JP4381876B2 - Digital frequency converter and digital frequency converter - Google Patents

Description

  The present invention relates to a digital frequency converter and a digital frequency converter for performing a desired signal frequency in the digital domain.

In recent years, highly advanced digital signal processing techniques are applied to many wireless transmission systems to which Code Division Multiple Access (CDMA) and Orthogonal Frequency Division Multiplexing (OFDM) are applied.
FIG. 9 is a diagram illustrating a configuration example of a transmitter in which digital signal processing is applied to generation of a transmission wave.

In the figure, the output of the baseband modulation processing unit 51 is connected to the input of the mixer 53 via the low-pass filter 52, and the output of the local oscillator 54 is connected to the local oscillation input of the mixer 53. The output of the mixer 53 is connected to a power amplifier (not shown) or an antenna system via a band filter 55.
In the transmitter having such a configuration, the FPGA (Field Programmable Gate Array) and the LSI (Large Scale Integration) provided in the baseband modulation processing unit 51 perform the modulation, An analog baseband signal is generated by performing processing such as code spreading and D / A conversion.

  For example, as shown in FIG. 10 (a), this baseband signal has a “zero-order hold-like waveform” in which the instantaneous value increases or decreases stepwise in synchronization with the clock signal used for the D / A conversion described above. Therefore, as shown in FIG. 10B, many harmonic components are included. Moreover, since the above-described baseband signal is generated as a train of pulses having a finite pulse width equal to the above-described clock signal period Ts, the frequency spectrum of these harmonic components is, for example, It does not become the same value regardless of the order (FIG. 10 (c)), and as the order becomes higher as shown by the dotted line in FIG. 10 (b), it decreases and becomes an integral multiple of the frequency fs of this clock signal. Has a minimum point (zero point) in frequency.

The low-pass filter 52 extracts the occupied band component of the baseband signal and removes the above-described harmonic component to generate a “baseband transmission wave signal”.
The local oscillator 54 steadily generates a local oscillation signal whose frequency is a specified value ft. The mixer 53 generates a signal having the frequency of the sum or difference of both frequencies by taking the product of the baseband transmission wave signal and the local oscillation signal. The band filter 55 extracts a component of the transmission wave signal distributed in a desired occupied band from this signal, and removes a spurious signal and a harmonic component that are not preferably added to the transmission wave signal, thereby converting the transmission wave signal. At the same time, the transmission wave signal is supplied to the above-described power amplifier or antenna system.

In addition, as a prior art relevant to this invention, there exists the following technique, for example.
A “digital modulator” that realizes quadrature modulation at high speed by utilizing a parallel / serial conversion circuit instead of a digital multiplier, as disclosed in Patent Document 1 described later.
As disclosed in Patent Document 2 to be described later, a high-speed quadrature modulation is realized by utilizing a parallel / serial conversion circuit instead of a digital multiplier, and the occupied band of the baseband signal is equal to the frequency of the clock signal. “Modulator” shifted to a higher frequency equal to a quarter
As disclosed in Patent Document 3 to be described later, quadrature modulation / demodulation is realized as signal processing in the complex domain, and a signal generated as a result of these quadrature modulation / demodulation is a signal having a desired frequency via an analog frequency converter. Complex modulation / demodulation system converted to
JP-A-8-23359 (Summary) JP-A-8-149170 (Summary) Japanese Patent Laid-Open No. 6-244883 (Summary)

By the way, in the above-described conventional example, the transmission wave signal supplied to the power amplifier and the antenna system is likely to have characteristic variations and fluctuations, as in the mixer 53, the local oscillator 54, and the band filter 55 described above, and there is nothing. It was difficult to make adjustments, and because of the large number of parts, it was generated by an analog circuit that is likely to hinder reliability improvement and high-density mounting.
However, such an analog circuit generally requires a lot of cost for improving and stabilizing the characteristics, and it is difficult to change the characteristics easily. Therefore, various characteristics such as wide band are flexible and inexpensive. It was a great hindrance to the realization of software defined radio technology.

  SUMMARY OF THE INVENTION An object of the present invention is to provide a digital frequency converter and a digital frequency conversion device that can achieve a desired frequency conversion accurately and stably by digital signal processing.

  In the first invention, the Fourier transform means Fourier transforms the discrete signal f (t) given by the sampling frequency fs in the order of the time series t in the digital domain, and from (−fs / 2) to (fs) on the frequency axis. Frequency component F (p) distributed in a plurality of p different bands belonging to the band up to / 2). The spectrum rearrangement unit interchanges the order on the frequency axis between the high-frequency group and the low-frequency group in which the frequency components F (p) distributed in a plurality of p different bands are divided into two. The inverse Fourier transform means performs inverse Fourier transform in the digital domain on the two groups whose order has been changed by the spectrum rearrangement means.

  The frequency spectrum of the discrete time signal obtained as a result of the inverse Fourier transform includes the above-described low-frequency group (or its harmonics) and the above-described high frequency adjacent to such a group on the frequency axis. Obtained as a set with a group of bands (or harmonics of the group), and a frequency that is an integral multiple of the sampling frequency fs is the lower end of the band in which these low band groups (or harmonics of the group) are distributed. Or it does not correspond except for the upper end of the band in which the high-frequency group (and its harmonics) are distributed.

  Therefore, the frequency conversion of the discrete signal f (t) is adjacent to the frequency axis in ascending order on the frequency axis in this way, and the sampling frequency fs is distributed in a band that cannot correspond to a frequency other than the lower end or the upper end frequency. If it is possible to extract a pair of a high frequency group (or a harmonic of the group) and a high frequency group (or a harmonic of the group), a Fourier transform unit, a spectrum rearrangement unit, and an inverse Fourier transform unit It is realized with high accuracy under the processing of the digital domain performed as described above.

  In the second invention, the Fourier transform means Fourier-transforms the discrete signal f (t) given at the sampling frequency fs in the order of the time series t in the digital domain, and from (−fs / 2) to (fs) on the frequency axis. Frequency component F (p) distributed in a plurality of p different bands belonging to the band up to / 2). The spectrum rearrangement means bisects the frequency component F ′ (p) distributed in the occupied band of the discrete signal f (t) among the frequency components F (p) described above at the lower end and the upper end of these bands. The high-frequency group and low-frequency group are rearranged. The inverse Fourier transform unit performs inverse Fourier transform on the two groups arranged at the lower end and the upper end by the spectrum rearrangement unit in the digital domain.

  The frequency spectrum of the discrete time signal obtained as a result of the inverse Fourier transform is a band in which the occupation band of the above-described discrete signal f (t) extends from (−fs / 2) to (fs / 2) on the frequency axis. Even if it corresponds to a part of the above, the low-frequency group (or its harmonics) and the above-mentioned high-frequency group (or its harmonics) adjacent to such a group on the frequency axis And a frequency that is an integral multiple of the sampling frequency fs is the lower end of the band in which these low-frequency groups (or harmonics of the group) are distributed, and the high-frequency group (and its It does not match other than the upper end of the band in which the group harmonics are distributed.

  Therefore, the frequency conversion of the discrete signal f (t) is thus adjacent to the frequency axis in ascending order on the frequency axis, and the sampling frequency fs is distributed in a band where the sampling frequency fs cannot correspond to a frequency other than the lower end or the upper end frequency. Can be extracted by a Fourier transform means, a spectrum relocation means, and an inverse Fourier transform means. It is realized with high accuracy under the processing of the digital domain performed as described above.

In the third invention, in the digital frequency converter according to the second invention, the low band group and the high band group are given to the sampling frequency fs and the occupied band width fb: 1: (fs / This is a component of each band obtained by dividing the occupied band by the ratio of fb-1).
When the low band group and the high band group are specified by dividing the occupied band of the discrete signal f (t) at such a ratio, on the frequency axis, “the low band group” Both (the lower end of the first band where the (or higher harmonics of the group) component is distributed) and "the upper end of the second band where the higher band group (or higher harmonics of the group) are distributed" And the distance from “the integral multiple of the sampling frequency fs and the frequency closest to the lower end and the upper end” are set to the same value.

Therefore, the result of the D / A conversion performed at the subsequent stage of the digital frequency converter according to the present invention is obtained as “a pulse train whose pulse width is not“ 0 ”(not an ideal impulse)”. Thus, distortion of the frequency spectrum that occurs in a band in the vicinity of a frequency that is an integral multiple of the sampling frequency fs is alleviated.
According to a fourth invention, in the digital frequency converter according to any one of the first to third inventions, the inverse Fourier transform means precedes the inverse Fourier transform, and the low frequency group and the high frequency group on the frequency axis. The boundary between the groups is changed, and the excess or deficiency of the bandwidth between these groups is corrected on a virtual annular frequency axis in which the upper end of the low band group and the lower end of the high band group are continuous.

  That is, when the direction and frequency in which the boundary between the low frequency group and the high frequency group is changed are appropriately given, the low frequency group (or the harmonic of the group) extracted as a result of the frequency conversion is given. Wave) and high frequency groups (or harmonics of the high frequency group), these frequencies do not include the sampling frequency fs, and are further performed after the filter that extracts the result of the frequency conversion. A band suitable for conversion and other processing is set.

Therefore, it is possible to reduce the scale and power consumption of the hardware arranged at the subsequent stage of the digital frequency converter according to the present invention, and to reduce the cost and size.
In the fifth invention, in the digital frequency converter according to the first to fourth inventions, the inverse Fourier transform means is configured such that the discrete output signal generated as a result of the inverse Fourier transform is a train of pulses having a finite pulse width. Due to being given, the high frequency group and the low frequency group are respectively multiplied by the reciprocal of the ratio of errors occurring in the frequency spectrum prior to the inverse Fourier transform.

That is, the above-described error is not included in the frequency spectrum of the band extracted as a result of the frequency conversion of the discrete signal f (t).
Therefore, the degree of freedom related to the selection of the sampling frequency fs is improved and the accuracy of frequency conversion is increased.
In the sixth invention, the plurality of digital frequency converters correspond to the digital frequency converters according to any of the first to fifth inventions described above, and are given by the sampling frequency fs in the order of time series t. The discrete signal f (t) is individually responded with different phases. The selection means sequentially selects the partial discrete output signals that ensure the desired accuracy under different phase combinations among the multiple discrete output signals obtained as a result of the inverse Fourier transform individually by the multiple digital converters. To do.

The partial discrete signals sequentially selected in this way are either or both of the start points and the end points of a plurality of point sequences to be subjected to Fourier transform in the order of time series in the point sequence indicating the discrete signal f (t). Even if one of the values is not “0”, it is obtained as a point sequence that accurately follows the discrete signal f (t).
Therefore, according to the present embodiment, the accuracy and linearity of frequency conversion are improved and maintained high regardless of the permutation of the values of the point sequence indicating the discrete signal f (t).

As described above, in the first and second inventions, the frequency conversion is realized with high accuracy under the digital domain processing performed as described above by the Fourier transform unit, the spectrum rearrangement unit, and the inverse Fourier transform unit. .
In the third invention, the result of the D / A conversion performed after the digital frequency converter according to the present invention is “a pulse train whose pulse width is not“ 0 ”(not an ideal impulse)”. As a result, the distortion of the frequency spectrum that occurs in a band in the vicinity of a frequency that is an integral multiple of the sampling frequency fs is alleviated.

Furthermore, in the fourth invention, it is possible to reduce the scale and power consumption of the hardware arranged at the subsequent stage of the digital frequency converter according to the present invention, and to reduce the cost and size.
In the fifth invention, the degree of freedom related to the selection of the sampling frequency fs is improved, and the accuracy of frequency conversion is increased.
Furthermore, in the sixth invention, the accuracy and linearity of the frequency conversion are improved and kept high regardless of the permutation of the values of the discrete signal f (t) given as a point sequence.

  Therefore, in an electronic device or system to which the present invention is applied, the configuration can be simplified and the performance can be stabilized without increasing the cost, and flexible adaptation to various specifications or changes in specifications is possible. Become.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[First embodiment]
FIG. 1 is a diagram showing first to third embodiments of the present invention.
In the figure, the output of the baseband signal processing unit 10 is connected to the input of a bandpass filter 20, and the output of the bandpass filter 20 is connected to the input of an unillustrated power amplifier or an antenna system.

The baseband signal processing unit 10 includes a baseband modulation processing unit 12, an FFT unit 13, a connection conversion unit 14, an IFFT unit 15, and a D / A conversion unit 16 that are connected in cascade.
FIG. 2 is a diagram for explaining the operation of the first embodiment of the present invention.
The operation of the first embodiment of the present invention will be described below with reference to FIG. 1 and FIG.

The baseband modulation processing unit 12 generates a baseband signal in the digital domain by performing processing such as modulation, code spreading, and A / D conversion in addition to transmission path coding and interleaving of transmission information.
Each unit of the baseband signal processing unit 10 operates in synchronization with a clock signal having a frequency fs.

  As shown in FIG. 2 (a), the spectrum of the baseband signal described above is assumed to be distributed above and below the origin on the frequency axis, and the rearrangement of frequency slots to be described later is clarified. In addition, it is assumed that the power is smaller at the lower end and the upper end of the band (−fs / 2) to (fs / 2) where the sampling theorem is established with respect to the frequency fs of the clock signal than the other bands. .

The FFT unit 13 performs a fast Fourier transform on the baseband signal, and as shown in FIG. 2B, a sequence of a plurality of N (= 16) frequency slots indicating the frequency spectrum of the baseband signal in the digital domain. Is generated.
In FIG. 1, for these frequency slots, suffix numbers “−8”, “−7”,..., “0”, “+1”,. "FS-8", "FS-7", ..., "FS0", "FS + 1", ..., "FS + 7".

The connection conversion unit 14 divides these 16 frequency slots into “low group” (FIG. 1 (1)) composed of “FS-8” to “FS-1” in ascending order on the frequency axis, and “FS0” to It is divided into “high group” (FIG. 1 (2)) consisting of “FS + 7”, and the arrangement of these “low group” and “high group” on the frequency axis is switched (FIG. 1 (3), FIG. 2). (c)).
The IFFT unit 15 performs a baseband signal different from the above-described baseband signal by performing inverse fast Fourier transform on the set of the “high group” and the “low group” arranged higher than the “high group”. A signal (hereinafter referred to as “quasi-baseband signal”) is generated.

The D / A converter 16 D / A-converts the quasi-baseband signal in synchronization with the clock signal, thereby generating an analog signal having a waveform approximated by a staircase wave with the period of the clock signal.
The frequency spectrum of such an analog signal is generally weighted with a smaller power as the absolute value of the frequency is higher on the frequency axis as shown in FIG. 2 (d). However, the frequency spectrum is an integer multiple of the frequency fs of the clock signal. (Hereinafter referred to as “polar frequency”) and a group of harmonic components of these groups together with a high group distributed in a low band adjacent to the low band and a low group distributed in a high band adjacent to the pole frequency. As obtained.

The band-pass filter 20 extracts a low-group harmonic component and a high-group harmonic component distributed in a band twice or three times the frequency fs of the clock signal from the analog signal components (FIG. 2 ( e)) A harmonic signal composed of these harmonic components is supplied as a transmission wave signal to the above-described power amplification and antenna system.
That is, the components of the transmission wave transmitted through these power amplifiers and antenna systems are distributed in a band that is sandwiched between two polar frequencies 2fs and 3fs on the frequency axis and does not include any of the polar frequencies.

  Therefore, according to the present embodiment, the analog signal output by the D / A converter 16 under the complex operation performed in the digital domain by the FFT unit 13 and the IFFT 15 being linked via the connection converter 14. Are reliably distributed in a band that does not include the pole frequency, and are extracted by the band-pass filter 20 to be used for generation of a transmission wave having a desired frequency and power.

In addition, since the complex operation described above is realized regardless of the applied modulation scheme, multiple access scheme, channel arrangement, and frequency arrangement, in the electronic apparatus to which the present embodiment is applied, the hardware digitization ratio Is increased.
[Second Embodiment]
FIG. 3 is a diagram for explaining the operation of the second embodiment of the present invention.

The operation of the second embodiment of the present invention will be described below with reference to FIGS.
The feature of this embodiment lies in the following processing performed by the baseband modulation processing unit 12 and the connection conversion unit 14.
The baseband modulation processing unit 12 generates a baseband signal in the digital domain in the same manner as in the first embodiment described above. The occupied band of this baseband signal has a frequency as shown in FIG. It is distributed in a single sideband from “0” to (fs / 2) on the axis.

  The FFT unit 13 performs a fast Fourier transform on the baseband signal, and as shown in FIG. 3 (b), a plurality of N (= 16) frequency slots “FS” indicating the frequency spectrum of the baseband signal in the digital domain. -8 ”,“ FS-7 ”,...,“ FS0 ”,“ FS + 1 ”,...,“ FS + 7 ”(here, in the same manner as in the first embodiment described above (−fs / 2 ) Through (fs / 2) is assumed to have a bandwidth of (fs / 16) divided into 16 equal parts).

The connection converter 14 assigns these frequency slots (“FS-8”, “FS-7”,..., “FS0”, “FS + 1”,..., “FS + 7”) to the first described above. Similar to the embodiment, it is divided into a “low group” (FIG. 1 (4)) and a “high group” (FIG. 1 (5)). Since it does not correspond to a large occupied band, it is discarded.
Moreover, the connection conversion part 14 performs the following process.
Of the 16 inputs of the IFFT unit 15, the “high group” is equally divided into the ascending order of the frequency by the 1st to 4th inputs and the 13th to 16th inputs corresponding to the ascending order of the frequencies. “Sub-low group” (set of “FS0” to “FS + 3”) (FIG. 1 (6)) and “sub-high group” (set of “FS + 4” to “FS + 7”) ) (FIG. 1 (7)), respectively (FIG. 1 (8), FIG. 3 (c)).
A constant “0” is given to the remaining inputs (fifth to twelfth inputs) of the IFFT unit 15 (FIG. 1 (9)).

  The IFFT unit 15 includes the above-mentioned “sub-high group”, “sub-low group” arranged higher than this “high group”, and these “sub-high group” and “sub-low group” on the frequency axis. A baseband signal different from the above-described baseband signal (hereinafter referred to as “quasi-standard”) is obtained by performing an inverse fast Fourier transform on a set of “invalid group” having a value set to the above constant “0”. Baseband signal ”).

The D / A converter 16 D / A-converts the quasi-baseband signal in synchronization with the clock signal, thereby generating an analog signal having a waveform approximated by a staircase wave with the period of the clock signal.
In general, the frequency spectrum of such an analog signal is weighted smaller as the absolute value of the frequency is higher on the frequency axis, as shown in FIG. Obtained as a set of high and sub-low groups and the harmonic components of these groups.

The band-pass filter 20 extracts the harmonic components of the sub-high group and the sub-low group that are distributed in a band that is two to three times the frequency fs of the clock signal among the frequency components of the analog signal (FIG. 3 (e )), A modulated wave signal composed of these harmonic components is supplied to the power amplification and antenna system described above.
Therefore, according to the present embodiment, the components of the analog signal generated by the D / A conversion unit 16 are the FFT unit 13 and the IFFT 15 even when the occupied band of the baseband signal is distributed in a single sideband. Are linked as described above via the connection converter 14, so that they are arranged in a band sandwiched between two polar frequencies 2 fs and 3 fs on the frequency axis.

In the present embodiment, the invalid group component (noise, etc.) included in the baseband signal is set to a constant “0” prior to the inverse fast Fourier transform performed by the IFFT unit 15.
Therefore, compared to the case where such processing is not performed, the steep frequency characteristics required for the vicinity of the boundary between the pass band and the stop band of the band filter 20 are allowed, and the cost is reduced. It is possible to reduce the size and weight.

In this embodiment, the sub-low group for the bisection of the single sideband of the baseband signal and the upper and lower ends of the bands (−fs / 2) to (fs / 2) where the sampling theorem holds. And the sub-high group are arranged by the connection converter 14.
However, the present invention is not limited to the case where the occupied band of the baseband signal is configured as a double sideband or a single sideband, and power can be distributed throughout the occupied band. Even if is discontinuous or can change, it is applicable under the following conditions.
・ The low group (sub-low group), the high group (sub-high group), and these harmonics are distributed in frequencies (bands) other than the extreme frequency, Frequency slots belonging to a high group (sub-high group) are allocated.
-Invalid groups (and harmonics of these invalid groups) in which no effective power is distributed in the occupied band are arranged on the frequency axis at and near the polar frequency.

In the present embodiment, the sub-low group and the sub-high group are configured by allocating four frequency slots included in the above-described high group in an ascending order of frequencies.
However, the present invention is not limited to such a configuration. For example, 1: {(fs / fb) for the frequency fb at the upper end of the high group (sub-high group) and the frequency fs of the clock signal described above. −1} in the ratio of the frequency slots included in the high group to the sub-low group and the sub-high group, “the lower end of the first band in which the harmonic components of the sub-high group are distributed” , “The upper end of the second band where the harmonic components of the sub-low group are distributed” and the distance on the frequency axis from each nearest pole frequency are set to the same value, so that the transmission quality is reduced. It may be kept to a minimum.
[Third embodiment]
The third embodiment of the present invention will be described below.

The operation of this embodiment will be described below with reference to FIGS.
The feature of the present embodiment is the following processing procedure performed by the connection conversion unit 14.
A deviation Δf on the frequency axis of a band in which it is desirable that the harmonic components of the sub-low group and the sub-high group described above are distributed is given to the connection conversion unit 14 in advance as a known value.
As for such deviation Δf, hereinafter, as shown in FIGS. 4A and 4B, the above-described band and the above-described harmonic component in the second embodiment described above are distributed. Assume that it is given as an integer i (−4 ≦ i ≦ 4) equal to the ratio (= δ / w) of the distance δ on the frequency axis with respect to the “standard band” and the width w of the frequency slot (= δ / w).

Similarly to the second embodiment, the connection conversion unit 14 discards the “low group” and once identifies the “sub-low group” and the “sub-high group”. The frequency slot combinations included in the “sub-low group” and “sub-high group” are corrected as follows.
When the above-described integer i is a positive number, “add a frequency slot located at the lower end of the“ sub-high group ”as a frequency slot adjacent to the lower frequency slot at the upper end of the“ sub-high group ”, The process of decrementing the integer i is repeated until the integer i becomes “0”.
When the above-mentioned integer i is a negative number, “add a frequency slot positioned at the upper end of the“ sub-low group ”as a frequency slot adjacent to the high frequency band of the lower frequency slot of the“ sub-high group ”, The process of decrementing the integer i is repeated until the integer i becomes “0” (FIGS. 4C to 4E).

The IFFT unit 15 includes the above-mentioned “sub-high group”, the “sub-low group” arranged higher than the “sub-high group”, and these “sub-high group” and “sub-low group” on the frequency axis. A baseband signal (hereinafter referred to as “quasi-baseband signal”) different from the above-described baseband signal is generated by performing an inverse fast Fourier transform on a set of “invalid group” sandwiched between
The D / A converter 16 D / A-converts the quasi-baseband signal in synchronization with the clock signal, thereby generating an analog signal having a waveform approximated by a staircase wave with the period of the clock signal.

The main frequency component of such an analog signal is obtained as a harmonic component of the adjacent sub-high group and sub-low group, and shifted on the frequency axis by the sum of the widths of the number of frequency slots equal to the integer i. Distributed in the band.
That is, if the above-described deviation Δf is appropriately given, the occupied band of the analog signal output by the band filter 20 is not only obtained without including the polar frequency component, but also shown in FIG. As described above, the band is set to a band suitable for frequency conversion and other processing performed in the subsequent stage of the band filter 20.

Therefore, according to the present embodiment, the processing amount of the digital signal processing unit 10 is effectively utilized for cost reduction and miniaturization in accordance with the reduction in hardware scale and power consumption.
Although this embodiment is applied to the second embodiment described above, it can be similarly applied to a fourth embodiment and a fifth embodiment described later.
[Fourth embodiment]
FIG. 5 is a diagram showing a fourth embodiment of the present invention.

This embodiment is composed of the following elements.
Cascade-connected baseband modulation processing unit 12, FFT unit 13, and connection conversion unit 14
A multiplier 31-(-N / 2) arranged at the subsequent stage of the connection converter 14 and individually corresponding to the above described (-N / 2) to (N / 2-1) frequency slots. ~ 31- (N / 2-1)
A coefficient setting unit 32 having N outputs individually connected to the multiplier inputs of these multipliers 31-(-N / 2) to 31- (N / 2-1)
IFFT section 15 having outputs individually connected to the inputs of multipliers 31-(-N / 2) to 31- (N / 2-1)
A D / A conversion unit 16 and a band filter 20 cascaded to the output of the IFFT unit 15
In FIG. 5, the baseband modulation processing unit 12, the FFT unit 13, the connection conversion unit 14, the multipliers 31-(-N / 2) to 31- (N / 2-1), the IFFT unit 15, and the D described above. The digital signal processing unit composed of the / A conversion unit 16 and the coefficient setting unit 32 is denoted by the reference numeral “10A”.

FIG. 6 is a diagram for explaining the operation of the fourth embodiment of the present invention.
The operation of the fourth embodiment of the present invention will be described below with reference to FIGS.
A feature of the present embodiment is that the following processing is performed by the coefficient setting unit 32 and the multipliers 31-(-N / 2) to 31- (N / 2-1).
The envelope component H (f) of the frequency spectrum of the analog signal, which is output by the D / A converter 16 and obtained by the waveform approximated by the staircase wave with the period Ts (= 1 / fs) of the clock signal, is Since the analog signal corresponds to a pulse train having a pulse width equal to the cycle Ts of the clock signal (different from an ideal impulse pulse width “0”), for example, as shown by a dotted line in FIG. It has a sharp minimum point (zero point) at the extreme frequency and decreases as the frequency increases.

Such an envelope component H (f) is generally given by the following equation with respect to the frequency f and the period Ts of the clock signal described above.
H (f) = (Ts · sinπfTs) / πfTs
Therefore, when the multipliers 31-(-N / 2) to 31- (N / 2-1) are not provided in the previous stage of the IFFT unit 15, a plurality of frequency slots input in parallel to the IFFT unit 15 are provided. Such an envelope component in between is an index n indicating the individual frequency slots in ascending order of frequency (with respect to the total number N of these frequencies (= 16), (−N / 2), (−N / 2 + 1) ,..., 0,..., (N / 2-2) or (N / 2-1))) and the harmonic order k with respect to the baseband are given by It is done.
H (fn) = [Ts · sinπ (kfs + fn) Ts] / π (kfs + fn) Ts
The coefficient setting unit 32 causes the multipliers 31-(-N / 2) to 31- (N / 2-1) to convert Cn expressed by the following equation with respect to such an envelope component H (fn) Correction coefficients C _{-N / 2} , C _{-N / 2 + 1} ,..., C ₀ ,..., C _{N / 2-2} , C _{N / 2-1} (FIG. 6A) shown in ascending order are given.
Cn = 1 / H (fn)
The multipliers 31-(-N / 2) to 31- (N / 2-1) are generated by the FFT unit 13 and rearranged by the connection conversion unit 14, and their correction coefficients C. The product of _{−N / 2} , C _{−N / 2 + 1} ,..., C ₀ ,..., C _{N / 2-2} , C _{N / 2-1} is given to the IFFT unit 15.

That is, the frequency spectrum of the quasi-baseband signal generated by the IFFT unit 15 is subjected to weighting for correcting the frequency spectrum error generated by the D / A conversion unit 16 in advance (FIG. 6). (b)).
As described above, according to the present embodiment, the frequency spectrum of the signal output from the band filter 20 is shown by the dotted line in FIG. 10C in the entire occupied band of the signal, as shown in FIG. It is obtained without any error due to the weighting shown.

Therefore, even when the frequency fs of the clock signal is not necessarily sufficiently high with respect to the width of the occupied band, the degree of freedom related to the selection of the frequency fs is secured and the accuracy of frequency conversion is increased.
In the present embodiment, multipliers 31-(-N / 2) to 31- (N / 2-1) are arranged between the connection conversion unit 14 and the IFFT unit.

However, the present invention is not limited to such a configuration. For example, the multipliers 31-(-N / 2) to 31- (N / 2-1) are arranged between the FFT unit 13 and the connection conversion unit 14. In the case of being arranged, in conjunction with the rearrangement of the frequency slots performed by the connection conversion unit 14, “there are given to these multipliers 31-(− N / 2) to 31- (N / 2-1)”. “Correction coefficients” may be rearranged.
In this embodiment, the values of the correction coefficients C _{-N / 2} , C _{-N / 2 + 1} , ..., C ₀ , ..., C _{N / 2-2} , C _{N / 2-1} are as described above. It is obtained in advance on the assumption that the result of the fast Fourier transform is obtained as the (1/2) th power value of the power included in each frequency slot.

However, the values of these correction coefficients C _{-N / 2} , C _{-N / 2 + 1} , ..., C ₀ , ..., C _{N / 2-2} , C _{N / 2-1} are, for example, stored in the FFT unit 13. When the power included in each frequency slot is directly obtained by an alternative Fourier transform unit, the square value of the above-described value may be applied.
Furthermore, in the present embodiment, the correction coefficient _{C -N / 2, C -N /} 2 + 1, ..., C 0, ..., C N / 2-2, the value of C N _{/ 2-1,} bandpass filter 20 Is obtained in advance as a value suitable for the pass band of, and is given as a constant to the multipliers 31-(-N / 2) to 31- (N / 2-1) by the coefficient setting unit 32.

However, the present invention is not limited to such a configuration, and correction coefficients C _{-N / 2} , C _{-N / 2 + 1} , ..., C ₀ , ..., C _{N / 2-2} , C _{N / 2-1} The value of may be updated one by one to a value suitable for the pass band of the band pass filter 20, for example.
[Fifth embodiment]
FIG. 7 is a diagram showing a fifth embodiment of the present invention.

The present embodiment includes a digital signal processing unit 10B including the following elements, and a band filter 20 connected in cascade to the digital signal processing unit 10B.
The digital signal processing unit 10B includes the following elements.
Baseband modulation processing unit 12
FFT units 13-1 and 13-2 connected in parallel to the output of the baseband modulation processing unit 12
Connection connection unit 14-1 and IFFT unit 15-1 that are cascade-connected to the output of the FFT unit 13-1
Connection connection unit 14-2 and IFFT unit 15-2 connected in series to the output of the FFT unit 13-2
A switch 41 having two inputs connected to the outputs of these IFFT units 15-1 and 15-2, respectively.
A D / A converter 16 that is cascade-connected to the output of the switch 41 and whose output is connected to the input of the band-pass filter 20 described above.
FIG. 8 is a diagram (2) for explaining the operation of the fifth embodiment of the present invention.

The fast Fourier transform is generally obtained with high accuracy on the assumption that the values of the discrete time function to be subjected to the fast Fourier transform are both “0” at the start point and the end point on the time axis.
However, for example, the baseband signal input to the FFT unit 13 in the first embodiment described above is actually given as a sequence of a plurality of N (= 16) values that are consecutive in time series. The values at the beginning and end of each column are not necessarily “0”.

Hereinafter, a system constituted by the FFT unit 13-1, the connection conversion unit 14-1, and the IFFT unit 15-1 is referred to as a “first system”, and the FFT unit 13-2, the connection conversion unit 14−. 2 and the IFFT unit 15-2 are referred to as a “second system”.
The first system is preceded by 8Ts (= (N / 2) Ts) on the time axis than the second system, and is a sequence of N (= 16) instantaneous values (hereinafter referred to as “first” By repeatedly performing the same processing as that performed by the FFT unit 13, the connection conversion unit 14, and the IFFT unit 15 in the first embodiment described above. A band signal (hereinafter referred to as a “first quasi-baseband signal”) is generated (FIG. 8A).

In the second system, a sequence of N (= 16) instantaneous values (hereinafter referred to as 8Ts (= (N / 2) Ts) on the time axis after the first FFT frame described above). , “Second FFT target frame”) is repeatedly captured and a quasi-baseband signal (hereinafter referred to as “second quasi-baseband signal”) is generated in the same manner as the first sequence (see FIG. 8 (b)).
The switch 41 synchronizes with the above-described clock signal and the above-described first and second FFT target frames, and the fifth to twelfth instantaneous values of the first quasi-baseband signal described above, The fifth to twelfth instantaneous values of the quasi-baseband signal are selected (FIG. 8C), and a sequence of these instantaneous values (hereinafter referred to as “specific instantaneous values”) is input to the D / A converter 16. give.

Of the instantaneous values of the first and second quasi-baseband signals, the first to fourth and thirteenth to sixteenth instantaneous values that do not correspond to the specific instantaneous value described above (shown with hunting in FIG. 8). ) May include more errors due to the above-mentioned assumption not being established than these specific instantaneous values.
However, as shown in the lower part of FIG. 8, the specific instantaneous values are sequentially obtained without overlapping and lacking on the time axis, and even if the above errors are included, the ratio of the errors is small.

That is, the instantaneous value of the analog signal output via the D / A conversion unit 16 and the band filter 20 corresponds to the baseband signal generated by the baseband modulation processing unit 12 with high accuracy.
Therefore, according to the present embodiment, the accuracy of frequency conversion is further improved and maintained high.

In the present embodiment, the first system and the second system are arranged in parallel downstream of the baseband modulation processing unit 12.
However, the present invention is not limited to such a configuration, and the configuration is the same as that of these systems in the subsequent stage of the baseband modulation processing unit 12, for example, in addition to the first system and the second system described above. In addition, a single system or a plurality of systems in which a part of the period of the FFT frame on the time axis is common to a part of the period of the FFT frame of any other system are connected in parallel. Among the instantaneous values of the quasi-baseband signal obtained at all outputs of the system, the specified instantaneous value is assured that the possibility of including the above-mentioned error is small or that the ratio of the error is small enough to be allowed Is selected by the switch 41, the accuracy and linearity of the frequency conversion may be further enhanced.

In each of the above-described embodiments, the frequency slots (“FS-8”, “FS-7”,..., “FS0”, “FS + 1”,..., “FS + 7”) The quasi-baseband signal is set to a desired position on the frequency axis with respect to the baseband signal.
However, the present invention is not limited to such a configuration. For example, the permutation of frequency slots may be changed as a process suitable for ensuring secrecy in the transmission system, the form of association with the receiving end, and the like.

Further, in each of the above-described embodiments, the present invention is applied to realize frequency conversion for changing the occupied band from the baseband region to a desired band.
However, the present invention is not limited to such frequency conversion, and can be similarly applied to frequency conversion from the intermediate frequency band to a desired band as long as the above-described processing in the digital domain can be realized.

Further, in each of the above-described embodiments, the present invention is applied to frequency conversion of a baseband signal having a limited occupation band.
However, the present invention can be applied regardless of the modulation scheme and the multiple access scheme applied to the generation of such a baseband signal, and the unmodulated unit that may be distributed in a known band. The present invention can also be applied to frequency conversion of one or a plurality of carrier signals.

Furthermore, the present invention can be applied to a software defined radio that is required to flexibly adapt to various changes such as a modulation scheme, a multiple access scheme, a frequency arrangement, and a channel configuration without changing the hardware configuration. Is preferred.
Further, the present invention is not limited to a wireless transmission system such as a mobile communication system or a satellite communication system, and can be applied to various devices in which frequency conversion is performed internally, for example.

Further, in each of the above-described embodiments, the processing for setting the band in which the second-order and third-order harmonic components of the baseband signal are distributed to the band sandwiched between two pole frequencies adjacent on the frequency axis is performed in the digital domain. In other words, a part of these harmonic components is extracted by the band-pass filter 20, whereby the frequency conversion is realized.
However, the order of such harmonic components may be higher if the desired level and quality are ensured.

In the above-described embodiments, the size of the fast Fourier transform performed by the FFT unit 13 (13-1, 13-2) is set to “16”.
However, such a size achieves a desired accuracy for achieving frequency conversion, and can be realized within the hardware configuration of the digital signal processing units 10, 10 </ b> A, and 10 </ b> B described above or in the range of surplus processing. If so, it may be set to a value larger than “16”.

Further, in each of the above-described embodiments, the fast Fourier transform and the fast inverse Fourier transform are performed, so that the frequency analysis of the baseband signal and the result of the frequency analysis are replaced and the inverse Fourier transform after the rearrangement is performed. ing.
However, the present invention is not limited to such a configuration, the desired accuracy for achieving the frequency conversion is achieved, and the hardware configuration of the digital signal processing units 10, 10A, and 10B described above or redundant processing is achieved. If it is feasible in the range, either or both of the following configurations may be applied.
A configuration in which a discrete Fourier transform (DFT) or a comb-type digital filter is applied instead of the fast Fourier transform. Further, the present invention is not limited to the above-described embodiment, and various forms of embodiments are possible within the scope of the present invention, and a part or all of the constituent devices are applicable. Improvements may be made.

Hereinafter, the invention disclosed in each of the above-described embodiments is arranged hierarchically and multifacetedly and listed as an additional item.
(Supplementary note 1) A discrete signal f (t) given at a sampling frequency fs in the order of time series t is Fourier-transformed in the digital domain, and on the frequency axis, the band extends from (−fs / 2) to (fs / 2). Fourier transform means for obtaining a frequency component F (p) distributed in different bands of a plurality of p belonging to,
Spectrum re-arrangement means for switching the order on the frequency axis between the high-frequency group and the low-frequency group obtained by dividing the frequency components F (p) distributed in the different bands of the plurality p;
A digital frequency converter, comprising: inverse Fourier transform means for performing inverse Fourier transform in the digital domain of the two groups whose order has been changed by the spectrum rearrangement means.
(Supplementary Note 2) The discrete signal f (t) given at the sampling frequency fs in the order of the time series t is Fourier-transformed in the digital domain, and in the frequency band (−fs / 2) to (fs / 2). Fourier transform means for obtaining a frequency component F (p) distributed in different bands of a plurality of p belonging to,
A high-frequency group and a low-frequency group in which the frequency component F ′ (p) distributed in the occupation band of the discrete signal f (t) is divided into two at the lower end and the upper end of the band. Spectrum rearrangement means for rearranging each group of regions,
A digital frequency converter, comprising: inverse Fourier transform means for performing inverse Fourier transform on the two groups arranged at the lower end and the upper end by the spectrum rearrangement means in a digital domain.
(Appendix 3) In the digital frequency converter described in Appendix 2,
The low-frequency group and the high-frequency group are:
It is a component of each band obtained by dividing the occupied band at a ratio of 1: (fs / fb-1) given to the sampling frequency fs and the width fb of the occupied band. Digital frequency converter.
(Appendix 4) In the digital frequency converter according to any one of appendices 1 to 3,
The inverse Fourier transform means includes
Prior to the inverse Fourier transform, the boundary between the low frequency group and the high frequency group on the frequency axis is changed, and the upper end of the low frequency group and the lower end of the high frequency group are continuous. A digital frequency converter characterized by correcting the excess or deficiency of bandwidth between these groups on a virtual annular frequency axis.
(Appendix 5) In the digital frequency converter according to any one of appendices 1 to 4,
The inverse Fourier transform means includes
Due to the fact that the discrete output signal generated as a result of the inverse Fourier transform is given as a train of pulses having a finite pulse width, the inverse of the ratio of errors occurring in the frequency spectrum is preceded by the inverse Fourier transform. A digital frequency converter characterized in that the high frequency group and the low frequency group are respectively multiplied.
(Supplementary note 6) A plurality of digital signals described in any one of supplementary notes 1 to 5 that individually respond with discrete phases to discrete signal f (t) given at sampling frequency fs in the order of time series t A frequency converter;
Selection for sequentially selecting partial discrete output signals that ensure desired accuracy under a combination of different phases, among a plurality of discrete output signals obtained individually as a result of inverse Fourier transform by the plurality of digital converters And a digital frequency converter.

It is a figure which shows 1st thru | or 3rd embodiment of this invention. It is a figure explaining operation | movement of 1st embodiment of this invention. It is a figure explaining operation | movement of 2nd embodiment of this invention. It is a figure explaining operation | movement of 3rd embodiment of this invention. It is a figure which shows 4th embodiment of this invention. It is a figure explaining operation | movement of 4th embodiment of this invention. It is a figure which shows 5th embodiment of this invention. It is a figure explaining the operation | movement of pseudo-fifth embodiment of this invention. It is a figure which shows the structural example of the transmitter to which the digital signal processing was applied to the production | generation of a transmission wave. It is a figure explaining operation | movement of a prior art example.

Explanation of symbols

10, 10A, 10B, 51 Digital signal processing unit 12 Baseband modulation processing unit 13 FFT unit 14 Connection conversion unit 15 IFFT unit 16 D / A conversion unit 20, 55 Band filter 31 Multiplier 32 Coefficient setting unit 41 Switch 52 Low band Filter 53 Mixer 54 Local oscillator

Claims (3)

A discrete signal f (t) given at the sampling frequency fs in the order of time series t is Fourier transformed in the digital domain, and a plurality of p belonging to the transmission band from (−fs / 2) to (fs / 2) on the frequency axis Fourier transform means for obtaining frequency components F (p) distributed in different bands of
Spectrum re-arrangement means for switching the order on the frequency axis between the high-frequency group and the low-frequency group obtained by dividing the frequency components F (p) distributed in the different bands of the plurality p;
A digital frequency converter, comprising: inverse Fourier transform means for performing inverse Fourier transform in the digital domain of the two groups whose order has been changed by the spectrum rearrangement means. A discrete signal f (t) given at the sampling frequency fs in the order of time series t is Fourier transformed in the digital domain, and a plurality of p belonging to the transmission band from (−fs / 2) to (fs / 2) on the frequency axis Fourier transform means for obtaining frequency components F (p) distributed in different bands of
A high-frequency group obtained by dividing frequency components F ′ (p) distributed in the occupied band of the discrete signal f (t) among the frequency components F (p) at the lower end and the upper end of the transmission band; Spectrum rearrangement means for rearranging the low-frequency group, and
A digital frequency converter, comprising: inverse Fourier transform means for performing inverse Fourier transform on the two groups arranged at the lower end and the upper end by the spectrum rearrangement means in a digital domain. A plurality of digital frequency converter individually responsive, and according to claim 1 or claim 2 in discrete signal f (t) in different phases given by the sampling frequency fs in order of time sequence t,
Selection for sequentially selecting partial discrete output signals that ensure desired accuracy under a combination of different phases, among a plurality of discrete output signals obtained individually as a result of inverse Fourier transform by the plurality of digital converters And a digital frequency converter.

Images