for each same band and obtaining the average thereof. It is a thing.

According to an eighth aspect of the present invention, there is provided the signal-to-noise power ratio improving method according to the second, fourth or sixth aspect of the present invention, wherein an addition process is performed for each same variable to obtain an average thereof. Instead of the procedure, the frequency component is converted into a frequency component by Fourier transform processing, the addition processing is performed for each same band, and the average is obtained.

According to a ninth aspect of the present invention, there is provided a signal to noise power ratio improving apparatus for suppressing a Gaussian noise component of an input signal. For the input signal capturing means to be generated and the same set of input signals from the input signal capturing means, one of the three variables of the fourth-order cumulant function is used as a variable and the other two are used as parameters, respectively. For each of the combinations of the two different parameters, the above variables are changed to obtain a fourth order cumulant set, and the fourth order cumulant set from the fourth order cumulant calculation processing means A Fourier transform processing means for converting each fourth-order cumulant into a frequency component and outputting the frequency component; And envelope detection means for envelope detection processing power, in which an adding processing means for obtaining an average by adding processing the output from the envelope detection processing means for each same band.

According to a tenth aspect of the present invention, there is provided a signal-to-noise power ratio improving method for suppressing a Gaussian noise component of an input signal. For the same set of input signals to be generated, one of the three variables of the fourth-order cumulant function is a variable, and the other two are parameters, respectively. For each of a plurality of different combinations of two parameters, , By changing the above variables, to obtain a set of fourth-order cumulants, and for each of the set of fourth-order cumulants,
Each fourth-order cumulant is converted into a frequency component by a Fourier transform process, an envelope detection process is performed, an addition process is performed for each same band, and an average is obtained.

[0033]

BEST MODE FOR CARRYING OUT THE INVENTION

Embodiment 1 FIG. FIG. 1 is a block diagram of a signal-to-noise power ratio improving apparatus according to a first embodiment of the present invention. The output of a shift register 1 is input to M fourth-order cumulant calculation circuits 2, and the output is added to an addition circuit. 7 is added. The configuration after the addition circuit 7 is the same as that of the conventional example of FIG. Further, the signal-to-noise power ratio calculation circuit 6 is merely connected for evaluating the improvement result. Here, an embodiment of the signal-to-noise power ratio improving method is also described.

The operation of the above signal-to-noise power ratio improving apparatus will be described with reference to FIG. Here, similarly to the conventional example, the input signal x (n) is converted into a complex sine wave signal having an amplitude A, an angular frequency ω, and an initial phase φ as shown in Expression (13), independently of the complex sine wave signal. It is assumed that Gaussian noise g (n) with zero mean and variance σ ² is added. In FIG. 1, up to the shift register 1 is the same as the conventional one. That is, the input signal x (n) sampled at certain time intervals is sequentially input to the shift register 1. The shift register 1 holds an input signal of N samples. In the conventional example of FIG.
While the input signals of N samples held for the four fourth-order cumulant calculation circuits 2 are output, in the present embodiment, the same input signals of N samples are supplied to the M fourth-order cumulant calculation circuits 2. Is output. In each fourth-order cumulant calculation circuit 2, three variables t1, t2,
Using one of t3 as a variable t and the other two variables as parameters, t is changed from-(N-1) to (N-1), and M sets of up to 2N-1 quartic Seek cumulant. That is, for example, if t1 is a variable t, the other two variables t2 and t3 are
For each quaternary cumulant calculation circuit 2, a quaternary cumulant is obtained as a constant of a different combination. that time,
t2 in the i-th (1 ≦ i ≦ M) fourth order cumulant calculation circuit 2
Let ai be the value of t3 and bi be the value of t3. For each of the fourth cumulant calculation circuits 2, ai and bi are set so that C is a common arbitrary constant and satisfies the relationship shown in equation (15). In addition,
The relationship of equation (15) holds when x (n + t2) and x (n + t3) are obtained as complex conjugates, as shown in equation (1), where t1 is a variable t and 4 cumulant functions are shown in equation (1). Relationship, t2 or t3
Is a variable t, or when a fourth-order cumulant function is obtained by setting another position of the equation (1) as a complex conjugate, another relationship is obtained.

[0035]

(Equation 8)

For example, when C = 4, ai and bi are [a1 =
1, b1 = 3], [a2 = 0, b2 = 4], [a3 = 2, b3 = 2], [a4 =-
1, b4 = 5], ……. However, [a1 = 1, b
When [1 = 3] is used, a value obtained by exchanging a1 and b1 thereafter, that is, [ai = 3, bi = 1] is not used.

By setting ai and bi as described above, a fourth-order cumulant function when only a complex sine wave signal is used as an input signal (a complex sine wave having the same frequency as the input signal)
Can be made constant irrespective of the values of ai and bi, as shown in equation (16).

[0038]

(Equation 9)

That is, when only the complex sine wave signal is used as the input signal, the values of the fourth order cumulants for the same t, which are obtained by the M fourth order cumulant calculation circuits 2, can be made the same.

On the other hand, Gaussian noise g with zero mean and variance σ ²
When only (n) is an input signal, M sets of fourth-order cumulants determined by the M fourth-order cumulant calculation circuits 2 (theoretically, the fourth-order cumulant for Gaussian noise is zero, as described above) However, the remaining signal is generated because the number of samples of the input signal is finite.) If ai and bi are constants of different combinations for each fourth-order cumulant calculation circuit 2, 4 for the same t The value of the next cumulant differs for each fourth-order cumulant calculation circuit 2, and the average of the sum of those values approaches zero.

The M sets of fourth-order cumulants thus obtained are output to the adder circuit 7. However, since the amplitude of the fourth-order cumulant is the fourth power of the amplitude of the input signal as shown in the equation (11), the fourth-order cumulant having the fourth root of the amplitude is supplied to the adding circuit 7. Output. The adder circuit 7 adds M sets of fourth-order cumulants for each t, and takes an average for each t. By performing such processing, it is possible to make only the fourth-order cumulant for the Gaussian noise component close to zero without changing the value of the fourth-order cumulant for the complex sine wave signal component. That is, the signal-to-noise power ratio can be improved.

The output signal of the adding circuit 7 is input to the window function circuit 3. In the window function circuit 3, in order to reduce the influence of the finite number of output signals of the adding circuit 7 in the fast Fourier transform circuit 4, weights are multiplied by a window function,
The result is output to the fast Fourier transform circuit 4. In the fast Fourier transform circuit 4, the output signal of the window function circuit 3 is
The signal is separated into components for each band determined by the number of output signals of the adder circuit 7 and the sampling time of the input signal x (n), and output to the square detection circuit 5. Thereafter, as in the conventional example, the square detection circuit 5
, The power of the signal in each band is obtained and output to the signal-to-noise power ratio calculation circuit 6. Signal to noise power ratio calculation circuit 6
Then, if the frequency of the complex sine wave signal of the input signal x (n) is known or predicted, the average value of the power outside the frequency band is calculated from the power value of the frequency band. The ratio between the subtracted value and the average value of the power other than the frequency band is calculated. If the frequency of the complex sine wave signal of the input signal x (n) is not known, a value obtained by subtracting the average value of the power of the other bands from the largest power value and a band other than the band having the largest power value And the ratio of the average power to the average value.

FIG. 2 shows an example of an output signal of the square-law detection circuit 5 in the conventional signal-to-noise-power ratio improving apparatus of FIG. However, the signal-to-noise power ratio SNRin of the input signal is -5 [dB], the number N of samples of the input signal is 32, and the normalized frequency of the input complex sine wave signal is 9/32. FIG. 3 shows an example of the output signal of the square detection circuit 5 when the same input signal as in FIG. 2 is input to the signal-to-noise power ratio improving apparatus of the embodiment shown in FIG. However, ten sets of fourth-order cumulants are obtained from one set of input signals. Where ai, bi
Is [a1 = 0, b1 = 0], [a2 = -1, b2 = 1]
[A3 = -2, b3 = 2], [a4 = -3, b4 = 3],…, [a10 = -9,
b10 = 9].

The conventional signal-to-noise power ratio calculation circuit 6 of FIG.
Signal-to-noise power ratio is 13.97 [dB],
The signal-to-noise power ratio in the signal-to-noise power ratio calculation circuit 6 of the first embodiment shown in FIG. 3 is 16.40 [dB], which means that the signal-to-noise power ratio improvement effect is improved by about 2.5 [dB]. I understand.

Embodiment 2 FIG. 4 is a configuration diagram of a signal-to-noise power ratio improving apparatus according to Embodiment 2 of the present invention.
The difference from the first embodiment is that a window function circuit and M fast Fourier transform circuits are provided after the M fourth-order cumulant calculation circuits 2 in the same manner as the fourth-order cumulant calculation circuit. The latter is that the window function circuit 3 and the fast Fourier transform circuit 4 are eliminated. Here, an embodiment of the signal-to-noise power ratio improving method is also described.

The operation of the above signal-to-noise power ratio improving apparatus will be described with reference to FIG. Here, similarly to the first embodiment, the input signal x (n) is converted into a complex sine wave signal having an amplitude A, an angular frequency ω, and an initial phase φ as shown in Expression (13). Gaussian noise g with zero mean and variance σ ²
(n) has been added. In FIG. 4, the configuration up to the fourth cumulant calculation circuit 2 is the same as that of the first embodiment. That is, the input signal x (n) sampled at certain time intervals is sequentially input to the shift register 1. The shift register 1 holds the input signals of N samples, and outputs the same input signals of N samples to the M fourth-order cumulant calculation circuits 2. In each fourth-order cumulant calculation circuit 2, three variables t1, t of the fourth-order cumulant function
2, one of t3 is a variable t, and the other two variables are parameters, and t is changed from-(N-1) to (N-1). Find the fourth cumulant. That is,
For example, if t1 is a variable t, the other two variables t2 and t3
Calculates a fourth-order cumulant as a constant of a different combination for each fourth-order cumulant calculation circuit 2. At this time, if the value of t2 in the i-th (1 ≦ i ≦ M) fourth-order cumulant calculation circuit 2 is ai and the value of t3 is bi, ai and bi are C Is an arbitrary constant,
The setting is made so as to satisfy the relationship shown in Expression (15).

In the first embodiment, the value is obtained in this manner.
While M sets of fourth-order cumulants are added for the same t, in the second embodiment, each fourth-order cumulant is converted into a frequency component and added for each same band. That is, the M sets of quaternary cumulants calculated by the M quaternary cumulant calculation circuits 2 are input to the corresponding window function circuits 3. However, as shown in equation (11), the amplitude of the fourth-order cumulant is the fourth power of the amplitude of the input signal. Output cumulant. In each window function circuit 3, the weight of the window function is reduced in order to reduce the effect of the finite number of fourth-order cumulants in the fast Fourier transform circuit 4 on the output signal of each fourth-order cumulant calculation circuit 2. And outputs the result to the corresponding fast Fourier transform circuit 4. Each fast Fourier transform circuit 4 converts the output signal of each window function circuit 3 into the number of each fourth-order cumulant and the input signal.
It is separated into components for each band determined by the sampling time of x (n), and is output to the addition circuit 7. The adder circuit 7 adds M signals for each band and calculates an average value for each band.

As described in the first embodiment, by setting the values of ai and bi so as to satisfy the relationship represented by the equation (15), the number of M signals when the input signal is only a complex sine wave signal is obtained. The values of the fourth-order cumulants for the same t obtained by the fourth-order cumulant calculation circuit 2 can be made the same. Therefore, when Fourier transform is performed on M sets of fourth-order cumulants obtained by the M fourth-order cumulant calculation circuits 2,
The values of the bands including the frequency components of the complex sine wave signal are all the same. On the other hand, Gaussian noise g with mean zero and variance σ ²
When only (n) is an input signal, ai and bi are set as constants of different combinations for each of the M quaternary cumulant calculation circuits 2 and M sets of quaternary cumulants are obtained. The values differ for each fourth-order cumulant, and the average of their sum approaches zero. Therefore, the value of each band when the M sets of fourth-order cumulants are subjected to Fourier transform also differs for each fourth-order cumulant, and the average of the sum of the values for each band approaches zero. Therefore, the input signal is a signal obtained by adding a Gaussian noise g (n) having a mean of zero and a variance of σ ^{2 to} the complex sine wave signal independently of the complex sine wave signal. Then, the M sets of fourth-order cumulants obtained under the conditions as in equation (15) are Fourier-transformed, added and averaged for each band, and the signal for the complex sine wave signal component does not change. Only the signal for the Gaussian noise component can be made close to zero, and as a result, the signal-to-noise power ratio can be improved.

The output signal of the adding circuit 7 is
Is input to The square detection circuit 5 obtains the power of the signal in each band and outputs it to the signal-to-noise power ratio calculation circuit 6.
In the signal-to-noise power ratio calculation circuit 6, when the frequency of the complex sine wave signal of the input signal x (n) is known or predicted, the frequency exists based on the power value of the frequency band. The ratio between the value obtained by subtracting the average value of the power outside the band and the average value of the power outside the band where the frequency exists is determined. If the frequency of the complex sine wave signal of the input signal x (n) is not known, a value obtained by subtracting the average value of the power of the other bands from the largest power value and a band other than the band having the largest power value And the ratio of the average power to the average value.

Embodiment 3 FIG. 5 is a configuration diagram of a signal-to-noise power ratio improving apparatus according to Embodiment 3 of the present invention.
The difference from the first embodiment is that M function shift circuits 8 are provided after M fourth-order cumulant calculation circuits. Here, an embodiment of the signal-to-noise power ratio improving method is also described.

The operation of the above signal-to-noise power ratio improving apparatus will be described with reference to FIG. Here, similarly to the first embodiment, the input signal x (n) is converted into a complex sine wave signal having an amplitude A, an angular frequency ω, and an initial phase φ as shown in Expression (13). Gaussian noise g with zero mean and variance σ ²
(n) has been added. In FIG. 5, M 4
The operation up to the input of the next cumulant calculation circuit 2 is the same as that of the first embodiment. That is, the input signal x (n) sampled at certain time intervals is sequentially input to the shift register 1. The shift register 1 holds the input signals of N samples, and outputs the same input signals of N samples to the M fourth-order cumulant calculation circuits 2. In each of the fourth-order cumulant calculation circuits 2, one of the three variables t1, t2, and t3 of the fourth-order cumulant function is set as a variable t, and the other two variables are set as parameters. N-1) to obtain M sets of up to 2N-1 quaternary cumulants. That is, for example, if t1 is a variable t, the other two variables t
2, t3 is to obtain the fourth cumulant as a constant of a different combination for each fourth cumulant calculation circuit 2. At this time, in the first embodiment, the values ai, bi of t2 and t3
(I = 1, 2,..., M) is restricted to a constant of a different combination that satisfies the relationship shown in Expression (15) for each fourth-order cumulant calculation circuit 2. On the other hand, in the present embodiment, ai and bi are calculated for each fourth-order cumulant calculation circuit 2,
What is necessary is just a constant of a different combination. However, as in the first embodiment, the values obtained by replacing the values of ai and bi are not used at the same time.

The M sets of fourth-order cumulants thus obtained are input to the corresponding function shift circuits 8. However, since the amplitude of the fourth-order cumulant is the fourth power of the amplitude of the input signal, as shown in the equation (11), the fourth-order root of the amplitude is given to each function shift circuit 8. Output cumulant. Each function shift circuit 8 shifts the fourth-order cumulant in the negative direction on the t-axis (ai + bi) with respect to the change of t obtained by each fourth-order cumulant calculation circuit 2. For example, when a fourth order cumulant is obtained as [a1 = 1, b1 = 3], the fourth order cumulant obtained for t = 10 is t = 10.
The fourth cumulant determined for 6. The direction and the amount of shift of the fourth cumulant are defined as x (n + t2) and x (n + t3), where t1 is a variable t and the 4 cumulant function is shown in equation (1). This relationship holds when calculated as a complex conjugate. When t2 or t3 is a variable t, or when a fourth-order cumulant function is calculated with another position of Equation (1) as a complex conjugate, the shift direction and shift amount are different. Becomes By performing such processing,
When only a complex sine wave signal is used as an input signal, the initial phase of a fourth-order cumulant function (a complex sine wave having the same frequency as the input signal) can be constant regardless of the values of ai and bi. In other words, when only the complex sine wave signal is used as the input signal, the number of the M fourth-order cumulant calculation circuits 2 is calculated.
All fourth order cumulant values for the same t can be the same. On the other hand, when only the Gaussian noise g (n) having zero mean and variance σ ² is used as the input signal, the value of the fourth-order cumulant for the same t is different for each fourth-order cumulant calculation circuit 2. The average of the sums approaches zero.

The output signals of the M sets of function shift circuits 8 thus obtained are output to the adder circuit 7. Adder circuit 7
So the same t your capital adds the output signal of the M sets of function shift circuit 8, taking the average of each t. By performing such a process, it is possible to make only the fourth-order cumulant for the Gaussian noise component close to zero without changing the value of the fourth-order cumulant for the complex sine wave signal component. That is, it is possible to make improved signal-to-noise power ratio.

In the subsequent processing, the output signal of the adding circuit 7 is input to the window function circuit 3 as in the first embodiment. The window function circuit 3 multiplies the weight by the window function in order to reduce the influence of the finite number of output signals of the adder circuit 7 in the fast Fourier transform circuit 4, and divides the result by the fast Fourier transform circuit 4. Output to The fast Fourier transform circuit 4 separates the output signal of the window function circuit 3 into components for each band determined by the number of output signals of the adder circuit 7 and the sampling time of the input signal x (n), and outputs it to the square detection circuit 5. I do. The square detection circuit 5 obtains the power of the signal in each band and outputs it to the signal-to-noise power ratio calculation circuit 6. In the signal-to-noise power ratio calculation circuit 6, when the frequency of the complex sine wave signal of the input signal x (n) is known or predicted, the frequency exists based on the power value of the frequency band. The ratio between the value obtained by subtracting the average value of the power outside the band and the average value of the power outside the band where the frequency exists is determined. If the frequency of the complex sine wave signal of the input signal x (n) is not known, a value obtained by subtracting the average value of the power of the other bands from the largest power value and a band other than the band having the largest power value And the ratio of the average power to the average value.

Embodiment 4 FIG. 6 is a configuration diagram of a signal-to-noise power ratio improving apparatus according to Embodiment 4 of the present invention.
The difference from the second embodiment described with reference to FIG. 4 is that M function shift circuits 8 are provided after M fourth-order cumulant calculation circuits 2. Here, an embodiment of the signal-to-noise power ratio improving method is also described.

The operation of the signal to noise power ratio improving apparatus will be described with reference to FIG. However, as in the second embodiment, the input signal x (n) is a complex sine wave signal having an amplitude A, an angular frequency ω, and an initial phase φ as shown in Expression (13), and the complex sine wave signal Gaussian noise g with independent zero mean and variance σ ²
(n) has been added. In FIG. 6, M 4
The operation up to the input of the next cumulant calculation circuit 2 is the same as that of the second embodiment. That is, the input signal x (n) sampled at certain time intervals is sequentially input to the shift register 1. The shift register 1 holds the input signals of N samples, and outputs the same input signals of N samples to the M fourth-order cumulant calculation circuits 2. In each of the fourth-order cumulant calculation circuits 2, one of the three variables t1, t2, and t3 of the fourth-order cumulant function is set as a variable t, and the other two variables are set as parameters. N-1) to obtain M sets of up to 2N-1 quaternary cumulants. That is, for example, if t1 is a variable t, the other two variables t
2, t3 is to obtain the fourth cumulant as a constant of a different combination for each fourth cumulant calculation circuit 2. At this time, in the second embodiment, the values ai, bi of t2 and t3
(I = 1, 2,..., M) is restricted to a constant of a different combination that satisfies the relationship shown in Expression (15) for each fourth-order cumulant calculation circuit 2. On the other hand, in the present embodiment, ai and bi are calculated for each fourth-order cumulant calculation circuit 2,
What is necessary is just a constant of a different combination. However, as in the second embodiment, the values obtained by replacing the values of ai and bi are not used at the same time.

The M sets of fourth-order cumulans determined as described above are input to the corresponding function shift circuits 8. However, since the amplitude of the fourth-order cumulant is the fourth power of the amplitude of the input signal, as shown in the equation (11), the fourth-order root of the amplitude is given to each function shift circuit 8. Output cumulant. Each function shift circuit 8 shifts the fourth-order cumulant in the negative direction on the t-axis (ai + bi) with respect to the change of t obtained by each fourth-order cumulant calculation circuit 2. The direction and the amount of shift of the fourth cumulant are defined as x (n + t2) and x (n + t3), where t1 is a variable t and the 4 cumulant function is shown in equation (1). This relationship holds when calculated as a complex conjugate. When t2 or t3 is a variable t, or when a fourth-order cumulant function is calculated with another position of Equation (1) as a complex conjugate, the shift direction and shift amount are different. Becomes By performing such processing, the initial phase of the fourth-order cumulant function (a complex sine wave having the same frequency as the input signal) when only the complex sine wave signal is used as the input signal is constant regardless of the values of ai and bi. Can be That is, when only the complex sine wave signal is used as the input signal, the values of the fourth order cumulants for the same t obtained by the M fourth order cumulant calculation circuits 2 can be all the same. On the other hand, when only the Gaussian noise g (n) having zero mean and variance σ ² is used as the input signal, the same t
The value of the fourth-order cumulant differs for each fourth-order cumulant calculation circuit 2, and the average of those values approaches zero.

The output signals of the M sets of function shift circuits 8 thus obtained are input to the corresponding window function circuits 3. In each window function circuit 3, the output signal of each function shift circuit 8 is applied to the output signal of the fast Fourier transform circuit 4.
In order to reduce the influence of the finite number of next cumulants, the weight of the window function is multiplied, and the result is output to the corresponding fast Fourier transform circuit 4. In the subsequent processing, as in the second embodiment, in each fast Fourier transform circuit 4, the output signal of each window function circuit 3 is converted to the number of output signals of the function shift circuit 8 and the input signal x (n). The signal is separated into components for each band determined by the sampling time, and output to the adding circuit 7. In the addition circuit 7, M
The signals are added, and an average value is obtained for each band. By performing such processing, the signal for the Gaussian noise component can be made close to zero without changing the signal for the complex sine wave signal component, and as a result, the signal-to-noise power ratio can be improved. . The output signal of the addition circuit 7 is input to the square detection circuit 5. The square detection circuit 5 obtains the power of the signal in each band and outputs it to the signal-to-noise power ratio calculation circuit 6. Signal to noise power ratio calculation circuit 6
Then, if the frequency of the complex sine wave signal of the input signal x (n) is known or predicted, the average value of the power outside the frequency band is calculated from the power value of the frequency band. The ratio between the subtracted value and the average value of the power other than the frequency band is calculated. If the frequency of the complex sine wave signal of the input signal x (n) is not known, a value obtained by subtracting the average value of the power of the other bands from the largest power value and a band other than the band having the largest power value And the ratio of the average power to the average value.

Embodiment 5 FIG. FIG. 7 is a configuration diagram of a signal-to-noise power ratio improving apparatus according to Embodiment 5 of the present invention.
The difference from the first embodiment is that M phase correction circuits 9 are provided after M fourth-order cumulant calculation circuits. Here, an embodiment of the signal-to-noise power ratio improving method is also described.

The operation of the above signal-to-noise power ratio improving apparatus will be described with reference to FIG. However, as in the first embodiment, the input signal x (n) is converted into a complex sine wave signal having an amplitude A, an angular frequency ω, and an initial phase φ as shown in Expression (13). Gaussian noise g with independent zero mean and variance σ ²
(n) has been added. In FIG. 7, M 4
The operation up to the input of the next cumulant calculation circuit 2 is the same as that of the first embodiment. That is, the input signal x (n) sampled at certain time intervals is sequentially input to the shift register 1. The shift register 1 holds the input signals of N samples, and outputs the same input signals of N samples to the M fourth-order cumulant calculation circuits 2. In each of the fourth-order cumulant calculation circuits 2, one of the three variables t1, t2, and t3 of the fourth-order cumulant function is set as a variable t, and the other two variables are set as parameters. N-1) to obtain M sets of up to 2N-1 quaternary cumulants. That is, for example, if t1 is a variable t, the other two variables t
2, t3 is to obtain the fourth cumulant as a constant of a different combination for each fourth cumulant calculation circuit 2. At this time, in the first embodiment, the values ai, bi of t2 and t3
(I = 1, 2,..., M) is restricted to a constant of a different combination that satisfies the relationship shown in Expression (15) for each fourth-order cumulant calculation circuit 2. In contrast, in the present embodiment,
ai and bi may be constants of different combinations for each fourth-order cumulant calculation circuit 2. However, Embodiment 1
Similarly to the above, the values obtained by replacing the values of ai and bi are not used at the same time.

The M sets of fourth-order cumulants thus obtained are input to the corresponding phase correction circuits 9. However, since the amplitude of the fourth-order cumulant is the fourth power of the amplitude of the input signal as shown in equation (11), the fourth-order cumulant having the fourth root of the amplitude is applied to each phase correction circuit 9. Output cumulant. Each phase correction circuit 9 multiplies each quadratic cumulant by the angular frequency ω of the complex sine wave signal obtained from the frequency information of the desired signal shown in FIG. 7 and the value obtained by equation (17) using ai and bi. And phase correction. Equation (17)
Is the relationship that holds when x (n + t2) and x (n + t3) are obtained as complex conjugates, as shown in equation (1), where t1 is the variable t and the four cumulant functions are shown in equation (1). When t3 is a variable t, or when a fourth-order cumulant function is obtained with another position of the expression (1) as a complex conjugate, another expression is obtained.

[0062]

(Equation 10)

By performing such processing, the initial phase of the fourth-order cumulant function (a complex sine wave having the same frequency as the input signal) when only the complex sine wave signal is used as the input signal is changed to the values of ai and bi. It can be kept constant. That is, when only a complex sine wave signal is used as an input signal, M
The values of the fourth-order cumulants for the same t obtained by the four fourth-order cumulant calculation circuits 2 can be all the same. On the other hand, Gaussian noise g (n) with mean zero and variance σ ²
If only the input signal is used, the value of the fourth-order cumulant for the same t differs for each fourth-order cumulant calculation circuit 2, and the average of the sum thereof approaches zero.

The output signals of the M sets of phase correction circuits 9 thus obtained are output to the addition circuit 7. The adder circuit 7 adds the output signals of the M sets of the phase correction circuits 9 for each of the same t, and takes an average for each of the t. By performing such a process, it is possible to make only the fourth-order cumulant for the Gaussian noise component close to zero without changing the value of the fourth-order cumulant for the complex sine wave signal component. That is, the signal-to-noise power ratio can be improved.

In the subsequent processing, the output signal of the adding circuit 7 is input to the window function circuit 3 as in the first embodiment. The window function circuit 3 multiplies the weight by the window function in order to reduce the influence of the finite number of output signals of the adder circuit 7 in the fast Fourier transform circuit 4, and divides the result by the fast Fourier transform circuit 4. Output to The fast Fourier transform circuit 4 separates the output signal of the window function circuit 3 into components for each band determined by the number of output signals of the adder circuit 7 and the sampling time of the input signal x (n), and outputs it to the square detection circuit 5. I do. Thereafter, similarly to the conventional example, the power of the signal in each band is obtained in the square detection circuit 5 and output to the signal-to-noise power ratio calculation circuit 6. In the signal-to-noise power ratio calculation circuit 6, when the frequency of the complex sine wave signal of the input signal x (n) is known or predicted, the frequency exists based on the power value of the frequency band. The ratio between the value obtained by subtracting the average value of the power outside the band and the average value of the power outside the band where the frequency exists is determined. If the frequency of the complex sine wave signal of the input signal x (n) is not known, a value obtained by subtracting the average value of the power of the other bands from the largest power value and a band other than the band having the largest power value And the ratio of the average power to the average value.

Embodiment 6 FIG. FIG. 8 is a configuration diagram of a signal-to-noise power ratio improving apparatus according to Embodiment 6 of the present invention.
The difference from the second embodiment is that M phase correction circuits 9 are provided after M fourth-order cumulant calculation circuits 2. Here, an embodiment of the signal-to-noise power ratio improving method is also described.

The operation of the above signal-to-noise power ratio improving apparatus will be described with reference to FIG. However, as in the second embodiment, the input signal x (n) is a complex sine wave signal having an amplitude A, an angular frequency ω, and an initial phase φ as shown in Expression (13), and the complex sine wave signal Gaussian noise g with independent zero mean and variance σ ²
(n) has been added. In FIG. 8, the operation up to the input of M phase correction circuits 9 is the same as that of the second embodiment. That is, the input signal x (n) sampled at certain time intervals is sequentially input to the shift register 1.
The shift register 1 holds the input signals of N samples, and outputs the same input signals of N samples to the M fourth-order cumulant calculation circuits 2. In each fourth-order cumulant calculation circuit 2, three variables t1, t2,
Using one of t3 as a variable t and the other two variables as parameters, t is changed from-(N-1) to (N-1), and M sets of up to 2N-1 quartic Seek cumulant. That is, for example, if t1 is a variable t, the other two variables t2 and t3 are
For each quaternary cumulant calculation circuit 2, a quaternary cumulant is obtained as a constant of a different combination. At that time, in the second embodiment, the values ai, bi (i = 1, 2,.
M) is given by equation (15) for each fourth-order cumulant calculation circuit 2.
There is a restriction that constants of different combinations satisfy the relationship shown by. On the other hand, in the present embodiment, ai, bi
May be a constant of a different combination for each fourth-order cumulant calculation circuit 2. However, as in the second embodiment, the values obtained by replacing the values of ai and bi are not used at the same time.

The M sets of fourth-order cumulants thus obtained are input to the corresponding phase correction circuits 9. However, since the amplitude of the fourth-order cumulant is the fourth power of the amplitude of the input signal as shown in equation (11), the fourth-order cumulant having the fourth root of the amplitude is applied to each phase correction circuit 9. Output cumulant. Each phase correction circuit 9 corrects the phase by multiplying each fourth-order cumulant by the value obtained by Expression (17) using the angular frequency ω of the complex sine wave signal and ai and bi. By performing such processing, the initial phase of the fourth-order cumulant function (a complex sine wave having the same frequency as the input signal) when only the complex sine wave signal is used as the input signal is constant regardless of the values of ai and bi. Can be That is, when only the complex sine wave signal is used as the input signal, the values of the fourth order cumulants for the same t obtained by the M fourth order cumulant calculation circuits 2 can be all the same. On the other hand, when only the Gaussian noise g (n) having zero mean and variance σ ² is used as the input signal, the value of the fourth order cumulant for the same t is 4
The next cumulant calculation circuit 2 differs, and the average of the sum thereof approaches zero.

The output signals of the M sets of phase correction circuits 9 obtained as described above are input to the corresponding window function circuits 3. Each window function circuit 3 multiplies the output signal of each phase correction circuit 9 by a weight of a window function in order to reduce the influence of the finite number of fourth-order cumulants in the fast Fourier transform circuit 4. , And outputs the result to the corresponding fast Fourier transform circuit 4. Subsequent processing is similar to that of the second embodiment described above. In each fast Fourier transform circuit 4, the output signal of each window function circuit 3 is
The signal is separated into components for each band determined by the number of output signals of the phase correction circuit 9 and the sampling time of the input signal x (n), and output to the addition circuit 7. The adder circuit 7 adds M signals for each band and calculates an average value for each band. By performing such processing, the signal for the Gaussian noise component can be made close to zero without changing the signal for the complex sine wave signal component, and as a result, the signal-to-noise power ratio can be improved. . The output signal of the addition circuit 7 is input to the square detection circuit 5. The square detection circuit 5 obtains the power of the signal in each band and outputs it to the signal-to-noise power ratio calculation circuit 6. In the signal-to-noise power ratio calculation circuit 6, when the frequency of the complex sine wave signal of the input signal x (n) is known or predicted, the frequency exists based on the power value of the frequency band. The ratio between the value obtained by subtracting the average value of the power outside the band and the average value of the power outside the band where the frequency exists is determined. If the frequency of the complex sine wave signal of the input signal x (n) is not known, a value obtained by subtracting the average value of the power of the other bands from the largest power value and a band other than the band having the largest power value And the ratio of the average power to the average value.

Embodiment 7 FIG. FIG. 9 is a configuration diagram of a signal-to-noise power ratio improving apparatus according to Embodiment 7 of the present invention.
The difference from the second embodiment is that the M square detection circuits 5 provided after the M fast Fourier transform circuits 4 and the square detection circuit 5 provided after the addition circuit 7 are eliminated.
Here, an embodiment of the signal-to-noise power ratio improving method is also described.

The operation of the signal to noise power ratio improving apparatus will be described with reference to FIG. However, as in the second embodiment, the input signal x (n) is a complex sine wave signal having an amplitude A, an angular frequency ω, and an initial phase φ as shown in Expression (13), and the complex sine wave signal Gaussian noise g with independent zero mean and variance σ ²
(n) has been added. In FIG. 9, M 4
The operation up to the output of the next cumulant calculation circuit 2 is the same as that of the second embodiment. That is, the input signal x (n) sampled at certain time intervals is sequentially input to the shift register 1. The shift register 1 holds the input signals of N samples, and outputs the same input signals of N samples to the M fourth-order cumulant calculation circuits 2. In each fourth-order cumulant calculation circuit 2, the third-order cumulant function 3
One of the variables t1, t2, and t3 is set as a variable t, and the other two variables are set as parameters, and t is changed from-(N-1) to (N-1) to obtain a maximum of 2N- Find one quaternary cumulant.
That is, for example, when t1 is a variable t, the other two variables t2 and t3 determine the fourth cumulant as constants of different combinations for each fourth cumulant calculation circuit 2. At this time, in the second embodiment, the values ai, bi of t2 and t3
(I = 1, 2,..., M) is restricted to a constant of a different combination that satisfies the relationship shown in Expression (15) for each fourth-order cumulant calculation circuit 2. On the other hand, in the present embodiment, ai and bi are calculated for each fourth-order cumulant calculation circuit 2,
What is necessary is just a constant of a different combination. However, as in the second embodiment, the values obtained by replacing the values of ai and bi are not used at the same time.

The M sets of fourth-order cumulants determined in this way are output to the corresponding window function circuits 3. However, as shown in equation (11), the amplitude of the fourth-order cumulant is the fourth power of the amplitude of the input signal. Output cumulant. In each window function circuit 3, in order to reduce the influence of the finite number of fourth-order cumulants in the fast Fourier transform circuit 4 on the fourth-order cumulants obtained in each fourth-order cumulant calculation circuit 2, The weight of the function is multiplied, and the result is output to the corresponding fast Fourier transform circuit 4. Each fast Fourier transform circuit 4 classifies the output signal of each window function circuit 3 into components for each band determined by the number of fourth-order cumulant function values and the sampling time of the input signal x (n). In the second embodiment,
While the output signal of each fast Fourier transform circuit 4 has been output to the addition circuit 7, in the present embodiment, the output signal of the fast Fourier transform circuit 4 is output to the corresponding square detection circuit 5, respectively. . Each square detection circuit 5 converts the signal of each band into a power value irrelevant to the initial phase, and outputs the result to the addition circuit 7. The adding circuit 7 adds M signals for each band and takes an average. By performing such a process, it is possible to reduce only the variance of the signal with respect to the Gaussian noise component without changing the signal with respect to the complex sine wave signal component, and as a result, as is apparent from the equation (14), , The signal to noise power ratio can be improved. The output signal of the addition circuit 7 is output to the signal-to-noise power ratio calculation circuit 6, and the signal-to-noise power ratio calculation circuit 6
If the frequency of the complex sine wave signal of the input signal x (n) is known or predicted, the average value of the power outside the frequency band is subtracted from the power value of the frequency band. The ratio between the value and the average value of the power other than the band in which the frequency exists is determined. If the frequency of the complex sine wave signal of the input signal x (n) is not known, a value obtained by subtracting the average value of the power of the other bands from the largest power value and a band other than the band having the largest power value And the ratio of the average power to the average value.

In the above embodiment, in the signal-to-noise power ratio improving apparatus and the signal-to-noise power ratio improving method, an example of an apparatus in which M fourth-order cumulant calculation circuits and the like are provided in parallel or M sets of parallel circuits Although the present invention has been described with reference to a simple processing example, in the present invention, it is not essential that there are M pieces of hardware, and that parallel processing is not essential, and that an apparatus and a method of sequentially and repeatedly processing software may be used. Needless to say.

[0074]

According to the first or second aspect of the present invention, for a periodic signal which is a desired signal in an input signal, one of three variables of the fourth-order cumulant function is a variable and the other two are variables. Are used as parameters, and the above two parameters are different from each other in such a manner that the initial phases of the fourth-order cumulant function for the desired signal in the input signal are all the same.
For each combination of the three parameters, when the above variables are changed to obtain a set of fourth-order cumulants, the initial phase of the fourth-order cumulant function becomes constant, so that the values of the fourth-order cumulants for the same variables are all the same. Can be. For Gaussian noise, the value of the fourth-order cumulant for the same variable differs for each combination of a plurality of different two parameters, and the average of those values approaches zero. Is obtained by adding each of the same variables and calculating the average, which has the effect of improving the signal-to-noise power ratio.

According to the third or fourth aspect of the present invention, for a periodic signal which is a desired signal in an input signal, for each set of fourth-order cumulants, each fourth-order cumulant has two parameters of each set. If the initial phase of the fourth-order cumulant function is not set so that the initial phases of the fourth-order cumulant function for the desired signal in the input signal are all the same, then the initial phase of the fourth-order cumulant function is given. Since it is constant, the values of the fourth-order cumulants for the same variable can all be the same. Therefore, similar to the first or second aspect of the present invention, if the addition processing is performed for each of the same variables for each set of the fourth-order cumulants and the average is obtained, there is an effect of improving the signal-to-noise power ratio.

According to the fifth or sixth aspect of the present invention, for a periodic signal that is a desired signal in the input signal, for each set of fourth-order cumulants, each of the four-order cumulants has two parameters of each set. When the phase correction is performed based on the frequency information of the desired signal in the input signal, the four parameters need not be set so that the initial phases of the fourth-order cumulant functions for the desired signal in the input signal are all the same. Since the initial phase of the second-order cumulant function is constant, the values of the fourth-order cumulants for the same variable can be all the same. Therefore, similar to the first or second aspect of the present invention, if the addition processing is performed for each of the same variables for each set of the fourth-order cumulants and the average is obtained, there is an effect of improving the signal-to-noise power ratio.

According to the seventh or eighth aspect of the present invention, instead of the means or procedure for adding and processing the average for each of the same variables and obtaining the average thereof, the frequency is converted into frequency components by the Fourier transform processing, and then the frequency is changed for each of the same bands. Even if the addition processing is performed in the area and the average is obtained, there is an effect of improving the signal-to-noise power ratio as in the first to sixth aspects of the present invention.

According to the ninth or tenth aspect of the present invention, 4
For each set of next cumulants, each fourth order cumulant is converted to a frequency component by Fourier transform processing,
When the envelope detection process is performed, the addition process is performed for each of the same bands, and the average is obtained. As for the periodic signal which is the desired signal in the input signal, two parameters are calculated by the fourth order cumulant function for the desired signal in the input signal. Can be treated as a value not related to the initial phase of the fourth-order cumulant function even if the initial phases are not all the same, and the variance of the signal with respect to the Gaussian noise component can be reduced. It has the effect of improving.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a signal-to-noise power ratio improving apparatus according to Embodiment 1 of the present invention.

FIG. 2 is a diagram illustrating an example of an output signal of a square detection circuit by a conventional signal-to-noise power ratio improving apparatus.

FIG. 3 is a diagram illustrating an example of an output signal of a square detection circuit by the signal-to-noise power ratio improving apparatus according to the first embodiment of the present invention;

FIG. 4 is a configuration diagram of a signal-to-noise power ratio improving apparatus according to Embodiment 2 of the present invention.

FIG. 5 is a configuration diagram of a signal-to-noise power ratio improving apparatus according to Embodiment 3 of the present invention.

FIG. 6 is a configuration diagram of a signal-to-noise power ratio improving apparatus according to Embodiment 4 of the present invention.

FIG. 7 is a configuration diagram of a signal-to-noise power ratio improving apparatus according to a fifth embodiment of the present invention.

FIG. 8 is a configuration diagram of a signal-to-noise power ratio improving apparatus according to Embodiment 6 of the present invention.

FIG. 9 is a configuration diagram of a signal-to-noise power ratio improving apparatus according to Embodiment 7 of the present invention.

FIG. 10 is a configuration diagram of a conventional signal-to-noise power ratio improving apparatus.

[Explanation of symbols]

1 shift register, 4th order cumulant calculation circuit,
3 window function circuit, 4 fast Fourier transform circuit, 5 square detection circuit, 6 signal-to-noise power ratio calculation circuit, 7 addition circuit, 8 function shift circuit, 9 phase correction circuit.

Continuation of the front page (56) References JP-A-6-21838 (JP, A) JP-A-5-14241 (JP, A) U.S. Pat. No. 5,370,053 (US, A) "The general source separation probe lemme" Lacome, J. L; Ga eta, M .; Spectrum Estimation and Modeling, 1990. , Fifth ASSP Workshopon, 1990 Page (s) 154-158 "Cumulant techniques for the estima- tion and cancelation of intersymbol interference. Nikias, C .; L. Spectrum Estimation and Modeling, 1988. , Fourth ASPS Works on, 1988 Page (s) 204-207 (58) Fields investigated (Int. Cl. ⁷ , DB name) G01S ⁷ /00-7/42 H03H 21/00 H04B 15/00

Claims (10)

(57) [Claims] 1. A signal-to-noise power ratio improving apparatus for suppressing a Gaussian noise component of an input signal, comprising: an input signal capturing means for sampling an input signal to generate a set of input signals; , One of the three variables of the fourth-order cumulant function as a variable and the other two as parameters, respectively, and the above two parameters are defined as 4 to the desired signal in the input signal.
Fourth-order cumulant calculation processing means for obtaining a set of fourth-order cumulants by changing the above variables for each of a plurality of different combinations of two parameters such that the initial phases of the second-order cumulant functions are all the same; A signal-to-noise-to-noise ratio improvement device, comprising: addition processing means for adding each of the sets of fourth-order cumulants from the fourth-order cumulant calculation processing means for each of the same variables to obtain an average thereof. 2. A signal-to-noise power ratio improving method for suppressing a Gaussian noise component of an input signal, wherein the input signal is sampled to generate a set of input signals. One of the three variables of the next cumulant function is a variable, and the other two are parameters, respectively, and the above two parameters are set to 4 for the desired signal in the input signal.
For each of a plurality of different combinations of two parameters in which the initial phase of the second order cumulant function is all the same, the above variables are changed to obtain a set of fourth order cumulants. A signal-to-noise power ratio improving method, wherein an addition process is performed for each of the same variables and an average thereof is obtained. 3. A signal-to-noise power ratio improving apparatus for suppressing a Gaussian noise component of an input signal, comprising: an input signal capturing means for sampling an input signal to generate a set of input signals; , One of the three variables of the fourth-order cumulant function is a variable, and the other two are parameters, respectively. For each combination of a plurality of different two parameters, Is changed to obtain a set of quaternary cumulants, and for each set of quaternary cumulants from the quaternary cumulant calculation processing means, each quaternary cumulant is assigned to each of the above two parameters. Function shift processing means for giving the amount of function shift obtained based on the same variable and the output from the function shift processing means for the same variable. A signal-to-noise-power ratio improving apparatus, comprising: addition processing means for obtaining an average by performing addition processing on the signals. 4. A signal-to-noise power ratio improving method for suppressing a Gaussian noise component of an input signal, wherein the input signal is sampled to generate a set of input signals. One of the three variables of the next cumulant function is a variable, and the other two are parameters, respectively. For each of a plurality of different combinations of two parameters, the above variables are changed to obtain a fourth-order cumulant set. , For each of the above set of quaternary cumulants,
A signal-to-noise power ratio improvement method, characterized in that each fourth-order cumulant is provided with a function shift of an amount obtained based on each of the above two parameters of each set, and an addition process is performed for each same variable to obtain an average thereof. 5. A signal-to-noise power ratio improving apparatus for suppressing a Gaussian noise component of an input signal, comprising: an input signal capturing means for sampling an input signal to generate a set of input signals; , One of the three variables of the fourth-order cumulant function is a variable, and the other two are parameters, respectively. For each combination of a plurality of different two parameters, And a fourth-order cumulant calculation processing means for obtaining a set of fourth-order cumulants by changing the above-mentioned parameters. For each of the fourth-order cumulant sets from the fourth-order cumulant calculation processing means, Phase compensation processing means for performing phase correction based on frequency information of a desired signal in the input signal, and output from the phase processing means. A signal-to-noise-to-noise-ratio improving apparatus, comprising: addition processing means for performing addition processing on the same variable for each variable and obtaining an average thereof. 6. A signal-to-noise power ratio improving method for suppressing a Gaussian noise component of an input signal, wherein the input signal is sampled to generate a set of input signals. One of the three variables of the next cumulant function is a variable, and the other two are parameters, respectively. For each of a plurality of different combinations of two parameters, the above variables are changed to obtain a fourth-order cumulant set. , For each of the above set of quaternary cumulants,
A signal pair comprising: performing a phase correction on each fourth-order cumulant based on the above two parameters of each set and frequency information of a desired signal in the input signal, and performing an addition process for each same variable to obtain an average thereof; Noise power ratio improvement method. 7. A signal-to-noise power ratio improving apparatus according to claim 1, wherein each set of inputs to said addition processing means is converted into a frequency component and output instead of said addition processing means. A signal-to-noise power ratio improving apparatus, comprising: a Fourier transform processing means for performing the above processing; and a frequency domain addition processing means for adding an output from the Fourier transform processing means for each same band to obtain an average thereof. 8. The method for improving a signal-to-noise ratio according to claim 2, 4 or 6, wherein a frequency component is converted to a frequency component by a Fourier transform instead of a procedure of adding each variable and calculating an average thereof. A signal-to-noise power ratio improving method, wherein an addition process is performed for each of the same bands to obtain an average thereof. 9. A signal-to-noise power ratio improving apparatus for suppressing a Gaussian noise component of an input signal, comprising: an input signal capturing means for sampling an input signal to generate a set of input signals; , One of the three variables of the fourth-order cumulant function is a variable, and the other two are parameters, respectively. For each combination of a plurality of different two parameters, Is changed to obtain a set of fourth-order cumulants, and for each set of fourth-order cumulants from the fourth-order cumulant calculation processing means, each fourth-order cumulant is converted into a frequency component and output. Fourier transform processing means, envelope detection processing means for performing envelope detection processing on the output from the Fourier transform processing means, A signal-to-noise power ratio improving apparatus, comprising: an addition processing means for adding an output from the envelope detection processing means for each of the same bands and obtaining an average thereof. 10. A signal-to-noise power ratio improving method for suppressing a Gaussian noise component of an input signal, wherein the input signal is sampled to generate a set of input signals. One of the three variables of the next cumulant function is a variable, and the other two are parameters,
For each of a plurality of different combinations of two parameters, the above variables are changed to obtain a fourth-order cumulant set, and for each of the fourth-order cumulant sets, each fourth-order cumulant is converted into a frequency component by Fourier transform processing. A signal-to-noise power ratio improving method characterized by performing conversion, performing envelope detection processing, performing addition processing for each of the same bands, and calculating an average thereof.