(19)
(11)EP 2 658 124 B1

(12)EUROPEAN PATENT SPECIFICATION

(45)Mention of the grant of the patent:
20.05.2020 Bulletin 2020/21

(21)Application number: 11851747.3

(22)Date of filing:  18.08.2011
(51)International Patent Classification (IPC): 
H03H 17/02(2006.01)
(86)International application number:
PCT/JP2011/069098
(87)International publication number:
WO 2012/086262 (28.06.2012 Gazette  2012/26)

(54)

DIGITAL FILTER CIRCUIT AND DIGITAL FILTER CONTROL METHOD

DIGITALFILTERSCHALTUNG UND DIGITALFILTERSTEUERVERFAHREN

CIRCUIT DE FILTRE NUMÉRIQUE ET PROCÉDÉ DE COMMANDE D'UN FILTRE NUMÉRIQUE


(84)Designated Contracting States:
AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR

(30)Priority: 21.12.2010 JP 2010284137

(43)Date of publication of application:
30.10.2013 Bulletin 2013/44

(73)Proprietor: NEC Corporation
Tokyo 108-8001 (JP)

(72)Inventor:
  • SHIBAYAMA, Atsufumi
    Tokyo 108-8001 (JP)

(74)Representative: Betten & Resch 
Patent- und Rechtsanwälte PartGmbB Maximiliansplatz 14
80333 München
80333 München (DE)


(56)References cited: : 
WO-A1-00/19654
JP-A- 2010 232 857
JP-A- 8 023 262
  
  • GRANDMAISON M-E ET AL: "Frequency domain filter using an accurate reconfigurable FFT/IFFT core", CIRCUITS AND SYSTEMS, 2004. NEWCAS 2004. THE 2ND ANNUAL IEEE NORTHEAST WORKSHOP ON MONTREAL, CANADA 20-23 JUNE 2004, PISCATAWAY, NJ, USA,IEEE, 20 June 2004 (2004-06-20), pages 165-168, XP010742477, DOI: 10.1109/NEWCAS.2004.1359048 ISBN: 978-0-7803-8322-7
  • John G. Proakis ET AL: "Filtering of Long Data Sequences" In: "Digital Signal Processing; Principles, Algorithms, and Applications", 1 January 1996 (1996-01-01), Prentice Hall, Inc., Upper Saddle River, New Jersey, USA, XP055059889, ISBN: 978-0-13-373762-2 pages 430-433, * page 430 - page 431; figure 5.10 *
  • John G. Proakis ET AL: "6.2.3 Use of the FFT Algorithm in Linear Filtering and Correlation" In: "Digital Signal Processing; Principles, Algorithms, and Applications", 1 January 1996 (1996-01-01), Prentice Hall, Inc., New Jersey, USA, XP055530174, ISBN: 978-0-13-373762-2 pages 477-479,
  
Note: Within nine months from the publication of the mention of the grant of the European patent, any person may give notice to the European Patent Office of opposition to the European patent granted. Notice of opposition shall be filed in a written reasoned statement. It shall not be deemed to have been filed until the opposition fee has been paid. (Art. 99(1) European Patent Convention).


Description

TECHNICAL FIELD



[0001] The present invention relates to an arithmetic processing circuit in digital signal processing and relates to a digital filter circuit and a digital filter control method in particular.

BACKGROUND ART



[0002] As a technology which compensates for waveform distortion in signal transmission in wireless communication and cable communication, a frequency domain equalization (FDE) technology is known. In FDE, first, a signal data in the time domain is converted to data in the frequency domain by Fast Fourier Transform (FFT). Then filter processing for equalization is performed and the signal data in the frequency domain is reconverted to the signal data in the time domain by Inverse FFT (IFFT). In FDE, by the above-mentioned processing, the waveform distortion of the signal is compensated for.

[0003] Meanwhile, in FFT, it is assumed that a signal repeats periodically, it is a problem that a computation distortion occurs in a signal around both ends of a processing block of FFT. As a technology which solves this problem, for example, an overlap FDE method disclosed in patent documents 1, 2 and 3 can be mentioned. In the overlap FDE method, FFT processing is performed by giving an overlap of a predetermined data between adjacent processing blocks, and then after performing filter processing and reconverting to a time domain signal by IFFT processing, only the signal data of the both ends of the processing block in which the computation-distortion occurs is removed.

[0004] As the other related arts, a digital filter device, described in patent document 4, which changes the sampling period of digital data is mentioned.

[0005] Document "Frequency domain filter using an accurate reconfigurable FFT/IFFT core", by Grandmaison M-E et al, published during the 2nd Annual IEEE Northeast Workshop on Circuits and Systems, NEWCAS 2004, in Montreal, Canada on 20-23 June 2004, discloses a frequency domain filter architecture with emphasis on bus width requirements to minimize quantization noise. The focus is on the design of efficient and reconfigurable FPGA implementations of frequency-domain filters. The hardware requirements and accuracy performances for a typical communication filter, the raised cosine filter, are presented.

[PRIOR ART DOCUMENT]


[PATENT DOCUMENT]



[0006] 

[Patent document 1] Japanese Patent Application Laid-Open No. 2006-304192

[Patent document 2] Japanese Patent Application Laid-Open No. 2010-130355

[Patent document 3] Japanese Patent Application Laid-Open No. 2007-201523

[Patent document 4] Japanese Patent Application Laid-Open No. 1997-135151


SUMMARY OF THE INVENTION


[PROBLEM TO BE SOLVED BY THE INVENTION]



[0007] However, in the overlap FDE method, at a part of overlapping FFT processing, it is necessary to overlap same input signals and perform the FDE processing. In other words, because of increase in the amount of operations to the overlapped part, the FDE processing requires a higher throughput of processing with the overlapped part than the throughput of the input-output of the signal data.

[0008] For example, when the FFT block size is 4096 data and the number of overlap is 1024 data the processing throughput of FDE needs to be 4/3 times the input-output throughput of the data.

[0009] As a method to improve the processing throughput of FDE for the increase in the amount of the operation of this overlapped part, the method comprising the step of parallelizing a FDE processing circuit can be considered. However, the method comprising the step of parallelizing the circuit has a problem that the circuit scale and the electric power consumption increase.

[0010] An object of the present invention is to provide a digital filter circuit and a digital filter control method capable of solving the problem mentioned above.

[MEANS FOR SOLVING THE PROBLEM]



[0011] The present invention is defined in the independent claims. The dependent claims define advantageous embodiments thereof.

[0012] A digital filter circuit according to the present invention includes: an overlap addition means for generating a block including N data (N is a positive integer) from input data in a time domain by giving an overlap of M data (M is a positive integer) between the block and the previous block; an FFT processing means for transforming the generated block to a block in a frequency domain by performing FFT processing; a filter computation means for performing filter processing to the transformed block in the frequency domain and outputting a filtered block, an IFFT means for transforming the filtered block to a block in the time domain by IFFT processing; an overlap removal means for generating an output signal by removing a total of M data from both ends of the transformed block in the time domain and generating an output signal; and a clock generation means for setting a frequency of a filter processing clock signal based on the number M of overlaps, wherein the filter processing clock signal drives the data output unit of the overlap addition means, the FFT means , the filter computation means, the IFFT means, and the data input unit of the overlap removal means. The number N is determined based on the number M so that the power consumption is minimized, and the frequency of the filter processing clock signal is determined based on the number N and the number M. The clock generation means sets the frequency of the filter processing clock signal at N/(N-M) times a frequency of an input-output clock signal which drives a data input unit of the overlap addition means and a data output unit of the overlap removal means.

[0013] A digital filter control method according to the present invention includes: generating a block including N units of data (N is a positive integer) from input data in a time domain by performing overlap addition processing which causes the block to give an overlap of M data (M is a positive integer) between the block and the previous block; transforming the generated block to a block in a frequency domain by performing FFT processing; performing filter processing to the transformed block in the frequency domain and generating a filtered block, transforming the filtered block to a block in the time domain by IFFT processing; performing overlap removal processing which generates an output signal by removing a total of M data from both ends of the transformed block in the time domain; and setting the frequency of a filter processing clock signal based on a value of the number M of the overlaps, wherein the filter processing clock signal relates to the output of the overlap addition processing, the FFT processing, the filter processing, the IFFT processing, and the input of the overlap removal processing. The number N is determined based on the number M, so that the power consumption is minimized, and the frequency of the filter processing clock signal is determined based on the number N and the number M. The frequency of the filter processing clock signal is set to N/(N-M) times a frequency of an input-output clock signal which relates to the input of the overlap addition processing and the output of the overlap removal processing.

[EFFECT OF THE INVENTION]



[0014] According to the present invention, in filter processing in a frequency domain of an overlap FDE method, the circuit scale and the electric power consumption can be reduced.

BRIEF DESCRIPTION OF THE DRAWINGS



[0015] 

[Fig. 1] is a block diagram showing an exemplary configuration of a digital filter circuit.

[Fig. 2] is a sequence diagram showing an example of operation of the digital filter circuit shown in Fig. 1.

[Fig. 3] is a timing chart showing an example of operation of the digital filter circuit shown in Fig. 1.

[Fig. 4] is a block diagram showing an exemplary configuration of a digital filter circuit according to a first exemplary embodiment of the present invention.

[Fig. 5] is a graph showing an example of power consumption of the digital filter circuit shown in Fig. 4.

[Fig. 6] is a table diagram showing an exemplary configuration of a setup table provided in the digital filter circuit shown in Fig. 4.

[Fig. 7] is a block diagram showing an exemplary configuration of a digital filter circuit.


EXEMPLARY EMBODIMENTS FOR CARRYING OUT OF THE INVENTION



[0016] Fig. 1 is a block diagram showing an exemplary configuration of a digital filter circuit 100. The digital filter circuit 100 is a frequency domain filter circuit which performs filter processing in a frequency domain. Specifically, first, after transforming a signal, in a time domain, inputted as an input data to data in a frequency domain by FFT, the digital filter circuit 100 performs filter processing. After that, the digital filter circuit 100 retransforms the data to a signal data in the time domain by IFFT, and outputs the signal data as an output data.

[0017] For example, the digital filter circuit 100 refers to an overlap number setting signal 28 which is given from a higher level circuit (not shown) such as a CPU (Central Processing Unit) or the like, and determines the number of the overlap.

[0018] The digital filter circuit 100 includes an overlap addition circuit 10, a FFT circuit 11, a filter computation circuit 12, an IFFT circuit 13, an overlap removal circuit 14 and a clock generation circuit 15 at least.

[0019] The overlap addition circuit 10 successively generates a block including N data from an input data which is an input signal on the time domain and outputs the data to the FFT circuit 11. At that time, the overlap addition circuit 10 gives an overlap of M data between each block and a previous block. Here, both of N and M are positive integers together. The number M of overlapped data is determined based on the overlap number setting signal 28. The overlap addition circuit 10 inputs an input data synchronizing with a clock D. And the overlap addition circuit 10 outputs the generated block to the FFT circuit 11 synchronizing with a filter clock 25.

[0020] Further, the overlap addition circuit 10 can be composed by a dual port memory, for example. At that time, writing to the dual port memory may be performed synchronizing with the clock D, and reading from the dual port memory may be performed synchronizing with the filter clock 25.

[0021] The FFT circuit 11 performs FFT to the input signal, in the time domain, outputted from the overlap addition circuit 10 and overlapped with M data, so that the input signal is transformed into a signal in the frequency domain and outputs the signal to the filter computation circuit 12. The FFT circuit 11 operates synchronizing with the filter clock 25.

[0022] The filter computation circuit 12 performs filter processing to the signal, in the frequency domain, transformed by the FFT circuit 11, and outputs the signal to the IFFT circuit 13. For example, the filter computation circuit 12 can be composed by a complex number multiplier when the digital filter circuit 100 carries out equalization processing of signal distortion in a communication channel. The filter computation circuit 12 operates synchronizing with the filter clock 25.

[0023] The IFFT circuit 13 performs IFFT to the signal, in the frequency domain, outputted from the filter computation circuit 12 after filter processing, so that the signal is retransformed in a signal in the time domain and outputs the signal in the time domain to the overlap removal circuit 14. The IFFT circuit 13 operates synchronizing with the filter clock 25.

[0024] The overlap removal circuit 14 removes a total of M data from the both ends of each block including N data that are the signal, in the time domain, retransformed by the IFFT circuit 13 and takes out only a midsection of the block, and outputs the midsection of the block as an output data. The number M of removed data is determined based on the overlap number setting signal 28.

[0025] The overlap removal circuit 14 inputs data from the IFFT circuit 13 synchronizing with the filter clock 25. And the overlap removal circuit 14 outputs the output data synchronizing with the clock D.

[0026] Further, the overlap removal circuit 14 can be composed by a dual port memory, for example. At that time, writing to the dual port memory may be performed synchronizing with the filter clock 25, and reading from the dual port memory may be performed synchronizing with the clock D.

[0027] The clock generation circuit 15 generates the filter clock 25 which is a new clock signal from a clock S and outputs the filter clock 25 to the overlap addition circuit 10, the FFT circuit 11, the filter computation circuit 12, the IFFT circuit 13 and the overlap removal circuit 14.

[0028] The clock generation circuit 15 determines a frequency of the filter clock 25 based on the inputted overlap number setting signal 28.

[0029] Further, when the frequency of the clock S is higher than that of the filter clock 25, for example, the clock generation circuit 15 can be composed by a frequency divider and can generate the filter clock 25 by dividing the clock S. On the other hand, when the frequency of the clock S is lower than that of the filter clock 25, for example, the clock generation circuit 15 can be composed by a PLL (Phase Locked Loop) circuit, and can generate the filter clock 25 by multiplying the clock S.

[0030] Hereinafter, operation of the digital filter circuit 100 shown in Fig. 1 will be described with reference to Fig. 2 and Fig. 3. Fig. 2 is a sequence diagram showing an example of operation of the digital filter circuit shown in Fig. 1. Fig. 3 is a timing chart showing an example of operation of the digital filter circuit shown in Fig. 1. In the following description, processing steps (1)-(5) correspond to processing steps (1)-(5) in Fig. 2 and Fig. 3, respectively.

(1) Overlap Addition Processing



[0031] The overlap addition circuit 10 successively generates a block including N data from an input data which is an input signal in the time domain. At that time, the overlap addition circuit 10 gives an overlap of each of the blocks and a previous block for only M data.

[0032] When the input data is set to x [i] (i=0, 1, ...),
a block of N data can be expressed by
x [j] (j=m(N-M)-N to m(N-M)-1).

[0033] Here, N is FFT block size and M is the number of overlaps. M is a positive integer.

(2) FFT Processing



[0034] The FFT circuit 11 performs a Fast Fourier Transform (FFT) to a block including signal data in the time domain and transforms the block into a block including a signal data in the frequency domain.

[0035] When the block including the signal N data in the time domain is anew set to
x [n] (n=0, 1... , N-1),
a block X[k] in the frequency domain after FFT processing can be given by Equation 1.


(3) Frequency Domain Filter Processing



[0036] The filter computation circuit 12 performs filter processing to each signal data, in the frequency domain, which composes a block after FFT processing.

[0037] A block X'[k] after filter processing to a block X[k] before filter processing is given by



[0038] In Equation 2, H (k) indicates a filter factor.

(4) IFFT Processing



[0039] The IFFT circuit 13 performs IFFT to a block including a signal data in the frequency domain after filter processing and retransforms the block to a block including a signal data in the time domain.

[0040] A block y[n] after IFFT processing for the block X'[k] before IFFT processing is given by Equation 3.


(5) Overlap Removal Processing



[0041] The overlap removal circuit 14 extracts a midsection from the block y[n] including N signal data after IFFT processing leaving M/2 overlapped data in the head and end of the block and generates y'[j] (j=M/2toN-1-M/2) signal data sequence of (N-M) units by removing the overlaps.

[0042] As mentioned above, the digital filter circuit 100 adds an overlap of M data to N data and performs filter processing, and removes the overlap of M data again. Accordingly, the processing throughput, which is higher for the overlap than the throughputs of the input and the output of data, is required for the filter processing. Specifically, the processing throughput, which is N/ (N-M) times or more the throughputs of the input and the output of data, is required for the filter processing.

[0043] In this example, the throughputs of the input and the output of data in the digital filter circuit 100 are determined by the frequency of the clock D. The reason is that the data input in the overlap addition circuit 10 and the data output in the overlap removal circuit 14 operate synchronizing with the clock D.

[0044] On the other hand, the processing throughput of the filter processing in the digital filter circuit 100 is determined by the frequency of the filter clock 25. The reason is that the data output in the overlap addition circuit 10, the data input in the overlap removal circuit 14, the FFT circuit 11, the filter computation circuit 12 and the IFFT circuit 13 operate synchronizing with the filter clock 25.

[0045] Here, in this example, the filter processing is performed with the throughput of N/(N-M) times as many as the throughputs of the data input and the data output by making the frequency of the filter clock 25 N/(N-M) times the clock D.

[0046] Specifically, when a frequency of the clock D is set to fD, the FFT block size is set to N, and the number of overlap is set to M, the frequency fF of the filter clock 25 is set by the following Equation 4.



[0047] At that time, the digital filter circuit 100 performs filter processing in every cycle of the filter clock 25 to the inputted data in every cycle of the clock D, and outputs the data, for which filter processing was performed, in every cycle of the clock D.

[0048] Generally, when throughputs of both input/output data and filter processing are not identical in a digital filter circuit, flow controls are needed, such as a control of suspension and starting of an input-output of data and a control of suspension and starting of filter processing. On the other hand, according to the digital filter circuit 100 of this example, this flow control is unnecessary. Thereby, because the circuit is simple, circuit scale and power consumption can be reduced.

[0049] Specifically, for example, when the overlap addition circuit 10 and the overlap removal circuit 14 are realized by a dual port memory, it does not occur that the memory is full of data, or empty thereof. Accordingly, a flow control which suspends or starts writing into the memory and reading from the memory is unnecessary. Because the circuit can be simplified, as a result, the circuit scale and the power consumption can be reduced.

[0050] According to the example described above, in the frequency domain filter processing using overlap FFT, in accordance with the required number M of overlaps and the FFT block size N, the operation frequency of the filter processing is set to be N/ (N-M) times as high as the operating frequency of the input-output of data.

[0051] Therefore, the digital filter circuit 100 according to this example operates in accordance with the required number of overlaps by the operating frequency enough necessary to perform filter processing to an input signal, which is inputted, without delaying. For this reason, the power consumption can be reduced.

[0052] Moreover, because a flow control for an input-output of data and filter processing is not needed, the required memory size can be reduced and the circuit configuration can be simplified in the digital filter circuit 100. Therefore, the circuit scale and the power consumption can be reduced.

[FIRST EXEMPLARY EMBODIMENT]



[0053] Fig. 4 is a block diagram showing an exemplary configuration of a digital filter circuit 101 according to a first exemplary embodiment of the present invention. The digital filter circuit 101 is a frequency domain filter circuit which performs filter processing in a frequency domain like the digital filter circuit 100 of the first exemplary embodiment. Basic operation of a frequency domain filter circuit has been already described in the previous example.

[0054] The digital filter circuit 101 refers to the overlap number setting signal 28 which is given from a higher level circuit such as a CPU like the digital filter circuit 100 of the previous example and determines the number of overlap.

[0055] The feature of the digital filter circuit 101 is that a setup table 36 is added to the digital filter circuit 100 of the previous example. Further, while the basic function of each component besides the setup table 36 is identical with each component in the digital filter circuit 100 of Fig. 1, it is but different from the previous example in a part of functions according to an addition of the setup table 36. Accordingly, codes which are different from Fig. 1 will be attached in Fig. 4 about each above-mentioned component.

[0056] The digital filter circuit 101 includes at least an overlap addition circuit 30, a FFT circuit 31, a filter computation circuit 32, an IFFT circuit 33, an overlap removal circuit 34, a clock generation circuit 35 and a setup table 36.

[0057] The overlap addition circuit 30 generates successively a block including N data from an input data which is an input signal in the time domain and outputs the block to the FFT circuit 31. At that time, the overlap addition circuit 30 gives an overlap of M data between each of the blocks and a previous block. Further, like in the previous example, both N and M are positive integers. The number N of data which compose a block is determined based on a FFT block size setting signal 26 which is outputted from the setup table 36. Further, the number M of the overlapped data is determined based on the overlap number setting signal 28. The overlap addition circuit 30 inputs an input data synchronizing with the clock D. Further the overlap addition circuit 30 outputs the block which was generated synchronizing with the filter clock 25 to the FFT circuit 31.

[0058] Further, for example, the overlap addition circuit 30 can be composed by a dual port memory. At that time, writing to the dual port memory may be performed synchronizing with the clock D, and reading from the dual port memory may be performed synchronizing with the filter clock 25.

[0059] The FFT circuit 31 performs FFT to the input signal, in the time domain, outputted from the overlap addition circuit 30 and overlapped with M data, so that the input signal is transformed into a signal in the frequency domain and outputs the signal to the filter computation circuit 32. At that time, the process unit (the block size) of FFT is determined based on the FFT block size setting signal 26 which is outputted from the setup table 36. Here, the block size can be changed easily by changes of the number of rows of the butterfly circuit which composes the FFT circuit 31. The FFT circuit 31 operates synchronizing with the filter clock 25.

[0060] The filter computation circuit 32 performs filter processing to the signal in the frequency domain transformed by the FFT circuit 31, and outputs the signal to the IFFT circuit 33. For example, the filter computation circuit 32 can be composed by a complex number multiplier when the digital filter circuit 101 carries out equalization processing of signal distortion in a communication channel. At that time, the process unit (the block size) of the filter processing is determined based on the FFT block size setting signal 26 which is outputted from the setup table 36. The filter computation circuit 32 operates synchronizing with the filter clock 25.

[0061] The IFFT circuit 33 performs IFFT to the signal, in the frequency domain, outputted from the filter computation circuit 32 after filter processing, so that the signal is retransformed in the frequency in a signal in the time domain and outputs the signal to the overlap removal circuit 34. At that time, the process unit (the block size) of IFFT is determined based on the FFT block size setting signal 26 which is outputted from the setup table 36. Here, the block size can be changed easily by changes of the number of rows of the butterfly circuit which composes the IFFT circuit 33. The IFFT circuit 33 operates synchronizing with the filter clock 25.

[0062] The overlap removal circuit 34 removes a total of M data from the both ends of each block including N data which is the signal in the time domain retransformed by the IFFT circuit 33 and takes out only a midsection in the block, and outputs the midsection of the block as an output data. The number N of data of N which compose the block is determined based on the FFT block size setting signal 26 which is outputted from the setup table 36. Further, the number M of the removed data is determined based on the overlap number setting signal 28.

[0063] The overlap removal circuit 34 inputs data from the IFFT circuit 33 synchronizing with the filter clock 25. And the overlap removal circuit 34 outputs the output data synchronizing with the clock D.

[0064] Further, the overlap removal circuit 34 can be composed by a dual port memory, for example. At that time, writing to the dual port memory may be performed synchronizing with the filter clock 25, and reading from the dual port memory may be performed synchronizing with the clock D.

[0065] The clock generation circuit 35 generates the filter clock 25 which is a new clock signal from a clock S and outputs the filter clock 25 to the overlap addition circuit 30, the FFT circuit 31, the filter computation circuit 32, the IFFT circuit 33 and the overlap removal circuit 34.

[0066] A frequency of the filter clock 25 is determined based on the filter clock frequency setting signal 27 which the setup table 36 outputs. When the frequency of the clock S is higher than that of the filter clock 25, for example, the clock generation circuit 35 can be composed by a frequency divider and can generate the filter clock 25 by dividing the clock S. On the other hand, when the frequency of the clock S is lower than that of the filter clock 25, for example, the clock generation circuit 35 can be composed by a PLL (Phase Locked Loop) circuit, and can generate the filter clock 25 by multiplying the clock S.

[0067] The setup table 36 holds a set value of the FFT block size in accordance with the inputted overlap number setting signal 28 and outputs the set value of the FFT block size to the overlap addition circuit 30, the FFT circuit 31, the IFFT circuit 33 and the overlap removal circuit 34. The setup table 36 holds a set value of the filter clock frequency in accordance with the setting of the inputted number of the overlaps and outputs the set value of the filter clock frequency to the clock generation circuit 35.

[0068] Hereinafter, operation of the digital filter circuit 101 shown in Fig. 4 will be described. The operation regarding filter processing of the digital filter circuit 101 is equal to the operation regarding the filter processing of the digital filter circuit 100 of the previous example shown in Fig. 2 and Fig. 3 by the processing steps (1)-(5).

[0069] In other words, the digital filter circuit 101 adds M overlaps to N data, performs filter processing, and removes M overlaps again like the digital filter circuit 100. Accordingly, the processing throughput, which is higher by the overlap than the throughputs of the input and the output of data, is required for the filter processing. Specifically, the processing throughput, which is N/(N-M) times or more the throughputs of the input and the output of data, is required for the filter processing.

[0070] In this exemplary embodiment, the throughputs of the input and the output of data in the digital filter circuit 101 are also determined by the frequency of the clock D. The reason is that the data input in the overlap addition circuit 30 and the data output in the overlap removal circuit 34 operate synchronizing with the clock D.

[0071] On the other hand, the processing throughput of the filter processing in the digital filter circuit 101 is determined by the frequency of the filter clock 25. The reason is that the data output in the overlap addition circuit 30, the data input in the overlap removal circuit 34, the FFT circuit 31, the filter computation circuit 32 and the IFFT circuit 33 operate synchronizing with the filter clock 25.

[0072] Here, in this exemplary embodiment, the filter processing is also performed with the throughput of N/(N-M) times as many as the throughputs of the data input and the data output by the frequency of the filter clock 25 is N/(N-M) times the clock D. Specifically, the frequency fF of the filter clock 25 is set by the above-mentioned Equation 4.

[0073] At that time, the digital filter circuit 101 performs filter processing in every cycle of the filter clock 25 to the inputted data in every cycle of the clock D, and outputs the data, which filter processing was performed to, in every cycle of the clock D.

[0074] Generally, when throughputs of both input/output data and filter processing are not identical in a digital filter circuit, flow controls are needed, such as a control of suspension and starting of an input-output of data and a control of suspension and starting of filter processing. On the other hand, according to the digital filter circuit 101 of this exemplary embodiment, this flow control is unnecessary. Thereby, because the circuit can be simplified , circuit scale and power consumption can be reduced.

[0075] Specifically, for example, when the overlap addition circuit 30 and the overlap removal circuit 34 are realized by a dual port memory, it does not occur that the memory is full of data, or empty thereof. Accordingly, a flow control which suspends or starts writing into the memory and reading from the memory is unnecessary. Because the circuit can be simplified, as a result, the circuit scale and the power consumption can be reduced.

[0076] Meanwhile, it is obvious from Equation 4 of the frequency fF of the above-mentioned filter clock 25 that the frequency of the filter clock 25 can be low when the number M of the overlaps is smaller and smaller, and the FFT block size N is larger and larger.

[0077] However, because the required number M of the overlaps is determined based on the impulse response length of the filter which is mainly realized in the digital filter circuit 101, it is difficult to set freely the required number M of the overlaps. Accordingly, if the FFT block size N is large against the required number M of the overlaps, the frequency of the filter clock 25 can be low. While lower Frequency leads to smaller power consumption, larger size N of FFT block leads to larger power consumption of the FFT circuit 31 and the IFFT circuit 33 themselves.

[0078] Accordingly, FFT block size N which gives smallest power consumption should exist for the required number M of the overlaps. In other words, the FFT block size N which gives smallest power consumption can be determined for the required number M of the overlaps. Further, when the number M of the overlaps and the number N of the FFT block size are determined, the frequency fF of the filter clock 25 is determined from Equation 4.

[0079] Hereinafter, a difference between the previous example and the first exemplary embodiment will be described.

[0080] Fig. 5 is a graph showing an example of power consumption of the digital filter circuit 101 for the FFT block size N of 2048 4096 and 8192, a case where the number M of the overlaps is changed by 128 units from 0 to 1280.

[0081] It can be understood from Fig. 5 that the FFT block sizes N where the power consumption is minimum among at least FFT block size N=2048, 4096 and 8192 are different depending on the number M of the overlaps. Specifically, when the number M of the overlaps is 0, 128 and 256 (area (a) in Fig. 5), the power consumption is minimized when the FFT block size N is 2048. When the number M of the overlaps is 384 and 512 (area (b) in Fig. 5), the power consumption is minimized when the FFT block size N is 4096. When the number M of the overlaps is 640, 768, 896, 1024, 1152 and 1280 (area (c) in Fig. 5), the power consumption is minimized when the FFT block size N is 8192.

[0082] The setup table 36 holds a set value of the FFT block size based on the FFT block size setting signal 26 and a set value of the filter clock frequency based on the filter clock frequency setting signal 27 corresponding to a set value of the number of the overlaps based on the overlap number setting signal 26. The setup table 36 outputs the FFT block size setting signal 26 and the filter clock frequency setting signal 27 corresponding to the inputted number of the overlaps.

[0083] Fig. 6 is a table diagram showing an exemplary configuration of the setup table 36. In Fig. 6, the FFT block size shown in Fig. 5 that minimizes power consumption is set to each overlap number settings of 128 steps from 0 to 1280 in the setup table 36. Further, a frequency of the filter clock 25 is set by Equation 4 from each overlap number settings and the FFT block size.

[0084] In other words, when the number of the overlaps required for the overlap number setting signal 28 is set, the setup table 36 outputs the FFT block size setting signal 26 which minimizes power consumption and the filter clock frequency setting signal 27 which realizes the required throughput of the filter processing.

[0085] As a result, the digital filter circuit 101 can operate under the setting of the FFT block size and the frequency of the filter clock 25 that minimize power consumption and to the inputted overlap number setting signal 28.

[0086] Further, the contents of the setup table 36 can be set with reference to the power consumption characteristics shown in Fig. 5. In this case, the power consumption characteristics may be obtained by a simulation or by measuring the power consumption of the actual circuit.

[0087] Further, the contents of the setup table 36 may be set when the digital filter circuit 101 is designed or is produced thereof and when a higher rank circuit such as CPU which controls the digital filter circuit 101 operates the digital filter circuit 101.

[0088] Further, the setup table 36 may not be a circuit for exclusive use. For example, the setup table may be installed on the main memory used by a CPU which controls the digital filter circuit 101.

[0089] According to the first exemplary embodiment described above, in the frequency domain filter processing using overlap FFT, the operation frequency of the filter processing is set to N/(N-M) times the operating frequency of the input-output of data in accordance with the required number M of the overlaps and the FFT block size N. Accordingly, the power consumption can be reduced because of being operated by the operating frequency necessary and sufficient to perform filter processing without delaying in accordance with the required number of overlaps.

[0090] Further, because a flow control regarding an input-output of data and filter processing is not needed, the required memory size is small and the circuit configuration can be simplified. So the circuit scale and power consumption can be reduced.

[0091] Further, according to the first exemplary embodiment, because the FFT block size is adjusted in accordance with the required number of the overlaps, the power consumption can be reduced.

[0092] Fig. 7 is a diagram showing another exemplary configuration of a digital filter circuit 1000. According to Fig. 7, the digital filter circuit 1000 includes an overlap addition means 1001, an FFT means 1002, a filter computation means 1003, an IFFT means 1004, an overlap removal means 1005 and a clock generation means 1006.

[0093] The overlap addition means 1001 generates a block including N data from an input data in the time domain by overlapping only M data with the previous block. Further, N and M are positive integers.

[0094] The FFT means 1002 transforms a block, which is generated by the overlap addition means 1001, by FFT processing to a block in the frequency domain.

[0095] The filter computation means 1003 performs filter processing to the block, in the frequency domain, transformed by the FFT means 1002.

[0096] The IFFT means 1004 transforms the block, to which the filter computation means 1003 performed the filter processing, to a block in the time domain by IFFT processing.

[0097] The overlap removal means 1005 generates an output signal by removing a total of M data from both ends of the block, in the time domain, transformed by the IFFT means 1004.

[0098] The clock generation means 1006 sets a frequency of a filter processing clock signal, which drives the data output unit of the overlap addition means 1001, the FFT means 1002, the filter computation means 1003, the IFFT means 1004, and a data input unit of the overlap removal means 1005, based on the value of the number M of the overlaps.

[0099] As described above, the frequency of the filter processing clock signal is set based on the value of the number M of the overlaps. Accordingly, in the digital filter circuit 1000 a flow control regarding the input-output of data and the filter processing is not needed to be performed. Therefore, in the digital filter circuit 1000 the required memory size is small and the circuit configuration can be simplified.

[0100] Accordingly, in the filter processing in a frequency domain such as an overlap FDE method, the circuit scale and the power consumption can be reduced.

[0101] Another exemplary configuration of a digital filter circuit comprises: an overlap addition means which generates a block including data of N data (N is a positive integer) from an input data in a time domain by giving an overlap of M data (M is a positive integer) between the block and the previous block; a FFT means which transforms the generated block to a block in a frequency domain by FFT processing; a filter computation means which performs filter processing to the transformed block in the frequency domain; an IFFT means which transforms the block, which the filter processing was performed to, to a block in the time domain by IFFT processing; and an overlap removal means which generates an output signal by removing a total of M data from both ends of the transformed block in the time domain, wherein a frequency of the filter processing clock signal, which drives a data output unit of the overlap addition means, the FFT means, the filter computation means, the IFFT means, and a data input unit of the overlap removal means, is set N/ (N-M) times a frequency of an input/output clock signal which drives a data input unit of the overlap addition means and a data output unit of the overlap removal means.

[0102] A digital filter control method according to an example is a control method for a digital filter which includes: an overlap addition means which generates a block including data of N data (N is a positive integer) from an input data in a time domain by giving an overlap of M data (M is a positive integer) between the block and the previous block; a FFT means which transforms the generated block to a block in a frequency domain by FFT processing; a filter computation means which performs filter processing to the transformed block in the frequency domain; an IFFT means which transforms the block, which the filter processing was performed to, to a block in the time domain by IFFT processing; and an overlap removal means which generates an output signal by removing a total of M data from both ends of the transformed block in the time domain, wherein a frequency of the filter processing clock signal, which drives a data output unit of the overlap addition means, the FFT means, the filter computation means, the IFFT means, and a data input unit of the overlap removal means, is set to N/ (N-M) times a frequency of an input/output clock signal which drives a data input unit of the overlap addition means and a data output unit of the overlap removal means.

[0103] A digital filter control method according to another example is a control method for a digital filter and includes: an overlap addition means which generates a block including data of N data (N is a positive integer) from an input data in a time domain by giving an overlap of M data (M is a positive integer) between the block and the previous block; a FFT means which transforms the generated block to a block in a frequency domain by FFT processing; a filter computation means which performs filter processing to the transformed block in the frequency domain; an IFFT means which transforms the block, which the filter processing was performed to, to a block in the time domain by IFFT processing; an overlap removal means which generates an output signal by removing a total of M data from both ends of the transformed block in the time domain; and a setup table means which holds FFT size setting indicating the number N of the data which composes the block and clock frequency setting, wherein filter processing is performed based on the FFT size setting and the clock frequency setting which the setup table means outputs according to the number M of the overlaps.

[0104] While the invention has been particularly shown and described with reference to exemplary embodiments thereof, the invention is not limited to these embodiments. It will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the scope of the present invention as defined by the claims.

[0105] This application is based upon and claims the benefit of priority from Japanese patent application No. 2010-284137, filed on December 21, 2010.

DESCRIPTION OF THE REFERENCE NUMERALS



[0106] 
10
overlap addition circuit
11
FFT circuit
12
filter computation circuit
13
IFFT circuit
14
overlap removal circuit
15
clock generation circuit
25
filter clock
26
FFT block size setting signal
27
filter clock frequency setting signal
28
overlap number setting signal
30
overlap addition circuit
31
FFT circuit
32
filter computation circuit
33
IFFT circuit
34
overlap removal circuit
35
clock generation circuit
36
setup table
100, 101, 1000
digital filter circuit
1001
overlap addition means
1002
FFT means
1003
filter computation means
1004
IFFT means
1005
overlap removal means
1006
clock generation means



Claims

1. A digital filter circuit (100) comprising:

an overlap addition means (10) for generating a block including N data, N being a positive integer, from input data (28) in a time domain by giving an overlap of M data, M being a positive integer, between the block and the previous block;

an FFT processing means (11) for transforming the generated block to a block in a frequency domain by performing FFT processing;

a filter computation means (12) for performing filter processing to the transformed block in the frequency domain and outputting a filtered block;

an IFFT means (13) for transforming the filtered block to a block in the time domain by IFFT processing;

an overlap removal means (14) for generating an output signal by removing a total of M data from both ends of the transformed block in the time domain;

characterized in that:

the digital filter circuit (100) further comprises a clock generation means (15) for setting a frequency of a filter processing clock signal (25) based on the number M of the overlaps, wherein the filter processing clock signal drives a data output unit of the overlap addition means, the FFT means, the filter computation means, the IFFT means, and a data input unit of the overlap removal means,

wherein the number N is determined based on the number M, so that the power consumption is minimized, and the frequency of the filter processing clock signal is determined based on the number N and the number M, wherein the clock generation means (15) sets the frequency of the filter processing clock signal (25) at N/(N-M) times a frequency of an input-output clock signal which drives a data input unit of the overlap addition means (10) and a data output unit of the overlap removal means (14).


 
2. The digital filter circuit (101) according to claim 1, further comprising:
a setup table means (36) for holding a FFT size setting indicating the number N of the data which constitute the block and a clock frequency setting corresponding to the number M of the overlaps; wherein the clock generation means (35) sets the frequency of the filter processing clock signal (25) based on the FFT size setting and the clock frequency setting which the setup table means outputs in accordance with an inputted set value of the number M of the overlaps.
 
3. A digital filter control method comprising the steps of:

generating a block including N data, N being a positive integer, from input data (28) in a time domain by performing overlap addition processing which causes the block to give an overlap of M data, M being a positive integer, between the block and the previous block;

transforming the generated block to a block in a frequency domain by performing FFT processing;

performing filter processing to the transformed block in the frequency domain and generating a filtered block;

transforming the filtered block to a block in the time domain by IFFT processing;

performing overlap removal processing which generates an output signal by removing a total of M data from both ends of the transformed block in the time domain; characterized in that:

the digital filter control method further comprises setting the frequency of a filter processing clock signal (25) based on the number M of the overlaps, wherein the filter processing clock signal relates to an output of the overlap addition processing, the FFT processing, the filter processing, the IFFT processing, and an input of the overlap removal processing,

wherein the number N is determined based on the number M, so that the power consumption is minimized, and the frequency of the filter processing clock signal is determined based on the number N and the number M, wherein the frequency of the filter processing clock signal (25) is set to N/(N-M) times a frequency of an input-output clock signal which relates to the input of the overlap addition processing and the output of the overlap removal processing.


 
4. The digital filter control method according to claim 3, further comprising the steps of:

holding FFT size setting indicating the number N of the data which constitute the block and a clock frequency setting corresponding to the number M of the overlaps; and

setting a frequency of the filter processing clock signal (25) based on the FFT size setting and the clock frequency setting in accordance with an inputted set value of the number M of the overlaps.


 


Ansprüche

1. Digitale Filterschaltung (100), umfassend:

eine Überlappungsadditionseinrichtung (10) zum Erzeugen eines Blocks mit N Daten, wobei N eine positive ganze Zahl ist, aus Eingangsdaten (28) in einem Zeitbereich, indem eine Überlappung von M Daten, wobei M eine positive ganze Zahl ist, zwischen dem Block und dem vorhergehenden Block gegeben wird;

eine FFT-Verarbeitungseinrichtung (11) zum Transformieren des erzeugten Blocks in einen Block in einem Frequenzbereich durch Ausführen einer FFT-Verarbeitung;

eine Filterberechnungseinrichtung (12) zum Durchführen einer Filterverarbeitung an dem transformierten Block in der Frequenzdomäne und zum Ausgeben eines gefilterten Blocks;

ein IFFT-Mittel (13) zum Transformieren des gefilterten Blocks in einen Block in der Zeitdomäne durch IFFT-Verarbeitung;

eine Überlappungsentfernungseinrichtung (14) zum Erzeugen eines Ausgangssignals durch Entfernen von insgesamt M Daten von beiden Enden des transformierten Blocks im der Zeitdomäne;

dadurch gekennzeichnet, dass

die digitale Filterschaltung (100) ferner eine Takterzeugungseinrichtung (15) zum Einstellen einer Frequenz eines Filterverarbeitungstaktsignals (25) auf der Grundlage der Anzahl M der Überlappungen umfasst, wobei das Filterverarbeitungstaktsignal eine Datenausgabeeinheit der Überlappungsadditionseinrichtung, die FFT-Einrichtung, die Filterberechnungseinrichtung, die IFFT-Einrichtung und eine Dateneingabeeinheit der Überlappungsentfernungseinrichtung ansteuert,

wobei die Zahl N auf der Grundlage der Zahl M bestimmt wird, so dass der Stromverbrauch minimiert wird, und die Frequenz des

Filterverarbeitungs-Taktsignals auf der Grundlage der Zahl N und der Zahl M bestimmt wird,

wobei die Takterzeugungseinrichtung (15) die Frequenz des Filterverarbeitungstaktsignals (25) auf das N/(N-M)-fache einer Frequenz eines Eingangs-Ausgangs-Taktsignals einstellt, das eine Dateneingabeeinheit der Überlappungsadditionseinrichtung (10) und eine Datenausgabeeinheit der Überlappungsentfernungseinrichtung (14) ansteuert.


 
2. Digitale Filterschaltung (101) nach Anspruch 1, weiter umfassend:
eine Setup-Tabelleneinrichtung (36) zum Halten einer FFT-Größeneinstellung, die die Anzahl N der Daten angibt, die den Block bilden, und einer Taktfrequenzeinstellung, die der Anzahl M der Überlappungen entspricht;
wobei die Takterzeugungseinrichtung (35) die Frequenz des Filterverarbeitungstaktsignals (25) auf der Grundlage der FFT-Größeneinstellung und der Taktfrequenzeinstellung, die die Setup-Tabelleneinrichtung entsprechend einem eingegebenen Einstellwert der Anzahl M der Überlappungen ausgibt, einstellt.
 
3. Digitales Filtersteuerverfahren, das die folgenden Schritte umfaßt:

Erzeugen eines Blocks mit N Daten, wobei N eine positive ganze Zahl ist, aus Eingangsdaten (28) in einem Zeitbereich durch Ausführen einer Überlappungsadditionsverarbeitung, die bewirkt, daß der Block eine Überlappung von M Daten, wobei M eine positive ganze Zahl ist, zwischen dem Block und dem vorhergehenden Block ergibt;

Transformieren des erzeugten Blocks in einen Block in einem Frequenzbereich durch Durchführen einer FFT-Verarbeitung;

Durchführen einer Filterverarbeitung für den transformierten Block in der Frequenzdomäne und Erzeugen eines gefilterten Blocks;

Transformieren des gefilterten Blocks in einen Block in der Zeitdomäne durch IFFT-Verarbeitung;

Durchführen einer Überlappungsentfernungsverarbeitung, die ein Ausgangssignal durch Entfernen von insgesamt M Daten von beiden Enden des transformierten Blocks in der Zeitdomäne erzeugt; dadurch gekennzeichnet, dass

das Digitalfiltersteuerverfahren ferner das Einstellen der Frequenz eines Filterverarbeitungstaktsignals (25) auf der Grundlage der Anzahl M der Überlappungen umfasst, wobei sich das Filterverarbeitungstaktsignal auf einen Ausgang der Überlappungsadditionsverarbeitung, der FFT-Verarbeitung, der Filterverarbeitung, der IFFT-Verarbeitung und einen Eingang der Überlappungsentfernungsverarbeitung bezieht,

wobei die Anzahl N auf der Basis der Anzahl M bestimmt wird, so dass die Leistungsaufnahme minimiert wird, und die Frequenz des Filterverarbeitungstaktsignals auf der Basis der Anzahl N und der Anzahl M bestimmt wird, wobei die Frequenz des Filterverarbeitungstaktsignals (25) auf das N/(N-M)-fache einer Frequenz eines Eingangs-Ausgangs-Taktsignals gesetzt wird, das sich auf den Eingang der Überlappungsadditionsverarbeitung und den Ausgang der Überlappungsentfernungsverarbeitung bezieht.


 
4. Digitales Filtersteuerungsverfahren nach Anspruch 3, das weiterhin die folgenden Schritte umfasst
Halten der FFT-Größeneinstellung, die die Anzahl N der Daten anzeigt, die den Block bilden, und einer Taktfrequenzeinstellung, die der Anzahl M der Überlappungen entspricht; und
Einstellen einer Frequenz des Filterverarbeitungs-Taktsignals (25) auf der Grundlage der FFT-Größeneinstellung und der Taktfrequenzeinstellung in Übereinstimmung mit einem eingegebenen Einstellwert der Anzahl M der Überlappungen.
 


Revendications

1. Circuit de filtre numérique (100) comprenant :

un moyen d'ajout de chevauchement (10) pour générer un bloc incluant N données, N étant un nombre entier positif, parmi des données d'entrée (28) dans un domaine temporel en donnant un chevauchement de M données, M étant un nombre entier positif, entre le bloc et le bloc précédent ;

un moyen de traitement FFT (11) pour transformer le bloc généré en un bloc dans un domaine fréquentiel en réalisant un traitement FFT ;

un moyen de calcul de filtre (12) pour réaliser un traitement de filtre sur le bloc transformé dans le domaine fréquentiel et pour délivrer un bloc filtré ;

un moyen IFFT (13) pour transformer le bloc filtré en un bloc dans le domaine de temps par un traitement IFFT ;

un moyen de suppression de chevauchement (14) pour générer un signal de sortie par la suppression d'un total de M données des deux extrémités du bloc transformé dans le domaine temporel ;

caractérisé en ce que :

le circuit de filtre numérique (100) comprend en outre un moyen de génération d'horloge (15) pour régler une fréquence d'un signal d'horloge de traitement de filtre (25) sur la base du nombre M des chevauchements, dans lequel le signal d'horloge de traitement de filtre pilote une unité de sortie de données du moyen d'ajout de chevauchement, le moyen FFT, le moyen de calcul de filtre, le moyen IFFT, et une unité d'entrée de données du moyen de suppression de chevauchement,

dans lequel le nombre N est déterminé sur la base du nombre M, de sorte que la consommation d'énergie soit minimisée, et la fréquence du signal d'horloge de traitement de filtre est déterminée sur la base du nombre N et du nombre M,

dans lequel le moyen de génération d'horloge (15) règle la fréquence du signal d'horloge de traitement de filtre (25) à N/(N-M) fois une fréquence d'un signal d'horloge d'entrée-sortie qui pilote une unité d'entrée de données du moyen d'ajout de chevauchement (10) et une unité de sortie de données du moyen de suppression de chevauchement (14).


 
2. Circuit de filtre numérique (101) selon la revendication 1, comprenant en outre :

un moyen de table de configuration (36) pour maintenir un réglage de taille FFT indiquant le nombre N des données constituant le bloc et un réglage de fréquence d'horloge correspondant au nombre M des chevauchements ;

dans lequel le moyen de génération d'horloge (35) règle la fréquence du signal d'horloge de traitement de filtre (25) sur la base du réglage de taille FFT et du réglage de fréquence d'horloge que le moyen de table de configuration délivre en fonction d'une valeur réglée entrée du nombre M des chevauchements.


 
3. Procédé de commande de filtre numérique comprenant les étapes de :

la génération d'un bloc incluant N données, N étant un nombre entier positif, parmi des données d'entrée (28) dans un domaine temporel, en réalisant un traitement d'ajout de chevauchement qui amène le bloc à donner un chevauchement de M données, M étant un nombre entier positif, entre le bloc et le bloc précédent ;

la transformation du bloc généré en un bloc dans un domaine de fréquentiel en réalisant un traitement FFT ;

la réalisation d'un traitement de filtre sur le bloc transformé dans le domaine de fréquentiel et la génération d'un bloc filtré ;

la transformation du bloc filtré en un bloc dans le domaine temporel par un traitement IFFT ;

la réalisation d'un traitement de suppression de chevauchement qui génère un signal de sortie par la suppression d'un total de M données des deux extrémités du bloc transformé dans le domaine de temps ;

caractérisé en ce que :

le procédé de commande de filtre numérique comprend en outre le réglage de la fréquence d'un signal d'horloge de traitement de filtre (25) sur la base du nombre M des chevauchements, dans lequel le signal d'horloge de traitement de filtre est relatif à une sortie du traitement d'ajout de chevauchement, du traitement FFT, du traitement de filtre, du traitement IFFT, et une entrée du traitement de suppression de chevauchement,

dans lequel le nombre N est déterminé sur la base du nombre M, de sorte que la consommation d'énergie soit minimisée, et la fréquence du signal d'horloge de traitement de filtre est déterminée sur la base du nombre N et du nombre M, dans lequel la fréquence du signal d'horloge de traitement de filtre (25) est réglée à N/(N-M) fois une fréquence d'un signal d'horloge d'entrée-sortie qui est relatif à l'entrée du traitement d'ajout de chevauchement et à la sortie du traitement de suppression de chevauchement.


 
4. Procédé de commande de filtre numérique selon la revendication 3, comprenant en outre les étapes de :

le maintien d'un réglage de taille FFT indiquant le nombre N des données constituant le bloc et d'un réglage de fréquence d'horloge correspondant au nombre M des chevauchements ; et

le réglage d'une fréquence du signal d'horloge de traitement de filtre (25) sur la base du réglage de taille FFT et du réglage de fréquence d'horloge en fonction d'une valeur réglée entrée du nombre M des chevauchements.


 




Drawing


























Cited references

REFERENCES CITED IN THE DESCRIPTION



This list of references cited by the applicant is for the reader's convenience only. It does not form part of the European patent document. Even though great care has been taken in compiling the references, errors or omissions cannot be excluded and the EPO disclaims all liability in this regard.

Patent documents cited in the description




Non-patent literature cited in the description