JPH09246983A - Digital signal processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent a higher harmonic component extraction part from entering a signal main line and to consider even level fluctuation in a digital signal processor aiming at noise elimination. SOLUTION: RZ digital (time sequential) signals gain controlled in an analog signal level control part 1 are Fourier developed in a fast Fourier transformation part 3. A frequency component detection part 2 extracts signal components from Fourier developed frequency axis signals and controls the analog signal level control part 1 so as to fix the level at all times. Further, the frequency component detection part 2 prepares signals from which higher harmonic components are eliminated and they are subtracted from the frequency axis signals in a subtractor part 4. Thereafter, they are restored to digital (time sequential) signals in an inverse fast Fourier transformation part 5.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital signal processing device, and more particularly to a digital signal processing device for removing noise generated in a digital signal transmission system.

[0002]

2. Description of the Related Art In a signal line in which a signal level fluctuates and has a lot of noise, for example, a DC component such as an RZ (return to zero) signal is large, a required transmission band is wide, and a signal format is weak against noise and level fluctuation For digital signal transmission, it is important to take measures against level fluctuation as well as noise reduction technology.

In recent years, with the development of LSI technology, FF
LSIs for T (Fast Fourier Transform) can be easily obtained, and FFT circuits can be easily manufactured with gate arrays.

Japanese Unexamined Patent Publication No. 1-257224 suggests a noise elimination technique for a digital signal using a fast Fourier transform and an inverse fast Fourier transform. JP 1
FIG. 4 shows an example of a conventional noise removing device disclosed in Japanese Patent Publication No. 257224.

In FIG. 4, an input digital signal which is a time-series signal is Fourier expanded into a frequency component (frequency axis signal) by a fast Fourier transform unit 3. Digital (time series)
When the signal is expanded into frequency components (frequency axis signals), it becomes discrete signals (signal components) composed of harmonic components of the fundamental frequency (clock frequency).

That is, since there is no signal component except near the harmonic component of the fundamental frequency, as shown on the left side of FIG. 5, a window for extracting each harmonic component from the Fourier-transformed signal (see FIG. 5). (Indicated by a broken line in the center of) is created by using, for example, a digital comb filter, and the high-frequency component extraction unit 6 performs Nth-order harmonic window processing (extraction of harmonic signal components), and then the inverse fast Fourier transform unit. Inverse Fourier transform processing is performed at 5 to restore the original digital (time series) signal (right side of FIG. 5).

As described above, the Fourier-transformed digital (time series) signal (frequency axis signal) exists only in the vicinity of the harmonic component of the fundamental frequency, and therefore exists outside the window shown by the broken line in FIG. All components can be considered noise components. Therefore, noise should be completely removed from the digital (time series) signal restored by performing the inverse Fourier transform process after performing the Nth harmonic window process.

However, since N is finite (it is practically impossible to make a window for extracting infinitely many harmonics), the restored digital (time-series) signal has a so-called low-pass characteristic with a slightly blunted waveform. It is inevitable.

[0009]

DISCLOSURE OF THE INVENTION Problems to be Solved by the Invention
In the case of the digital signal processing device described in Japanese Patent Application Laid-Open No. 257224/1998, since the harmonic component extraction unit is inserted in the signal main line, there is a problem that the waveform of the restored digital signal becomes dull, and further the level fluctuation of the signal There is no consideration.

It is an object of the present invention to provide a digital signal processing device in which the harmonic component extraction unit does not enter the signal main line and the level fluctuation is taken into consideration.

[0011]

A digital signal processing apparatus according to the present invention comprises a Fourier transform means for performing a Fourier expansion of an input digital time-series signal into a frequency axis signal and a noise component for removing a signal component from the frequency axis signal. Noise component extraction means for extracting a component, noise subtraction means for subtracting the noise component from the frequency axis signal, and inverse Fourier transform of the frequency axis signal from which the noise component has been subtracted to restore the digital time series signal And an inverse Fourier transform means for performing the inverse Fourier transform.

The digital signal processing device according to the present invention comprises:
Further, a signal component extracting means for extracting a fundamental frequency component of the signal component from the frequency axis signal, and a signal by controlling the gain of the digital time series signal or the frequency axis signal using the extracted fundamental frequency component Gain control means for controlling the level of the component to be constant is included.

[0013]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The operation of the present invention is as follows.
Fast Fourier transform of a digital (time series) signal is performed, the harmonic component is extracted from the frequency axis signal subjected to the fast Fourier transform, the level is detected, and the variable gain circuit is controlled to remove the level fluctuation and Digital (time series) by removing the signal (harmonic) component from the signal, subtracting it from the frequency axis signal, and performing inverse fast Fourier transform
Restore the signal.

An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing a configuration of an embodiment of a digital signal processing device according to the present invention, FIG. 3 is an explanatory diagram of a noise removing function, and the same parts as those in FIG. There is. In addition, overlapping description is omitted.

In FIG. 1, an RZ input digital (time series) signal (FIG. 3A) is applied to an analog signal level control section 1. The analog signal level control unit 1 is a gain control circuit capable of electrically linearly controlling the signal level of an input digital (time series) signal as an analog signal.

The signal whose gain has been controlled by the analog signal level control unit 1 is subjected to Fourier expansion (transformation) by the fast Fourier transform unit 3 so that a digital (time series) signal becomes a frequency domain component (frequency axis signal). Be expanded. The frequency component detection unit 2 is, for example, a digital filter, extracts a fundamental frequency component from the frequency axis signal subjected to Fourier expansion, and controls the analog signal level control unit 1 so that the level of the fundamental frequency component is always constant.

That is, the preset threshold level is compared with the level of the extracted signal component, and feedback control is performed so that this difference becomes zero.

On the other hand, the frequency component detecting section 2 further removes a harmonic component (signal component) from the frequency axis signal by a digital comb filter, for example, to remove the signal (including only the noise component as described above). Then, the subtracter unit 4 subtracts it from the Fourier-developed frequency axis signal output from the fast Fourier transform unit 3.

Therefore, the noise component is subtracted and removed here (FIG. 3 (b)). Thereafter, when the inverse fast Fourier transform unit 5 restores the digital (time series) signal, noise is removed, and the digital (time series) signal from which the level fluctuation is also removed (FIG. 3C) is faithfully restored.

In the case of the embodiment of the present invention shown in FIG. 1, since the frequency component detector 2 performs feedback control so that the output (signal) level of the analog signal level controller 1 becomes constant, the fast Fourier transform is performed. Since the full input dynamic range of the conversion unit 3 can be used efficiently (the input digital signal level is controlled to be constant, the input digital signal level can be selected to be almost the same as the input dynamic range). Although a high fast Fourier transform is possible, an analog gain control circuit (having some unstable elements as compared with a digital circuit) remains in the analog signal level control unit 1 and it is difficult to realize an LSI.

In the case of another embodiment of the present invention shown in FIG. 2, since the RZ input digital signal is directly added to the fast Fourier transform unit 1 (while the signal level variation is included), the signal level variation is taken into consideration. Since it is necessary to select the input digital signal level of the fast Fourier transform unit 1 so as not to exceed the input dynamic range, the use efficiency of the dynamic range is unavoidably lowered, but the frequency component detection unit 2 causes the fundamental frequency to decrease. The unstable element of the analog circuit can be removed by detecting the signal level of the component and digitally correcting the signal level fluctuation in the digital signal multiplier section 7 so that this signal level becomes equal to the set level.
It can be easily integrated into an LSI.

[0023]

As described above, according to the present invention, after the Fourier expansion for noise removal, the harmonic component extraction unit that causes rounding of the signal waveform does not enter the signal main line, and Fourier expansion is performed even for level fluctuations. By extracting the level of the signal component and correcting the signal level, it is possible to obtain a highly accurate noise removal circuit from which level fluctuations have been removed.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of the present invention.

FIG. 2 is a block diagram of another embodiment of the present invention.

FIG. 3 is an explanatory diagram of a noise removal method.

FIG. 4 is a block diagram showing an example of a conventional noise removal circuit using Fourier transform.

FIG. 5 is a diagram for explaining the operation of a conventional noise removal circuit using Fourier transform.

[Explanation of symbols]

 1 analog signal level control unit 2 frequency component detection unit 3 fast Fourier transform unit 4 subtractor unit 5 inverse fast Fourier transform unit 7 digital signal multiplier unit

Claims (3)

[Claims] 1. A Fourier transform means for performing a Fourier expansion of an input digital time series signal into a frequency axis signal, a noise component extracting means for removing a signal component from the frequency axis signal and extracting a noise component, and the frequency axis. It is characterized by including noise subtraction means for subtracting the noise component from the signal, and inverse Fourier transform means for inverse Fourier transforming the frequency axis signal from which the noise component is subtracted to restore the digital time series signal. Digital signal processor. 2. A signal component extracting means for extracting a fundamental frequency component of the signal component from the frequency axis signal,
Gain control means for controlling the gain of the digital time series signal or the frequency axis signal using the extracted fundamental frequency component so that the level of the signal component becomes constant. Item 1. The digital signal processing device according to Item 1. 3. The digital signal processing apparatus according to claim 1, wherein the noise component extracting means and the signal component extracting means are constituted by digital comb filters.