JP6296453B2 - Signal processing apparatus and signal processing method - Google Patents

Description

  The present invention relates to a signal processing apparatus and a signal processing method for removing noise during demodulation in an environment where noise is superimposed in broadcasting and communication using double sideband waves.

  As a method for demodulating a desired broadcast wave by removing the noise from the desired broadcast wave on which the noise is superimposed, techniques described in the following patent documents are known. The following Patent Document 1 discloses a first antenna that receives a broadcast wave stronger than the other antenna and obtains a first signal, and a second antenna that receives noise stronger than the other antenna and obtains a second signal. Is used to adjust the amplitude and phase of the second signal so that the signal level after synthesis is reduced, and synthesizes the first signal. In this technique, the amplitude and phase of the second signal are adjusted when the reception level of the desired broadcast wave is smaller than a certain threshold value. That is, when the noise power is larger than the power of the broadcast wave, the amplitude and phase of the second signal are adjusted so that the level of the combined signal becomes small. Since both antennas receive noise from the same noise source, the amplitude and phase of the noise received by both antennas differ according to the difference in distance from each antenna to the noise source. In order to compensate for this, the amplitude of the second signal is made to coincide with the amplitude of the first signal, the phase of the second signal is changed by π with respect to the phase of the first signal, and The second signal is synthesized in reverse phase. By adjusting the amplification factor and phase of the second signal in this way, a signal with noise canceled from the received broadcast wave can be obtained even when a desired broadcast wave can be received.

  Further, in the technique of Patent Document 2 below, in receiving an AM radio broadcast wave by a radio receiver mounted on a vehicle, pulsed noise emitted from an electronic device of the vehicle is mixed in the AM radio broadcast wave. This is a technology for removing sexual noise. In this technique, first, pulse noise in a band other than the AM radio broadcast wave band is detected, and the noise source is specified by obtaining the magnitude of the period of the pulse noise and the fluctuation range of the level. Depending on the noise source, in the broadcast wave on which pulse noise is superimposed, the harmonic band of the pulse noise near the broadcast wave frequency and the band of the noise period are attenuated for the time corresponding to the time width of the pulse noise. By doing so, the person who listens to AM radio broadcasting is prevented from giving discomfort due to pulse noise.

JP2012-257155A Patent No. 5012246

  The technique of Patent Document 1 is a technique that uses two antennas so that only noise can be received and is adjusted in advance so that the level of the combined signal of the received signals of the two antennas becomes small. For this reason, the technique of Patent Document 1 requires two antennas to cancel noise, and the setting for removing noise does not affect the adjustment of noise removal by the reception level of a desired broadcast wave. Thus, it is necessary to perform in an environment where the reception level is small. Further, since the amplitude and phase of the second signal are adjusted uniformly regardless of the frequency characteristics of the noise, the noise is not completely removed. Further, in the case of a DC-DC converter using the PWM method, the frequency characteristic of noise changes due to a change in pulse width even if the basic period does not change. For this reason, the method of Patent Document 1 cannot completely remove noise.

  The technique of Patent Document 2 detects pulse noise in a band other than the broadcast wave band, and attenuates an appropriate frequency band according to the type of noise for the time corresponding to the pulse width at the detection timing. Technology. Therefore, the broadcast wave is essentially attenuated only during the period of pulse noise. This causes discomfort to those who listen to AM radio broadcasts.

  Accordingly, an object of the present invention is to accurately remove noise having periodicity in a frequency space without affecting a desired signal.

In order to solve the above problems, the present invention provides a signal processing device that receives a double-sideband signal and removes noise superimposed on the RF band, and orthogonally demodulates the double-sideband signal to obtain a positive frequency band and a negative frequency. Fourier the demodulating means for demodulating the in-phase and quadrature components of the baseband and a band, the in-phase and quadrature components by Fourier transform to output a quadrature frequency components of the in-phase frequency components and complex function of the complex function Based on the transforming means, the in-phase frequency component and the quadrature frequency component, the transfer function calculating means for obtaining a time-averaged transfer function between the quadrature frequency component and the in-phase frequency component, and the frequency characteristic of the quadrature component based on the transfer function. The signal processing apparatus includes a synthesizing unit that corrects and synthesizes the in-phase component.
The gist of the present invention is to remove the noise component superimposed on the in-phase component based on the quadrature component, paying attention to the fact that the quadrature component does not include the signal component and only the noise component appears. The noise component can be removed on the frequency axis or the time axis.

1. Principle of the Invention The present invention removes noise from a demodulated signal using the following principle. When quadrature demodulation is performed on both sideband waves (for example, AM broadcast waves) that are not orthogonally multiplexed, a signal component and a noise component appear in the in-phase component of the baseband, no signal component appears in the orthogonal component, and only the noise component Appears. When noise is superimposed on the upper sideband, the phase of the quadrature component is delayed by π / 2 on the time axis in both the positive frequency band and the negative frequency band with respect to the noise component of the in-phase component. . That is, for the noise component in the positive frequency band, the phase of the quadrature component is rotated by −π / 2 with respect to the in-phase component, and for the noise component in the negative frequency band, the quadrature component is π / 2 with respect to the in-phase component. Only the phase is rotating. Note that the sign of the phase is defined as positive in the sign of the phase in the complex coordinate system, that is, the left rotation direction regardless of the positive frequency band and the negative frequency band. Also, the definition of phase advance and delay on the time axis is defined as the advance and delay with respect to the left rotation in the complex coordinate system for the positive frequency and the right rotation with respect to the negative frequency.

Regarding the noise superimposed on the upper sideband in the RF band, when the spectrum of the noise component in the positive frequency band of the in-phase component is ρ _U (ω), the spectrum of the noise component in the positive frequency band of the quadrature component is −jρ _U (ω). However, ρ _U is a complex number including an absolute value and a phase. j is a pure imaginary number.

Since both the in-phase component and the quadrature component after quadrature demodulation are real functions, the spectrum of the noise component in the baseband is in a complex conjugate relationship between the positive frequency band and the negative frequency band. Therefore, the spectrum of the noise component in the negative frequency band of the in-phase component is represented by ρ _U ^* (− ω), and the spectrum of the noise component in the negative frequency band of the quadrature component is represented by jρ _U ^* (− ω). . However, ω> 0 and * means a complex conjugate operation.

As for the noise superimposed on the lower sideband in the RF band, the noise component of the quadrature component has a phase of π on the time axis in both the positive frequency band and the negative frequency band with respect to the noise component of the in-phase component. Progressing by / 2. That is, for the noise component in the positive frequency band, the phase of the quadrature component is rotated by π / 2 with respect to the in-phase component, and for the noise component in the negative frequency band, the quadrature component is −π / 2 with respect to the in-phase component. Only the phase is rotating. When the spectrum of the noise component in the positive frequency band of the in-phase component is ρ _L ^* (ω), the spectrum of the noise component in the positive frequency band of the quadrature component has a relationship of jρ _L ^* (ω). Similarly, the spectrum of the noise component in the baseband is in a complex conjugate relationship between the positive frequency band and the negative frequency band. Therefore, the spectrum of the noise component in the negative frequency band of the in-phase component is represented by ρ _L (−ω), and the spectrum of the noise component in the negative frequency band of the quadrature component is represented by −jρ _L (−ω).

The above relationship is established for frequencies in all positive frequency bands and all negative frequency bands. In other words, if the spectrums -jρ _U (ω) and jρ _L ^* (ω) of the noise component in the positive frequency band of the orthogonal component are known, the spectrum of the noise component in the entire frequency band of the in-phase component is also uniquely determined. Because of the above relation, the spectrum of the noise component in the positive frequency band of the in-phase component becomes the spectrum of the noise component in the positive frequency band of the quadrature component −jρ _U (ω) and jρ _L ^* (ω), respectively ( It can be obtained by multiplying j) and (-j). Similarly, the spectrum of the noise component in the negative frequency band of the in-phase component becomes the spectrum jρ ^* _U (−ω) and −jρ _L (−ω) of the noise component in the negative frequency band of the quadrature component, respectively (−j). , (J) can be obtained by multiplication. The multiplication factor {(−j), (j)} in the negative frequency domain is also a complex conjugate of the multiplication factor {(j), (−j)} in the positive frequency domain.

  Note that multiplication by j means a phase rotation of π / 2 in the complex coordinate system, and on the time axis, the phase is advanced by π / 2 for the positive frequency band, and only π / 2 for the negative frequency band. It means delaying the phase. The multiplication of −j means −π / 2 phase rotation in the complex coordinate system. On the time axis, the phase is delayed by π / 2 for the positive frequency band, and only π / 2 for the negative frequency band. It means to advance the phase.

The spectrum of the noise component in the positive frequency band of the orthogonal component, −jρ _U (ω), jρ _L ^* (ω) can be generalized as follows. That is, the noise spectrum of the quadrature component is obtained by multiplying the spectrum {ρ _U (ω), ρ _L ^* (ω)} of the in-phase component for each frequency band by a factor of −j or j. . Therefore, if {ρ _U (ω), ρ _L ^* (ω)} is generalized to the total noise spectrum ρ (ω) in the positive frequency band, the spectrum of the noise component in the positive frequency band of the orthogonal component is jsign (ω ) Ρ (ω) and the general formula. Here, ρ (ω) is the spectrum (complex function) of the noise component in the positive frequency band of the in-phase component, and sign (ω) is a function of ω taking +1 or −1 at an arbitrary frequency ω. is there. Hereinafter, this sign (ω) is referred to as a sign function (real function). Since the sign function only represents the phase relationship between the in-phase component and the quadrature component, it does not vary with time. If the sign function sign (ω) and therefore jsign (ω) are known, the spectrum ρ (ω) of the noise component in the positive frequency band of the in-phase component can be obtained from the quadrature component. When ρ (ω) = Z (ω) (jsign (ω) ρ (ω)) is defined as a transfer function Z (ω) for converting from a quadrature component to an in-phase component, Z (ω) = − jsign (ω) The transfer function can be obtained. Similarly, the transfer function in the negative frequency band can be obtained by Z ^* (− ω) = jsign (−ω).

Accordingly, if the transfer function Z (ω) can be obtained from the in-phase frequency component and the quadrature frequency component, the noise component of the in-phase component can be obtained, and if this noise component is removed from the in-phase component, the in-phase component is noisy. The signal component is removed. In order to obtain the transfer function Z (ω), an in-phase frequency component and a quadrature frequency component are used, and the in-phase frequency component includes a signal component together with a noise component. Therefore, in order to eliminate the influence of the signal component when obtaining the transfer function Z (ω) between the in-phase component and the quadrature component, an average calculation is performed over a certain period of time. That is, since it can be considered that the signal component changes irregularly, if the time average is performed, the average value becomes small. Therefore, the signal component can be eliminated by the moving averaging process.
The present invention uses the above principle.

In the present invention, the time axis signal is Fourier-transformed to obtain the transfer function. If the transfer function is known, the noise component of the in-phase component on the time axis can be obtained by convolving and integrating the impulse response with the time axis orthogonal component. Therefore, a signal from which noise is removed can be obtained by subtracting the time characteristic of the result of convolution integration from the in-phase component on the time axis.
Therefore, in the present invention, the synthesizing means is equivalent to the equivalent means for obtaining the quadrature component on the time axis corrected by the convolution of the quadrature component on the time axis and the impulse response of the transfer function, and the in-phase component on the time axis. It is possible to adopt a configuration having noise removal means for removing the orthogonal component on the corrected time axis output from the means. For synthesis on the time axis, a transversal filter or other filter that realizes convolution integration can be used.

Further, noise can be removed on the frequency axis.
That is, in the above invention, the combining means subtracts the corrected orthogonal frequency component output from the equivalent means from the equivalent means for correcting the orthogonal frequency component by the transfer function and outputs the noise-removed in-phase frequency component from the in-phase frequency component. And a noise removing unit that performs the inverse Fourier transform on the noise-removed in-phase frequency component to obtain a demodulated signal on the time axis.

2. Transfer Function There are several configurations for obtaining the transfer function.
(1) Method of calculating transfer function from time average of complex conjugate product In the above invention, the transfer function calculating means multiplies one of the in-phase frequency component and the quadrature frequency component and the other complex conjugate for each frequency. Complex conjugate multiplication means, a time average calculation means for calculating the time average of the output of the complex conjugate multiplication means, and a code component for each frequency is extracted from the output of the time average calculation means, and the frequency characteristic of the code component is transmitted. And a code frequency characteristic extracting means as a function.

The in-phase frequency component F _r (ω) in the positive frequency band is represented by S (ω) + ρ (ω). Further, the orthogonal frequency component F _i (ω) in the positive frequency band is represented by jsign (ω) ρ (ω) as described above. F _r (ω) F _i ^* (ω) is S (ω) (− jsign (ω)) ρ ^* (ω) −jsign (ω) | ρ (ω) | ² * Means complex conjugate operation. Since S (ω) (− jsign (ω)) ρ ^* (ω) of the first term changes irregularly in time, if an average is taken, an average residual Δ ((min) including zero) Δ ( ω). | Ρ (ω) | ² of the time-average Av (| ρ (ω) | 2) is because it is the square of the time average of the absolute value of the noise spectrum, not a small amount. Av represents a time average value. Accordingly, since the time average {−jsign (ω) Av (| ρ (ω) | ² )} of the second term is an imaginary number, Av (| ρ (ω) | ² ) is the absolute value of Δ (ω) The sign {-jsign (ω)} can be determined for each frequency. Therefore, the transfer function Z (ω) = − jsign (ω) described above is obtained, and the transfer function can be obtained as Z ^* (− ω) = jsign (−ω) in the negative frequency band.

Note that F _r ^* (ω) F _i (ω) is S ^* (ω) (jsign (ω) ρ (ω)) + jsign (ω) | ρ (ω) | ² If this moving time average is taken, S ^* (ω) (jsign (ω) ρ (ω)) can be an average residual Δ (ω) of a minute amount including zero. Therefore, if the time average Av (| ρ (ω) | ² ) is larger than Δ (ω), the sign jsign (ω) can be determined for each frequency. Therefore, −jsign (ω) with the sign inverted can be obtained as the above-described transfer function Z (ω). In the negative frequency band, the transfer function can be obtained as Z ^* (− ω) = jsign (−ω). Therefore, one of the in-phase frequency component and the quadrature frequency component and the other complex conjugate are multiplied for each frequency, the time average of the multiplied value is calculated, and the code component for each frequency is extracted from the time average. The frequency characteristic of the code component can be used as a transfer function.
Even if the transfer function in the negative frequency domain is obtained as a conjugate complex number of the transfer function in the positive frequency domain, F _r (−ω) F _i ^* (− ω) or F _r (− ω) F _i ^* (-ω) of may be obtained in time average. As a result, Z ^* (− ω) = jsign (−ω) is obtained.

  In the above invention, the pure imaginary number j or -j is multiplied or divided by only one of the quadrature component and the in-phase component output from the demodulation means, or the quadrature frequency component and the in-phase frequency component output from the Fourier transform means. Only one of them may have a phase rotation means that newly uses a frequency component obtained by multiplying or dividing the pure imaginary number j or -j as an orthogonal frequency component. In this case, the above description holds true except that the sign of the moving time average of the product (one is complex conjugate) of the in-phase frequency component and the quadrature frequency component becomes ± sign (ω) and a real number. Since the product of the sign function and the ± j quadrature frequency component becomes an in-phase frequency component, or the product of the sign function and the quadrature frequency component becomes a ± j in-phase frequency component, −jsign (ω) is obtained as a transfer function. It is none other than that.

(2) Method for obtaining transfer function from ratio The in-phase frequency component F _r (ω) in the positive frequency band is S (ω) + ρ (ω). As described above, the orthogonal frequency component F _i (ω) in the positive frequency band is jsign (ω) ρ (ω). Therefore, the transfer function Z (ω) for the orthogonal frequency component of the in-phase frequency component of the noise component is F _r (ω) / F _i (ω) = (S (ω) + ρ (ω)) / (jsign (ω) ρ (Ω)) is defined as a frequency function of the sign of the imaginary part of the time average. The time average of S (ω) / (jsign (ω) ρ (ω)) is an average residual Δ (ω) of a minute amount including zero. Further, ρ (ω) / (jsign (ω) ρ (ω)) is −jsign (ω) even in the instantaneous value. Therefore, when the absolute value of the average residual Δ (ω) is smaller than 1, the transfer function Z (ω) can be obtained by the moving time average of the ratio of the in-phase frequency component to the orthogonal frequency component for each frequency. it can.
Further, the transfer function Y (ω) for the in-phase frequency component of the orthogonal frequency component in the positive frequency region is F _i (ω) / F _r (ω) = (jsign (ω) ρ (ω)) / (S (ω) + Ρ (ω)) is defined as a frequency function of the sign of the imaginary part of the time average. Also at this time, F _i (ω) / F _r (ω) = jsign (ω) / {[S (ω) / ρ (ω)] + 1}, and the time average of S (ω) / ρ (ω) Is an average residual Δ (ω) of a minute amount including zero. When the absolute value of the average residual Δ (ω) is smaller than 1, the sign of the imaginary part of the time average of F _i (ω) / F _r (ω) is sign (ω). Therefore, the transfer function Z (ω) may be obtained by the reciprocal of the transfer function Y (ω) obtained in this way.

Therefore, in the above invention, the transfer function calculating means calculates the time average of the frequency characteristics obtained by dividing the in-phase frequency component by the orthogonal frequency component for each frequency, or the time average of the frequency characteristics obtained by dividing the orthogonal frequency component by the in-phase frequency component. A time average calculating means for calculating, and a code frequency characteristic extracting means for extracting a code component for each frequency from the output of the time average calculating means and using the frequency characteristic of the code component as a transfer function can be provided. .
Further, when the absolute value of the average residual Δ (ω) is small, an in-phase frequency component of noise can be obtained from the transfer function and the orthogonal frequency component even if a transfer function that is not encoded is used. That is, in the above invention, the transfer function calculating means calculates the time average of the frequency characteristic obtained by dividing the in-phase frequency component by the orthogonal frequency component for each frequency, or the time average of the frequency characteristic obtained by dividing the orthogonal frequency component by the in-phase frequency component. A time average calculating means for calculating, and a time-averaged frequency characteristic output from the time average calculating means can be used as a transfer function.

When obtaining the transfer function Z (ω), if ρ (ω) is 0, the transfer function Z (ω) (including the case of obtaining Z (ω) = 1 / Y (ω)) diverges. In order to avoid this, the following configuration may be employed.
In other words, in the present invention, the transfer function calculating means or the combining means performs a calculation or combination on a frequency component whose absolute value of the orthogonal frequency component is greater than or equal to a predetermined threshold value, and the absolute value of the orthogonal frequency component is smaller than the predetermined threshold value It can be set as the structure which does not perform a calculation or a synthesis | combination with respect to a component.
That is, it is only necessary to extract only the spectrum of the frequency at which the absolute value of the orthogonal frequency component is equal to or greater than a predetermined threshold value and calculate the ratio with the in-phase frequency component. The predetermined threshold is such that the absolute value of the average residual Δ (ω) that is the time average of S (ω) / (jsign (ω) ρ (ω) or S (ω) / ρ (ω)) is 1 or less. What is necessary is just to set to the absolute value of (rho) ((omega)) which becomes. At other frequencies, the calculation of the transfer function Z (ω) and the synthesis for the in-phase frequency component are not executed. In this way, although noise components cannot be removed at frequencies that are not calculated, it is possible to prevent the quadrature noise components from being superimposed on the in-phase noise components in the same phase. Since the noise level at this frequency is originally small, it does not matter if the noise cannot be removed from the in-phase component.

Further, in the above invention, the synthesizing means, when the transfer function calculating means calculates the time average of the frequency characteristic obtained by dividing the in-phase frequency component by the orthogonal frequency component for each frequency, the absolute value of the time average is 2 When a synthesis operation is performed only on a small frequency component and the transfer function calculation means calculates a time average of frequency characteristics obtained by dividing the orthogonal frequency component by the in-phase frequency component, the absolute value of the time average is ½ A composition operation may be executed only for larger frequency components.
If the absolute value of the orthogonal frequency component is not 0, the above ratio can be calculated. However, when the absolute value of the average residual Δ (ω) of S (ω) / (jsign (ω) ρ (ω)) is 1 or more, the absolute value of the time average of the in-phase frequency component / orthogonal frequency component is 2 or more. When the absolute value of the ratio is 2 or more, the sign of the imaginary part of the ratio does not represent −jsign (ω). Since the absolute value of −jsign (ω) is 1, when the absolute value of the ratio is smaller than 2, the sign of the imaginary part of the ratio is not inverted with respect to −jsign (ω).
Further, when the absolute value of the average residual Δ (ω) of S (ω) / (ρ (ω) is 1 or more, | Δ (ω) +1 | may be 2 or more, and therefore, orthogonal The absolute value of the time average of the frequency component / in-phase frequency component may be ½ or less, in which case the sign of Δ (ω) +1 may be negative. If the absolute value is ½ or less, the sign of the imaginary part of the ratio will not represent jsign (ω) If the absolute value of the ratio is greater than ½, the sign of the imaginary part of the ratio is , Jsign (ω) is not inverted.
When the transfer function is obtained by the ratio calculation, if the upper sideband noise and the lower sideband noise overlap at the same frequency in the baseband, the time average of the ratio is -jsign (ω) Don't be. Therefore, the transfer function Z (ω) is obtained directly as the time average of the ratio without obtaining the sign function from the time average of the ratio. When the absolute value of the above average residual Δ (ω) is small and the time average of the ratio does not change greatly, this transfer function Z (ω) is superimposed on the upper sideband noise and the lower sideband noise at the same frequency. By multiplying the generated orthogonal frequency component, the noise of the in-phase frequency component superimposed on the upper sideband noise and the lower sideband noise at the same frequency can be obtained.

In the present invention, mask means for removing a band including direct current may be provided between the output of the Fourier transform means and the output of the transfer function calculation means.
In quadrature demodulation, when the demodulated carrier wave is not synchronized with the carrier wave of the received signal, a beat signal appears near the direct current, which also appears in the orthogonal frequency component. The transfer function can be obtained by removing this frequency component.

3. Synchronous demodulation In the present invention, the demodulating means controls the frequency and phase of the demodulated carrier so that the beat signal of the error frequency of the demodulated carrier with respect to the modulated carrier contained in the quadrature component after quadrature demodulation is zero. It is desirable to have a part. Synchronous detection can be performed, and noise can be reliably removed.

In addition, the demodulation means obtains a beat signal of the error frequency of the demodulated carrier wave with respect to the modulated carrier wave from the moving average of the baseband signal, and based on the beat signal, a signal obtained by correcting the fluctuation due to the beat signal of the baseband signal is newly added to the baseband signal. It is desirable to have synchronization means for making a band signal. Since the spectrum in the baseband is shifted in frequency by the beat frequency for both the in-phase component and the quadrature component, the signal component can be demodulated and noise can be reliably removed by correcting this frequency shift.
The synchronization unit may include a unit that corrects the delay of the instantaneous phase caused by the moving average of the baseband signal according to the time difference of the detected instantaneous phase and the moving average.

4). Method invention The present invention is a signal processing method for receiving a double-sideband signal and removing noise superimposed on the RF band, and orthogonally demodulating the double-sideband signal to have a positive frequency band and a negative frequency band. The baseband in-phase component and quadrature component are demodulated, and the in-phase component and quadrature component are Fourier transformed to obtain the in-phase frequency component and quadrature frequency component. Based on the in-phase frequency component and quadrature frequency component, the quadrature frequency component Is a signal processing method characterized in that a time-averaged transfer function between the in-phase frequency components is obtained, the frequency characteristics of the quadrature components are corrected based on the transfer function, and the in-phase components are synthesized. What has been described above for the device invention is also applicable to the method invention.

In the method invention, a code component for each frequency may be extracted from a time-averaged transfer function, and the frequency characteristic of the code component may be used as a transfer function for correcting the orthogonal component.
In the method invention, the synthesis is performed by removing a quadrature frequency component obtained by correcting the quadrature frequency component with a transfer function from the in-phase frequency component to obtain a noise-removed common-mode frequency component, and performing an inverse Fourier transform on the noise-removed common-mode frequency component, A demodulated signal on the time axis may be used. This is the same as the device invention.

  According to the present invention, since noise can be accurately removed during demodulation in an environment where noise is superimposed on the RF band, the detection accuracy and demodulation accuracy of a desired signal can be improved.

The block diagram of the signal processing apparatus which concerns on the specific Example 1 of this invention. FIG. 3 shows frequency characteristics of an input signal and a demodulated signal of the signal processing apparatus according to Embodiment 1. The frequency characteristic of the in-phase component in the baseband after the demodulation of the signal processing apparatus of Example 1, and a quadrature component. FIG. 6 is a configuration diagram of a signal processing device according to a second embodiment. FIG. 6 is a configuration diagram of a signal processing device according to a third embodiment.

  Hereinafter, the present invention will be described based on specific examples. The present invention is not limited to the following examples.

1. Constitution
A configuration of a signal processing apparatus 1 according to a specific embodiment of the present invention is shown in FIG. A present Example is a signal processing apparatus which suppresses the noise mixed in the AM radio receiver in HV (hybrid vehicle). It is assumed that the HV is equipped with a DC-DC converter that is controlled at a carrier frequency of 100 kHz. The AM radio broadcast wave is assigned a frequency band of 531 kHz to 1602 kHz. The switching noise generated from the DC-DC converter basically becomes a line spectrum string in the frequency space. This noise enters the AM radio broadcast band and gives noise to the AM radio broadcast wave. The present invention can remove noise even when noise is present in both the upper sideband and lower sideband of AM radio broadcasting. This embodiment is a signal processing device that cancels this type of noise entering the AM radio broadcast band. However, the present invention is not limited to such noise, and can be used in all environments in which noise is mixed in the RF band in double-sideband transmission that is not orthogonally multiplexed. Further, the noise to be mixed is not limited to the line spectrum but may be a continuous spectrum having a band.

  In the signal processing apparatus of this embodiment, the AM radio broadcast signal received by the antenna 11 is amplified by the amplifier 12, sampled by the A / D converter 13 at a constant period Δt, and converted into a digital value. A device processed by a CPU. Of course, it is possible to construct all or part of the analog circuit, but since it is easy to process digitally, this embodiment is based on digital processing. The configuration of FIG. 1 is expressed in blocks for each functional unit of digital processing. The signal output from the A / D converter 13 is a real number, but the orthogonal demodulator 20 and subsequent data processing are all performed in complex numbers. The quadrature demodulating unit 20 serving as a demodulating unit includes a mixer 21, a demodulated carrier wave generating unit 22, an in-phase component extracting unit 23, and a quadrature component extracting unit 24. A baseband signal is obtained by the orthogonal demodulator 20. In view of handling with a complex signal, this baseband signal has a positive frequency band corresponding to the upper sideband band and a negative frequency band corresponding to the lower sideband band.

  The quadrature demodulation unit 20 is provided with a phase synchronization processing unit 70. The phase synchronization processing unit 70 receives a baseband signal and calculates a moving average of the moving average calculating unit 71, a complex conjugate calculating unit 72 that calculates the complex conjugate of the output, and an amplitude that normalizes the amplitude of the output. A normalization unit 73 and a multiplication unit 74 that multiplies the output by the baseband signal.

  The in-phase component output from the in-phase component extraction unit 23 is input to a Fourier transform unit 45 (hereinafter referred to as “FFT”) to obtain an in-phase frequency component. Further, the orthogonal component output by the orthogonal component extraction unit 24 is input to the FFT 46 to obtain an orthogonal frequency component. The apparatus according to the present embodiment further includes a multiplication unit 26, a transfer function calculation unit 60, and a synthesis unit 80. The multiplier 26 is a part that multiplies the orthogonal frequency component by j and constitutes a phase rotation means. The transfer function calculation unit 60 is a part that calculates a transfer function for the quadrature frequency component of the in-phase frequency component, and has a complex conjugate calculation unit 47, a multiplication unit 48, a time average calculation unit 49, and a code frequency characteristic extraction unit 50. ing. The synthesizing unit 80 is a part that corrects the quadrature frequency component using the transfer function obtained by the transfer function calculating unit 60 and subtracts the corrected frequency component from the in-phase frequency component to remove noise. Unit 81, subtracting unit 82, and inverse Fourier transform unit (hereinafter referred to as “IFFT”) 83.

  The orthogonal frequency component output from the FFT 46 is multiplied by j by the multiplication unit 26 (the phase is rotated by π / 2) and input to the complex conjugate calculation unit 47 to obtain a complex conjugate orthogonal frequency component. The complex conjugate quadrature frequency component and the in-phase frequency component are input to the multiplication unit 48, multiplied by each component for each frequency, and output to the time average calculation unit 49 as a complex product frequency component. The time average calculation unit 49 calculates a time average within the past fixed time T with respect to the current time t. The output of the time average calculation unit 49 is input to the code frequency characteristic extraction unit 50 to calculate a time average transfer function composed of the above-described sign function of -jsign (ω). The time average transfer function and the orthogonal frequency component are multiplied for each frequency in the multiplication unit 81. The output of the multiplier 81 is an in-phase noise component predicted from the quadrature noise component (hereinafter referred to as “estimated in-phase noise component”). The subtracting unit 82 subtracts the estimated in-phase noise component from the in-phase frequency component for each frequency to obtain a noise-removed in-phase frequency component from which noise has been removed. The noise-removed in-phase frequency component is input to IFFT 83 and subjected to inverse Fourier transform to obtain a demodulated signal S (t) from which noise on the time axis has been removed.

この受信信号ｒ（ｔ）のフーリエ変換であるスペクトルは図２（ａ）に示すようになり、上側帯波帯域と下側帯波帯域とを有している。Ｓ^- は、下側帯波のスペクトル、Ｓ⁺ は、上側帯波のスペクトルであり、Ａは搬送波の振幅、ρ_U ，ρ_L は、放送局から受信装置までにおいて、それぞれ、ＲＦ帯域の上側帯波帯域、下側帯波帯域に重畳された雑音のスペクトルである。Ａは実数、Ｓ^- 、Ｓ⁺ 、ρ_U ，ρ_L は角周波数ω（以下、単に、「周波数」と記す）に関する複素関数である。すなわち、絶対値と位相とを含んでいる。ωに関して、Ｓ^- 、Ｓ⁺ の絶対値は等しく、位相は反転関係にある。したがって、Ｓ^- 、Ｓ⁺ は相互に複素共役関数である。Ｓ^- （ｔ）、Ｓ⁺ （ｔ）は、それぞれ、Ｓ^- 、Ｓ⁺ の逆フーリエ変換であり、時間に関する複素関数である。また、Ｓ^- （ｔ）、Ｓ⁺ （ｔ）は、相互に複素共役の関係にあり、したがって、Ｓ^- （ｔ）＋Ｓ⁺ （ｔ）は実関数である。ω_c は、変調時の搬送波の周波数、ω_c ＋ω_U ，ω_c −ω_L は、それぞれ、上側帯波、下側帯波に重畳した雑音の周波数である。ω_c ＞０、ω_U ＞０、ω_L ＞０として定義する。 Next, the operation of the signal processing apparatus will be described.
1. The spectrum noise of the received signal is superimposed on the upper sideband and the lower sideband from the broadcasting station to the receiving device. The reception signal r (t) output from the antenna 11 is expressed by equation (1).

A spectrum that is a Fourier transform of the received signal r (t) is as shown in FIG. 2A, and has an upper sideband and a lower sideband. S ⁻ is the spectrum of the lower sideband, S ⁺ is the spectrum of the upper sideband, A is the amplitude of the carrier wave, and ρ _U and ρ _L are the upper band of the RF band from the broadcasting station to the receiving device, respectively. This is a spectrum of noise superimposed on the waveband and the lower sideband. A is a real number, and S ⁻ , S ⁺ , ρ _U , and ρ _L are complex functions relating to an angular frequency ω (hereinafter simply referred to as “frequency”). That is, it includes an absolute value and a phase. With respect to ω, the absolute values of S ⁻ and S ⁺ are equal, and the phases are in an inversion relationship. Therefore, S ⁻ and S ⁺ are mutually complex conjugate functions. S ⁻ (t) and S ⁺ (t) are inverse Fourier transforms of S ⁻ and S ⁺ , respectively, and are complex functions related to time. Further, S ⁻ (t) and S ⁺ (t) are in a complex conjugate relationship with each other, and therefore S ⁻ (t) + S ⁺ (t) is a real function. ω _c is the frequency of the carrier wave during modulation, and ω _c + ω _U and ω _c −ω _L are the frequencies of noise superimposed on the upper side band and the lower side band, respectively. It is defined as ω _c > 0, ω _U > 0, and ω _L > 0.

信号成分の直交成分は存在しないので、複素空間では、直交復調は、（１）式で表される複素関数の実部の受信信号にｅｘｐ［−ｊ（ω_c ＋Δω）ｔ］を掛ける演算を行うことに等しい。したがって、ミキサー２１の出力する復調した後のベースバンドの信号は、（３）式で表される。なお、復調結果には１／２の係数が係るので、表現を簡単にするために、ｘ（ｔ）は、直交復調の結果の２倍で定義する。ベースバンド信号に、ｅｘｐ（−ｊΔωｔ）の因子が現れる。なお、ミキサー２１の出力には２ω_c の高調波成分が含まれるので、実際には、ローパスフィルタにより高調波成分は除去されている。

このベースバンド信号ｘ（ｔ）が移動平均演算部７１によりその移動平均が演算される。移動平均の結果は、（４）式で与えられる。

すなわち、移動平均により、（３）式の第２項及び第３項の周波数はΔωに比べて大きいので、移動平均により、この項は０となる。 A wave propagating in space is represented by the real part of r (t). Therefore, the sampled received signal (data) output from the A / D converter 13 is a real number sequence. Next, the received signal is demodulated orthogonally.
1. Synchronous Demodulation The frequency of the demodulated carrier wave output from the demodulated carrier wave generator 22 is assumed to be larger by Δω than the frequency ω _{c of the} modulated carrier wave. That is, the demodulated carrier wave L (t) is expressed by equation (2).

Since there is no quadrature component of the signal component, in the complex space, quadrature demodulation is performed by multiplying the received signal in the real part of the complex function expressed by equation (1) by exp [−j (ω _c + Δω) t]. Equivalent to doing. Therefore, the demodulated baseband signal output from the mixer 21 is expressed by equation (3). Since the demodulation result has a factor of 1/2, x (t) is defined as twice the result of the orthogonal demodulation in order to simplify the expression. A factor of exp (−jΔωt) appears in the baseband signal. Since the output of the mixer 21 includes 2ω _c harmonic components, the harmonic components are actually removed by the low-pass filter.

The moving average of the baseband signal x (t) is calculated by the moving average calculator 71. The result of the moving average is given by equation (4).

That is, due to the moving average, the frequency of the second term and the third term in the equation (3) is larger than Δω, so that this term becomes 0 by the moving average.

なお、移動平均により、上記のΔωｔの位相量θを決定している。しかし、この値は、平均期間Ｔにおける平均値Θであって、雑音除去の演算を行う現在時刻ｔでの位相量ではない。そこで、平均演算を行うタイミング毎に平均位相量Θは演算されるで、平均演算毎の変化量Δφを求める。平均位相量Θは、時間区間Ｔの中点（現時刻ｔに対してＴ／２時間前）での値と見做すことができる。とすると、現時刻ｔでの位相量θ（ｔ）は、Θ＋ΔφＭ、Ｍは中点から現時刻までの平均演算を行う点数である。このように、平均演算により求められた位相量を補正して、現時刻の位相Δωｔを精密に求めて、（４）式のΔωｔとすれば、より完全に復調時の同期を実現することができ、正確な信号の復調が可能となる。 Next, the complex conjugate of the equation (4) is obtained by the complex conjugate computing unit 72, and the normalized signal of the equation (5) is obtained by the amplitude normalizing unit 73. Since A + S ⁺ (t) + S ⁻ (t) in equation (4) is a real number, −jΔωt is obtained from equation (4) by −jΔωt = tan ⁻¹ (real part / imaginary part), so exp ( jΔωt).

The phase amount θ of Δωt is determined by moving average. However, this value is the average value Θ in the average period T, and is not the phase amount at the current time t at which the noise removal calculation is performed. Therefore, the average phase amount Θ is calculated every time the average calculation is performed, and the change amount Δφ for each average calculation is obtained. The average phase amount Θ can be regarded as a value at the midpoint of the time interval T (T / 2 hours before the current time t). Then, the phase amount θ (t) at the current time t is Θ + ΔφM, and M is the number of points to perform the average calculation from the middle point to the current time. As described above, if the phase amount obtained by the average calculation is corrected and the phase Δωt at the current time is accurately obtained and Δωt in the equation (4) is used, synchronization at the time of demodulation can be realized more completely. And accurate demodulation of the signal becomes possible.

この処理により、復調搬送波の周波数が変調搬送波の周波数に対して偏差Δωを有していても、その偏差による影響を排除することができる。
なお、上記の説明では、受信信号に含まれる復調搬送波と、変調搬送波との位相差Δφは、明示していないが、（２）〜（５）式におけるｊΔωｔをｊΔωｔ＋ｊΔφとおいて、位相誤差Δφを考慮して、（６）式を演算すると、Δφは消去されるので、Δφが存在しても、（６）式が得られる。すなわち、周波数誤差だけでなく位相誤差も、補償されることになる。 Next, the multiplier 74 can multiply the baseband signal by the normalized signal of the formula (5) to obtain the synchronized baseband signal x _sync (t) of the formula (6).

By this processing, even if the frequency of the demodulated carrier wave has a deviation Δω with respect to the frequency of the modulated carrier wave, the influence of the deviation can be eliminated.
In the above description, the phase difference Δφ between the demodulated carrier wave and the modulated carrier wave included in the received signal is not clearly shown, but the phase error Δφ is calculated by setting jΔωt to jΔωt + jΔφ in equations (2) to (5). Considering this, if the expression (6) is calculated, Δφ is deleted, so that even if Δφ exists, the expression (6) can be obtained. That is, not only the frequency error but also the phase error is compensated.

Therefore, the baseband signal demodulated by the phase synchronization processing unit 70 is expressed by equation (7). That is, the output signal x _sync (t) of the mixer 74 can be expressed by equation (7), and its spectrum is as shown in FIG. 2B, and has a baseband positive frequency band and negative frequency band. ing. Noise has a spectrum of ρ _{U at} the frequency ω _U in the positive frequency band and ρ _L at −ω _L in the negative frequency band.

すなわち、同相成分抽出部２３の出力する同相成分ｘ_r （ｔ）が（８）式で、直交成分抽出部２４の出力する直交成分ｘ_i （ｔ）が（９）式で、表現される。 The real part of equation (7) is the in-phase component in quadrature demodulation, and the imaginary part is the quadrature component in quadrature demodulation.
The in-phase component is expressed by equation (8), and the quadrature component is expressed by equation (9).

That is, the in-phase component x _r (t) output from the in-phase component extraction unit 23 is _{expressed by} equation (8), and the quadrature component x _i (t) output by the quadrature component extraction unit 24 is expressed by equation (9).

直交成分ｘ_i （ｔ）は、ＦＦＴ４６に入力し、フーリエ変換されて直交周波数成分Ｘ_i （ω）が、（１１）式のように求められる。

The in-phase component x _r (t) is input to the FFT 45 and subjected to Fourier transform, and the in-phase frequency component X _r (ω) is obtained as shown in Equation (10).

The quadrature component x _i (t) is input to the FFT 46 and subjected to Fourier transform to obtain the quadrature frequency component X _i (ω) as shown in equation (11).

S ⁺ (ω) is a function defined only in the region of ω> 0, and S ⁻ (ω) is a function defined only in the region of ω <0. δ (ω−ω _U ) or the like is a delta function of 1 when ω = ω _U and 0 otherwise. The in-phase component has a signal component and a noise component, but the quadrature component has no signal component and only a noise component. The spectrum of the in-phase component represented by the equation (10) is as shown in FIG. In the positive frequency band, the spectrum (ρ _U / 2) and (ρ _L ^* / 2) of the noise component appear in the spectrum S ^{+ of the} signal component and the frequencies ω _u and ω _L , respectively, and in the negative frequency band. , Noise component spectra (ρ _U ^* / 2) and (ρ _L / 2) appear in the signal component spectrum S ⁻ and the frequencies −ω _u and −ω _L , respectively. ρ ^* is a complex conjugate of ρ and a spectrum obtained by inverting the phase of ρ. Both the in-phase component x _r (t) and the quadrature component x _i (t) are real functions.

The spectrum of the orthogonal component represented by the equation (11) is as shown in FIG. In the positive frequency band, spectrums (−jρ _U / 2) and (jρ ^* _L / 2) of noise components of orthogonal components appear at frequencies ω _u and ω _L , respectively. That is, this noise component has the same amplitude as each noise component of the in-phase component, but the phase is rotated by −π / 2, π / 2 with respect to the in-phase component (on the time axis, π / 2 is delayed and advanced). In the negative frequency band, spectrums (jρ _U ^* / 2) and (−jρ _L / 2) of noise components of orthogonal components appear at frequencies −ω _u and −ω _L , respectively. That is, the noise component has the same amplitude as the noise component of the in-phase component, but the phase is rotated by π / 2 and −π / 2 with respect to the in-phase component, respectively (on the time axis, π / 2 is delayed and advanced). Further, in both the in-phase component and the quadrature component, the spectra in the positive frequency band and the negative frequency band are in a complex conjugate relationship, that is, a relationship in which the phases are inverted.

2. Next, the quadrature frequency component X _i (ω) is input to the phase rotation unit 26 that rotates the phase by π / 2 (multiply by j) and is multiplied by j to be phase rotation orthogonal. A frequency component jX _i (ω) is obtained. The phase rotation orthogonal frequency component jX _i (ω) is input to the complex conjugate calculation unit 47, and the complex conjugate orthogonal frequency component (jX _i (ω)) ^* is calculated. Next, the multiplication unit 48 calculates a product for each frequency of the in-phase frequency component X _r (ω) output from the FFT 45 and the complex conjugate orthogonal frequency component (jX _i (ω)) ^* . The complex product frequency component X _r (ω) (jX _i (ω)) ^* , which is the calculation result, is expressed by equation (12). However, in order to simplify the expression, it is expressed by four times the complex product frequency component.

（１２）式において、信号成分Ｓが係る項は、信号は不規則に変化していると考えられるので、時間平均をとれば、雑音の大きさに対して十分に小さく０と見做すことができる。雑音成分については振幅の２乗の時間平均であるので、雑音が存在する以上、０とはならない。 Next, in the complex product frequency component, the time average calculation unit 49 calculates the time average of the past T period (time moving average) with respect to the current time t. The result is expressed by equation (13) as time average R (ω).

In the equation (12), the term related to the signal component S is considered that the signal is irregularly changed. Therefore, if the time average is taken, it is assumed that the term is sufficiently small with respect to the magnitude of the noise. Can do. Since the noise component is the time average of the square of the amplitude, it does not become 0 as long as noise exists.

The time average R (ω) of the equation (13) is input to the code frequency characteristic extraction unit 50, and the code for each frequency is extracted. Av represents a time average. Each term of equation (13) is a real number, and is a time-averaged real spectrum of the square of the amplitude appearing at the frequency position of each noise spectrum. Therefore, it is possible to execute a positive / negative determination for each frequency at which a value exists. As a result, the code frequency characteristic extraction unit 50 outputs the code frequency characteristic composed of sign (ω) as the time average transfer function W (ω). The time average transfer function W (ω) is expressed by equation (14). In addition, since the sign cannot be determined when the noise superimposed on the upper sideband and the noise superimposed on the lower sideband appear at the same frequency in the baseband, ρ _U and ρ _L overlap in this embodiment. The case is not assumed.

次に、乗算部８１において、位相回転部２６の出力する位相回転直交周波数成分ｊＸ_i （ω）と時間平均伝達関数Ｗ（ω）との周波数毎の積が演算される。その結果、推定同相雑音成分Ｑ（ω）が、（１５）式で求められる。

Next, the multiplication unit 81 calculates a product for each frequency of the phase rotation orthogonal frequency component jX _i (ω) output from the phase rotation unit 26 and the time average transfer function W (ω). As a result, the estimated common-mode noise component Q (ω) is obtained by the equation (15).

Next, the subtracting unit 82 subtracts the estimated in-phase component Q (ω) from the in-phase frequency component X _r (ω) output from the FFT 45 for each frequency. As is clear from the comparison between the equations (10) and (15), the in-phase frequency component Aδ (ω) + S ⁺ (ω) + S ⁻ (ω) from which noise is removed is obtained. This in-phase frequency component is input to IFFT 83 and output as a demodulated signal S (t) on the time axis from which noise has been removed.
Even if the average residual Δ (ω) exists in the time average R (ω), if the absolute value is smaller than the time average of the square of the absolute value of the noise spectrum, the time average R (ω) is The code frequency characteristic can be extracted by extracting the code for frequencies greater than | Δ (ω) |.

In the above embodiment, the phase rotation unit 26 for multiplying the output of the FFT 46 by j is provided. However, the position where j is multiplied may be after the orthogonal component extraction unit 24. That is, even if j is multiplied by the orthogonal component x _i (t) on the time axis, the result is the same. Further, as apparent from the equation (12), the in-phase component x _r (t) that is the output of the in-phase component extraction unit 23 or the in-phase frequency component X _r (ω) that is output from the FFT 45 is multiplied by −j. The same result is obtained. Alternatively, multiplication of j and complex conjugate calculation may be performed on the in-phase frequency component, and the resulting product of (jX _r (ω)) ^* and quadrature frequency component X _i (ω) may be obtained.
Further, the complex product frequency component X _r (ω) (X _i (ω)) ^* or (X _r (ω)) ^* (X _i (ω)) is calculated without performing multiplication of j. Also good. As is apparent from the equation (11), the orthogonal frequency component X _i (ω) is only multiplied by a factor of 1 / j in common, so that the complex product frequency component X _r (ω) (X _i ( ω)) ^* is expressed by an expression obtained by multiplying j in common on the right side of the expression (12). Therefore, the time average R (ω) in the equation (13) and the time average transfer function W (ω) in the equation (14) are also expressed by equations obtained by multiplying each term by j. Since each term in equation (13) is an imaginary number, the sign can be determined. Therefore, when the orthogonal frequency component X _i (ω) is multiplied by the time average transfer function W (ω) for each frequency, the estimated common-mode noise component Q (ω) of the equation (15) is obtained. Therefore, in FIG. 1, even when the phase rotation unit 26 is not present, the in-phase frequency component from which the noise component is removed can be similarly obtained, and the demodulated signal S (t) from which the noise on the time axis is removed is obtained. Can do.
In the present embodiment, it is assumed that the calculation up to the above synthesis is performed in all frequency bands, but only the frequency where the absolute value of the orthogonal frequency component (X _i (ω) is equal to or greater than a predetermined threshold is executed. The predetermined threshold may be set to a value such that the square of the absolute value of the orthogonal noise component is equal to or greater than the absolute value of the average residual Δ (ω). The in-phase component is prevented from being synthesized in the same phase.
In the above embodiment, the upper sideband noise and the lower sideband noise are described as a single line spectrum for the sake of simplicity. However, the above principle holds true for each frequency component, whether it is a large number of line spectrum sequences or a continuous spectrum. Therefore, the present invention can also be applied when the noise is a large number of line spectrum sequences or continuous spectra.
In the above embodiment, it is assumed that the calculation is performed in all positive and negative frequency bands. However, the code frequency characteristic W (ω) is obtained by the above calculation for the positive frequency band, and the code frequency characteristic in the negative frequency band is a complex conjugate thereof. It may be obtained as W (−ω) ^* . Of course, the opposite may be possible.

  Next, an example in which the transfer function is obtained by the ratio of the in-phase frequency component and the orthogonal frequency component will be described. FIG. 4 shows the configuration of the signal processing apparatus according to the second embodiment. Parts having the same functions as those in FIG. 1 are denoted by the same reference numerals. In this embodiment, the configuration of the transfer function calculation unit 70 is different from that of the first embodiment. The transfer function calculation unit 70 includes a ratio calculation unit 71, a time average calculation unit 72, and a code frequency characteristic extraction unit 73.

The in-phase frequency component X _r (ω) is a component obtained by removing the DC component A, and is expressed by a general formula as shown in Equation (16). S (ω) and ρ (ω) are the spectrum of the signal component and the spectrum of the noise component of the in-phase component defined in all positive and negative frequency bands. In addition, when the noise superimposed on the upper sideband and the noise superimposed on the lower sideband do not overlap in the baseband, as described above, the quadrature component is the sign function jsing (ω for each frequency with respect to the in-phase component. Only the factor of) is different. Therefore, the orthogonal frequency component X _i (ω) is expressed by equation (17) using jsing (ω) and ρ (ω).

この比率周波数特性Ｖ（ω）を現時刻ｔに対して過去所定時間Ｔの時間平均が時間平均演算部７２で演算され、ｊsign（ω）は時間的に変動しないので、時間平均Ｒ（ω）は、（１９）式で表される。

（１８）式の第１項は、不規則に変化する信号成分Ｓ（ω）の時間平均であるので、長時間平均すれば、０となる。しかし、仮に、微小量の平均残差Δ（ω）があっても符号周波数特性を求めることができる。したがって、平均残差Δ（ω）の絶対値が１より小さい場合には、時間平均Ｒ（ω）から符号関数｛−jsing(ω) ｝を抽出することができる。時間平均Ｒ（ω）は符号周波数特性演算部７３に入力して、符号関数｛−jsing(ω) ｝が伝達関数Ｚ（ω）として決定される。 In the ratio calculation unit 71, the ratio frequency characteristic V (ω) of the in-phase frequency component X _r (ω) with respect to the quadrature frequency component X _i (ω) is calculated for each frequency as shown in equation (18).

The time average of the ratio frequency characteristic V (ω) with respect to the current time t in the past predetermined time T is calculated by the time average calculator 72, and jsign (ω) does not vary with time, so the time average R (ω) Is expressed by equation (19).

Since the first term of the equation (18) is the time average of the signal component S (ω) that changes irregularly, it becomes 0 when averaged for a long time. However, the code frequency characteristic can be obtained even if there is a minute average residual Δ (ω). Therefore, when the absolute value of the average residual Δ (ω) is smaller than 1, the sign function {−jsing (ω)} can be extracted from the time average R (ω). The time average R (ω) is input to the code frequency characteristic calculation unit 73, and the sign function {−jsing (ω)} is determined as the transfer function Z (ω).

推定同相雑音成分Ｑ（ω）は、同相成分の雑音成分ρ（ω）を表しているので、ＦＦＴ４５から出力される同相周波数成分Ｘ_r （ω）から推定同相雑音成分Ｑ（ω）を減算部８２で減算することで、雑音が除去された信号成分Ｓ（ω）を得ることができる。この信号成分Ｓ（ω）がＩＦＦＴ８３に入力して、時間軸上の雑音が除去された復調信号Ｓ（ｔ）を得ることができる。 Next, the orthogonal frequency component X _i (ω) is multiplied by the transfer function Z (ω) for each frequency by the multiplier 81 to obtain an estimated in-phase noise component Q (ω) expressed by the equation (20). .

Since the estimated in-phase noise component Q (ω) represents the noise component ρ (ω) of the in-phase component, the estimated in-phase noise component Q (ω) is subtracted from the in-phase frequency component X _r (ω) output from the FFT 45. By subtracting at 82, the signal component S (ω) from which noise has been removed can be obtained. This signal component S (ω) is input to IFFT 83, and a demodulated signal S (t) from which noise on the time axis has been removed can be obtained.

When the ratio frequency characteristic V (ω) is obtained by the equation (18), if the orthogonal frequency component X _i (ω) is 0, it diverges. For this reason, when calculating this ratio, the ratio is calculated only for the frequency whose absolute value of the orthogonal frequency component X _i (ω) is equal to or greater than a predetermined value, and for other frequencies, this ratio is 0, that is, the transfer function Z (Ω) is set to 0, or no subtraction operation is performed in the subtraction unit 82 without performing the operation. At this frequency, noise cannot be removed from the in-phase frequency component, but the noise component of the quadrature frequency component is prevented from being superimposed on the in-phase noise in phase.

  The predetermined threshold for determining the frequency when calculating the ratio may be given by the absolute value of the orthogonal frequency component such that the absolute value of the average residual Δ (ω) in equation (19) is smaller than 1. This is because when the absolute value of the average residual Δ (ω) is smaller than 1, the sign function {−jsing (ω)} can be determined from the equation (19).

In addition, at a frequency where the absolute value of the time average R (ω) of the ratio of the equation (19) is smaller than 2, the sign function {−jsing (ω)} can be determined from the sign of the imaginary part. In this case, since it means that the absolute value of the average residual Δ (ω) is smaller than 1, the sign function can be determined. At frequencies where the absolute value of the time average R (ω) is 2 or more, the transfer function Z (ω) is set to 0, or the composite operation of the in-phase frequency component X _r (ω) and the estimated in-phase noise component Q (ω) is performed. By not doing so, in-phase synthesis of noise is prevented as described above.

In addition, when the transfer function is obtained by the ratio calculation, when the upper sideband noise and the lower sideband noise are overlapped at the same frequency, the time average of the ratio is not −jsign (ω), Directly, the transfer function Z (ω) is given as the time average of the ratio. The time average R (ω) is expressed by equation (21), and the transfer function Z (ω) is expressed by equation (22).

However, ρ _r (ω) and ρ _i (ω) are full-band spectra of the in-phase noise component and the quadrature noise component. When the absolute value of the average residual Δ (ω) is small and the time average Av (ρ _r (ω) / ρ _i (ω)) does not change significantly, the estimated common-mode noise component Q (ω) is expressed as ρ _{Since i} (ω) Av (ρ _r (ω) / ρ _i (ω)), the estimated common-mode noise component Q (ω) is approximately ρ _r (ω). In this way, noise can be removed even if the time average R (ω) of the ratio is directly used as the transfer function Z (ω). Of course, at the frequency where the upper sideband noise and the lower sideband noise do not overlap, the transfer function Z (ω) in the equation (22) naturally becomes the sign function {−jsign (ω)}. In-phase noise can be removed from in-phase frequency components.
In the above embodiment, it is assumed that the calculation is performed in all positive and negative frequency bands. However, the sign function {−jsign (ω)} is obtained as the transfer function Z (ω) by the above calculation for the positive frequency band, and the negative frequency is obtained. The code frequency characteristic of the band may be obtained as its complex conjugate Z (−ω) ^* = jsign (ω). Of course, the opposite may be possible.
Further, the ratio calculation unit 71 obtains the ratio frequency characteristic V (ω) of the quadrature frequency component X _i (ω) with respect to the in-phase frequency component X _r (ω), extracts the sign function therefrom, and extracts it as the transfer function. The reciprocal 1 / Z (ω) may be used. Moreover, without extracting the sign function, or multiply the reciprocal 1 / V of the proportional frequency characteristic (omega) to the quadrature frequency components X _i (ω), orthogonal frequency components X _i (ω) The proportional frequency characteristic V ( The estimated common-mode frequency characteristic Q (ω) may be obtained by dividing by ω).
The above is true in the same way. The ratio calculation or the combination calculation may be performed only for the frequency where the absolute value of the orthogonal frequency component is equal to or greater than the predetermined threshold or the frequency where the absolute value of the ratio is greater than 1/2.

  If phase synchronization is not complete in the phase synchronization processing unit 70 and Δω exists, the band of −Δω to Δω centered on the direct current is affected. For this reason, as shown in FIG. 5, a mask processing unit 51 for removing the band may be provided. Also in the case of FIG. 4 of the second embodiment, a mask processing unit 51 may be provided after the ratio calculation unit 71 of FIG.

23 ... In-phase component extraction unit 24 ... Quadrature component extraction unit 60, 70 ... Transfer function calculation unit 80 ... Synthesis unit

Claims (17)

In a signal processing device that receives a double sideband signal and removes noise superimposed on the RF band, the baseband in-phase component having a positive frequency band and a negative frequency band is obtained by quadrature demodulation of the double sideband signal. And a demodulating means for demodulating into a quadrature component,
Fourier transform means for Fourier transforming the in-phase component and the quadrature component to output the in- phase frequency component of the complex function and the quadrature frequency component of the complex function ;
Transfer function computing means for obtaining a time-averaged transfer function between the quadrature frequency component and the in-phase frequency component based on the in-phase frequency component and the quadrature frequency component;
A signal processing apparatus comprising: combining means for correcting the frequency characteristic of the quadrature component based on the transfer function and combining it with the in-phase component. In a signal processing apparatus that receives a double sideband signal and removes noise superimposed on the RF band, the baseband in-phase component having a positive frequency band and a negative frequency band is obtained by quadrature demodulation of the double sideband signal. And a demodulating means for demodulating into a quadrature component,
Fourier transform means for Fourier transforming the in-phase component and the quadrature component to output an in-phase frequency component and a quadrature frequency component;
Transfer function computing means for obtaining a time-averaged transfer function between the quadrature frequency component and the in-phase frequency component based on the in-phase frequency component and the quadrature frequency component;
And combining means for correcting the frequency characteristic of the orthogonal component based on the transfer function and combining it with the in-phase component,
The transfer function calculating means includes
Complex conjugate multiplication means for multiplying one of the in-phase frequency component and the quadrature frequency component and the other complex conjugate for each frequency;
A time average calculating means for calculating a time average of an output of the complex conjugate multiplication means;
Code frequency characteristic extracting means for extracting a code component for each frequency from the output of the time average calculating means, and using the frequency characteristic of the code component as the transfer function;
A signal processing apparatus comprising:   Multiply or divide only one of the quadrature component and the in-phase component output by the demodulating unit by a pure imaginary number j or -j, or the quadrature frequency component and the in-phase frequency component output by the Fourier transform unit 3. The signal according to claim 2, further comprising: a phase rotation unit that uses a frequency component obtained by multiplying or dividing only one of them by a pure imaginary number j or −j as the quadrature frequency component or the in-phase frequency component. Processing equipment. In a signal processing apparatus that receives a double sideband signal and removes noise superimposed on the RF band, the baseband in-phase component having a positive frequency band and a negative frequency band is obtained by quadrature demodulation of the double sideband signal. And a demodulating means for demodulating into a quadrature component,
Fourier transform means for Fourier transforming the in-phase component and the quadrature component to output an in-phase frequency component and a quadrature frequency component;
Transfer function computing means for obtaining a time-averaged transfer function between the quadrature frequency component and the in-phase frequency component based on the in-phase frequency component and the quadrature frequency component;
And combining means for correcting the frequency characteristic of the orthogonal component based on the transfer function and combining it with the in-phase component,
The transfer function calculating means includes
A time average of frequency characteristics obtained by dividing the in-phase frequency component by the orthogonal frequency component for each frequency, or a time average calculating means for calculating a time average of the frequency characteristics obtained by dividing the orthogonal frequency component by the in-phase frequency component;
Code frequency characteristic extracting means for extracting a code component for each frequency from the output of the time average calculating means, and using the frequency characteristic of the code component as the transfer function;
A signal processing apparatus comprising: In a signal processing apparatus that receives a double sideband signal and removes noise superimposed on the RF band, the baseband in-phase component having a positive frequency band and a negative frequency band is obtained by quadrature demodulation of the double sideband signal. And a demodulating means for demodulating into a quadrature component,
Fourier transform means for Fourier transforming the in-phase component and the quadrature component to output an in-phase frequency component and a quadrature frequency component;
Transfer function computing means for obtaining a time-averaged transfer function between the quadrature frequency component and the in-phase frequency component based on the in-phase frequency component and the quadrature frequency component;
And combining means for correcting the frequency characteristic of the orthogonal component based on the transfer function and combining it with the in-phase component,
The transfer function calculating means includes
A time average of frequency characteristics obtained by dividing the in-phase frequency component by the orthogonal frequency component for each frequency, or a time average calculating means for calculating a time average of the frequency characteristics obtained by dividing the orthogonal frequency component by the in-phase frequency component;
Have
The signal processing apparatus, wherein the time-averaged frequency characteristic output from the time-average calculating means is the transfer function. The transfer function calculating means or the synthesizing means performs calculation or synthesis on a frequency component having an absolute value of the orthogonal frequency component equal to or greater than a predetermined threshold value, so that the absolute value of the orthogonal frequency component is smaller than the predetermined threshold value. 6. The signal processing device according to claim 1, wherein no arithmetic operation or synthesis is performed on the signal processing device. The synthesis means includes
When the transfer function calculation means calculates a time average of frequency characteristics obtained by dividing the in-phase frequency component by the orthogonal frequency component for each frequency, only the frequency component whose absolute value of the time average is less than 2 is used. Perform a composite operation,
When the transfer function calculation means calculates the time average of the frequency characteristics obtained by dividing the quadrature frequency component by the in-phase frequency component, only the frequency component whose absolute value of the time average is greater than ½ is combined and calculated The signal processing device according to claim 4, wherein the signal processing device is executed. The synthesis means includes
Equivalent means for correcting the orthogonal frequency component by the transfer function;
Noise removing means for subtracting the corrected quadrature frequency component output from the equivalent means from the in-phase frequency component to output a noise-removed in-phase frequency component;
Inverse Fourier transform means for performing inverse Fourier transform on the noise-removed in-phase frequency component to obtain a demodulated signal on the time axis;
The signal processing apparatus according to claim 1, comprising: In a signal processing apparatus that receives a double sideband signal and removes noise superimposed on the RF band, the baseband in-phase component having a positive frequency band and a negative frequency band is obtained by quadrature demodulation of the double sideband signal. And a demodulating means for demodulating into a quadrature component,
Fourier transform means for Fourier transforming the in-phase component and the quadrature component to output an in-phase frequency component and a quadrature frequency component;
Transfer function computing means for obtaining a time-averaged transfer function between the quadrature frequency component and the in-phase frequency component based on the in-phase frequency component and the quadrature frequency component;
And combining means for correcting the frequency characteristic of the orthogonal component based on the transfer function and combining it with the in-phase component,
The synthesis means includes
An equivalent means for obtaining an orthogonal component on the time axis corrected by convolution of the orthogonal component on the time axis and the impulse response of the transfer function;
Noise removing means for removing the corrected quadrature component on the time axis output from the equivalent means from the in-phase component on the time axis;
A signal processing apparatus comprising: The synthesis means includes
An equivalent means for obtaining an orthogonal component on the time axis corrected by convolution of the orthogonal component on the time axis and the impulse response of the transfer function;
Noise removing means for removing the corrected quadrature component on the time axis output from the equivalent means from the in-phase component on the time axis;
The signal processing apparatus according to claim 2 , wherein the signal processing apparatus includes: During the period from the output of the Fourier transform means to the output of the transfer function calculating unit, according to any one of claims 1 to 10, characterized in that it comprises a mask means for removing a band including a DC Signal processing device. The demodulation means has a phase-locked loop unit that controls the frequency and phase of the demodulated carrier so that the beat signal of the error frequency of the demodulated carrier contained in the quadrature component after quadrature demodulation is zero. the signal processing apparatus according to any one of claims 1 to 11, characterized in. The demodulating means obtains a beat signal of an error frequency of the demodulated carrier wave with respect to the modulated carrier wave from the moving average of the baseband signal, and newly corrects the fluctuation of the baseband signal due to the beat signal based on the beat signal. the signal processing apparatus according to any one of claims 1 to 12 characterized by having a synchronizing means for the baseband signal. 14. The synchronization unit according to claim 13 , further comprising: a unit that corrects a delay of an instantaneous phase caused by a moving average of the baseband signal in accordance with a time difference between the detected instantaneous phase and a moving average. Signal processing device. In a signal processing method for receiving a double sideband signal and removing noise superimposed on an RF band, a baseband in-phase component having a positive frequency band and a negative frequency band is obtained by orthogonally demodulating the double sideband signal. And the quadrature component,
The Fourier transform of the in-phase component and the quadrature component to obtain the in- phase frequency component of the complex function and the quadrature frequency component of the complex function ,
Based on the in-phase frequency component and the quadrature frequency component, a time-averaged transfer function between the quadrature frequency component and the in-phase frequency component is obtained,
Based on the transfer function, the frequency characteristic of the quadrature component is corrected and combined with the in-phase component. A baseband in-phase component having a positive frequency band and a negative frequency band in a signal processing method for receiving a double sideband signal and removing noise superimposed on the RF band by quadrature demodulation of the double sideband signal And the quadrature component,
Fourier transform of the in-phase component and the quadrature component to obtain an in-phase frequency component and a quadrature frequency component,
Based on the in-phase frequency component and the quadrature frequency component, a time-averaged transfer function between the quadrature frequency component and the in-phase frequency component is obtained,
Based on the transfer function, correct the frequency characteristics of the quadrature component and synthesize it into the in-phase component,
A signal processing method, wherein a code component for each frequency is extracted from the time-averaged transfer function, and the frequency characteristic of the code component is used as the transfer function used for correcting the orthogonal component. In the synthesis, a quadrature frequency component obtained by correcting the quadrature frequency component with the transfer function is removed from the common mode frequency component to obtain a noise-removed common mode frequency component,
The signal processing method according to claim 15 or 16 , wherein the noise-removed in-phase frequency component is subjected to inverse Fourier transform to obtain a demodulated signal on a time axis.

Images