WO1998018083A1 - A device and method for calculating fft - Google Patents
A device and method for calculating fft Download PDFInfo
- Publication number
- WO1998018083A1 WO1998018083A1 PCT/SE1997/001637 SE9701637W WO9818083A1 WO 1998018083 A1 WO1998018083 A1 WO 1998018083A1 SE 9701637 W SE9701637 W SE 9701637W WO 9818083 A1 WO9818083 A1 WO 9818083A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- calculations
- calculating unit
- memory
- values
- input values
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
- G06F17/141—Discrete Fourier transforms
- G06F17/142—Fast Fourier transforms, e.g. using a Cooley-Tukey type algorithm
Definitions
- the present invention relates to a method of calculating Fast Fourier Transform (FFT) and to a device for carrying out the method, and in particular to the interconnection of and the coaction between the calculating unit and the memory of said device.
- FFT Fast Fourier Transform
- the Discrete Fourier Transform (DFT) of a sequence f(n) is defined as:
- Cooley and Tukey introduced a novel method of calculating X(k) and x(n) respectively, resulting in a much shorter calculation time.
- the method was designated Fast Fourier Transform (FFT) and utilizes the circumstance that the same calculations reoccur at several positions.
- FFT Fast Fourier Transform
- The* number of additions and multiplications can be reduced to the order of log 2 N when N is a two-power.
- the method is based on a number of successive calculations with a series or column of values, where each pair of calculations is called a butterfly.
- the old series is no longer required each time a new series is calculated.
- This enables the memory, or store, in which the previous series is stored to be reused, this being referred to as calculating inplace.
- the method can be carried out with a memory of size N, disclosed in GB 1 546 173 and GB 2 006 485, for instance.
- the drawback with this is that memory accesses are always made to the same memory, which is not particularly effective.
- FFT can also be implemented with the use of two memories of size N, as disclosed in the technical report "An Energy-Efficient FFT Processor Architecture"
- the present invention addresses the problem of FFT calculations requiring either a great deal of time or a large amount of memory space.
- the object of the present invention is to provide a solution to this problem by organizing the interconnection and coaction between the calculating unit and memory, so as to enable two butterfly calculations to be commenced simultaneously. This is achieved by reading the input values from memory positions in the memories and intermediately storing these input values in a register in the calculating unit after reading, and/or by storing the output values to be written into memory positions in the memory intermediately in a register in the calculating unit prior to writing-in said values.
- the values are conveniently allocated or distributed so that each memory will contain essentially the same number of values. This enables, for instance, two memories to be used to a maximum, by reading from and writing into memory positions in the two memories simultaneously. Alternatively, values can be read from one memory at the same time as values are written into the other memory.
- One advantage afforded by the present invention is that the memories are utilized to a maximum, therewith enabling the memory space to be minimized to half the memory space required by the known solution that utilizes two memories. Another advantage is that the time required in this respect can be reduced in comparison with the time required by the known solution that utilizes a single memory.
- a butterfly designates a group of least two calculations, where each calculation includes at least one calculation step.
- Figure 1a is a graphic illustration of a known eight-point radix-2 decimation-in-time FFT algorithm.
- Figure 1b is a graphic illustration of the butterfly calculation in Figure 1a.
- Figure 2a is a graphic illustration of a known eight-point radix-2 decimation-in- frequency FFT algorithm.
- Figure 2b is a graphic illustration of the butterfly calculation in Figure 2a.
- Figure 3a is a graphic illustration of a known 64-point radix-4 decimation-in- frequency FFT algorithm.
- Figure 3b is a graphic illustration of the butterfly calculation in Figure 3a.
- Figure 4a and 4b is a graphic illustration of known technology.
- Figure 5 is a block schematic illustrating one embodiment of the invention.
- Figure 6 is a block schematic illustrating another embodiment of the invention.
- Figure 7 is a block schematic illustrating one embodiment of the calculating unit in
- Figure 5 or in Figure 6 is a block schematic illustrating another embodiment of the calculating unit in Figure 5 or in Figure 6.
- Figure 8b is a block schematic illustrating one embodiment of the summation unit in Figure 8a.
- Figure 9 is a block schematic illustrating a further embodiment of the calculating unit in Figure 5 or in Figure 6.
- FFT Fast Fourier Transform
- DFT Discrete Fourier Transform
- the invention is primarily intended for use in a communications system that employs multi-carrier wave modulation and where inverse FFT (I FFT) is used for modulation whereas FFT is used for demodulation, although the invention may, of course, be used in other areas.
- I FFT inverse FFT
- Multi-carrier wave modulation is a known method of transmitting a large number of Mbits over a telephone line, for instance.
- the method is based on sampling speech in a telephone conversation, resulting in a large number of bits.
- the bits to be transmitted are delivered to a transmitter in some complex number form, whereafter an IFFT is carried out.
- the IFFT gives a sum of orthogonal carrier waves or tones whose amplitudes and phase shifts are influenced by the complex numbers.
- FFT is carried out instead, wherewith the original bits are returned. Attenuation in the telephone line can be readily compensated for, by multiplying by a complex number on each carrier wave.
- Multi-carrier wave modulation is described in more detail in WO 95/03656, EP 0 653 858, EP 0 656 706, EP 0 683 576 and WO 95/05042, for instance.
- FIG. 1a is a graphic presentation of a known FFT algorithm having eight input values x(0)-x(7) and eight output values X(0)-X(7).
- Figure 1b illustrates a recurring group of calculations referred to as a butterfly for obvious reasons. Because a butterfly has two input values and two output values in this case, this FFT algorithm is called a radix-2 type algorithm. Each butterfly includes a complex multiplication and two complex additions according to:
- the input values x(0)-x(7) in column C1 are used to calculate a new series of values in column C2, which is used, in turn, to calculate column C3, which, in turn, is used to calculate column C4.
- Column C4 will then include the desired output values X(0)-X(7).
- the input values x(0)-x(7) in Figure 1a are in the correct order, wherewith the output values X(0)-X(7) becomes mixed. The opposite is also conceivable.
- Figure 2a illustrates a graphic presentation of another known radix-2-FFT algorithm. This algorithm is called a decimation-in-frequency (DIF) algorithm.
- DIF decimation-in-frequency
- Figure 2b shows a corresponding butterfly according to:
- the butterfly of the algorithm is shown schematically in Figure 3a, and in more detail in Figure 3b.
- x(l)' (x(j)-x(k)+x(l)-x(m))W(l,N)
- FFT algorithms having higher radix numbers are also known, although radix-2 and radix-4 are the most common.
- a common feature of all FFT algorithms is that the FFT is divided into a number of columns in which subtotals are calculated. These subtotals are then used as input values for the next column. The number of columns is decided by the size of the FFT to be calculated and also by the radix used.
- An FFT processor can thus be built-up around a calculating or computing unit that performs a butterfly calculation, and at least one data store.
- the bottleneck lies in the memory accesses. At least two memory accesses per clock cycle would be desirable as an optimum.
- Figures 4a and 4b illustrate an alternative method of organizing the memory and the calculating unit.
- Figure 4a and 4b illustrate solely the principle of transmitting data to and fro and shall not be interpreted literally.
- One calculating unit 100 and first and second memories 101 and 102 are used.
- the input values are first read from memory positions in a first memory 101 of said two memories, at the same time as the output values are written into memory positions in the other or second memory 102; see Figure 4a.
- the results of the calculation of a first column lie in the second memory 102.
- Data in the second memory 102 is then used as the input values to the calculating unit 100, while the output values for the next column are written into memory positions in the first memory 101.
- the memories 101 and 102 are thus switched until the whole of the FFT has been calculated.
- Each, of the memories 101 and 102 must have at least the same size as the FFT to be calculated.
- At least one of the columns will incorporate the problem that all input values lie in said one memory and all output values shall be written into the same memory, whereas the other memory is not used at all.
- the same problem arises when, for instance, placing odd memory positions in one memory and even positions in the other memory, although in another column.
- the invention is based on calculating butterflies from different parts of the column in the calculating unit at one and the same time. This enables values to be read from and written into the two memories alternately, in a manner to ensure that no memory will be overloaded.
- the memory requirement can be halved in comparison with the aforedescribed method, at the cost of a few more registers in the calculating unit for the intermediate storage of subresults.
- Table 1 illustrates an example of the configuration of a calculating process.
- the memories are referenced 111 and 112, which is in accord with Figures 5 and 6 described here below.
- Each row in the Table corresponds to a clock cycle. It will be noted that a calculation often includes more than one calculating step and that the Table thus gives only a simplified picture.
- the two memories are divided so that the first memory 112 will contain values that have odd numbers, while the second memory 111 will include values that have even numbers.
- Other divisions are possible.
- the values x(0), x(1), x(2) and x(3) can be inputted in one memory and the values x(4), x(5), x(6) and x(7) inputted in the other memory.
- Calculation of the first column is simple, by virtue of the two input values required for each butterfly lying in a respective memory.
- the calculating unit reads value x(0) from one memory position in the first memory 112 and value x(1) from a memory position in the second memory 111.
- the butterfly is calculated in the two subsequent clock cycles and results in the output values x'(0) and x'(1) that are to be written back into the memory positions 0 and 1 as the values x(0) and x(1) were read from. This cannot be done immediately, however, and it is necessary to store the output values x'(0) and x'(1) intermediately in registers until the values can be written-in two clock cycles later on.
- the first butterfly requires the values x'(0) and x'(2)
- the second butterfly requires the values x'(1) and x'(3). Consequently, the values x'(0) and x'(1) are read on the first instance and stored intermediately.
- the values x'(2) and x'(3) are then read.
- the first butterfly calculation with x'(0) and x'(2) can now be carried out, which takes two clock cycles, whereafter the second butterfly calculation can be carried out with x'(1) and x f (3). New values are read during this time period, in accordance with the aforegoing.
- the third column is calculated in the same way.
- the odd values are stored in a first memory 112 and the even values in a second memory 111.
- the values intended for calculation of a column will consequently appear in one and the same memory, with the exception of the values intended for the last column, for which these latter input values are found uniformly distributed in both memories.
- the four values are read-in sequentially in Table 5, nevertheless.
- Figure 5 illustrates a method of implementing the invention.
- a calculating unit 110 is connected to a first memory 111 and to a second memory 112 by means of two write busses 114a and 114b, and two read busses 113a and 113b.
- Half of the values are stored in memory positions in the first memory 111 and half in memory positions in the second memory 112. This enables reading and writing to be effected simultaneously in both memories 111 and 112. It will be noted that only half of the memory space in the two memories 111 and 112 is required, in comparison with Figures 4a and 4b.
- Figure 6 shows another embodiment of the invention.
- a calculating unit 110 is connected to a first memory 111 and to a second memory 112 by means of a common write bus 114 and a common read bus 113.
- the calculating unit 110 writes to the second memory 112 at the same time as it reads from the first memory 111.
- One common write bus 114 and one common read bus 113 will therefore suffice to this end.
- the Figure is similar to Figure 5 in other respects.
- Figure 7 is a block schematic that illustrates a conceivable design of the calculating unit in Figure 5 and in Figure 6 in the case of radix-2 DIF. Both of the two input values pass to a subtractor 122 and to an adder 123. The value delivered by the substractor 122 is multiplied in a complex multiplier 125 by a weight coefficient delivered from a weight coefficient memory 124.
- registers are also required for the intermediate storage of the input values and the output values respectively, so that two butterfly calculations can be commenced simultaneously.
- the number of registers required will depend on the number of clock cycles needed between reading and calculating and between calculating and writing.
- the registers shown in Figure 7 are illustrated purely schematically as an intermediate storage device 121 into which input values are entered, and an intermediate storage device 126 from which the output values are delivered. It is not necessary to always store both incoming and outgoing values.
- Figure 8a is a block schematic illustrating a conceivable design of the calculating unit 110 in Figure 5 or in Figure 6 in the case radix-4 DIF.
- the four input values required are delivered to four summation units 131 , whereafter the four summated values are delivered singly from the summation units 131 , via a multiplexor 132, to a complex multiplier 125, where the summated values are multiplied by corresponding weight coefficients from a weight coefficient memory 124 prior to being resent to the memories.
- FIG. 8b is a detailed illustration of a design of a summation unit 131.
- the summation unit 131 includes a multiplier 133, an adder 134, a summation register 135 and a multiplexor 136.
- the first input value is multiplied in the multiplier 133 by 1 , -1 , i or -i depending on which of the values in the butterfly shall be calculated. No conventional advanced multiplier is required for such an operation, since all that is required is a change of sign or a change of place.
- the result obtained in the multiplier 133 is delivered to the adder 134.
- the multiplexor 136 chooses what the adder 134 shall add to the summation register 135. On the first occasion, the value 0 is added to the result from the multiplier 133 and placed in the summation register 135. On the remaining three occasions, the value is instead placed in the summation register 135 together with the result from the multiplier 133.
- Figure 9 shows a more concrete example of the register layout. This Figure illustrates implementation of Table 6 with the aid of a calculating unit 110. It will be noted that each row in the Table corresponds to a clock cycle, but that the presentation of the calculation in the Table is simplified. In Figure 9, a value can be moved from one register to another during a clock cycle, with possible intermediate calculations, or may remain in the same register.
- the values 0, 16, 32 and 64 are read-in one at a time and placed in four first registers 141 , 142, 143, 144.
- the values 0, 16, 32 and 64 are then moved to four second registers 145, 146, 147, 148, while new values 1 , 17, 33 and 49 are read-in one at a time and placed in said four registers 141 , 142, 143, 144.
- the calculating unit 110 utilizes the fact that the same additions reoccur.
- the intermediate result is therefore stored intermediately in four intermediate storage registers 154, 155, 156, 157.
- the values from respective registers 146 and 150 are multiplied by one of the values 1 or -i or by 1 or i in two simple multipliers 149 and 150. That is to say, the multiplication by 1 is the same as no multiplication at all, and the multiplication by i or -i is merely a question of a place change and sign change, if any, in the data word of the value.
- the addition and subtraction is performed with the aid of two multiplexors 151 and 152 and an adder-subtractor 153, wherewith the values in the intermediate storage registers 154, 155, 156 and 157 will be 0+32; 0-32; 16+64; -16i+64i respectively.
- the values 1 , 17, 33, 49 in the first registers 141 , 142, 143 and 144 are transferred to the second registers 145, 146, 147 and 148, while new values 2, 18, 34, 50 are read into the first register 141 , 142, 143, 144. All these values will later be processed analogously with the first values 0, 16, 32 and 64.
- the values in the intermediate storage registers 154, 155, 156 and 157 are now added to and subtracted from each other through the medium of two further multiplexors 158 and 159 and a further adder-substractor 161 , and are stored intermediately in a register 161.
- the value in the register 161 will then assume the values 0+16+32+64; 0-16i-32+64i; 0-16+32-64 and 0+16i-32-64i respectively.
- the value currently in register 161 will then be multiplied in a complex multiplier 125 by a weight coefficient taken from a weight coefficient memory 124. This weight coefficient is different for the different calculations.
- the resulting output values can then be written one at a time via terminating intermediate storage registers 162, 163, 164 and 165, and a terminating multiplexor 166.
- Calculating units for radix-2 and radix-4 DIT may, of course, be obtained in a similar way, although with multiplication by weight coefficients prior to addition and subtraction.
- FFT algorithms having higher radix numbers can also be implemented in a similar way.
Landscapes
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Discrete Mathematics (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Complex Calculations (AREA)
Abstract
Description
Claims
Priority Applications (5)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP97909781A EP1008060B1 (en) | 1996-10-21 | 1997-09-29 | A device and method for calculating fft |
US09/284,752 US6356926B1 (en) | 1996-10-21 | 1997-09-29 | Device and method for calculating FFT |
DE69737699T DE69737699T2 (en) | 1996-10-21 | 1997-09-29 | DEVICE AND METHOD FOR FFT CALCULATION |
CA002269464A CA2269464A1 (en) | 1996-10-21 | 1997-09-29 | A device and method for calculating fft |
AU47304/97A AU4730497A (en) | 1996-10-21 | 1997-09-29 | A device and method for calculating fft |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
SE9603862-5 | 1996-10-21 | ||
SE9603862A SE507529C2 (en) | 1996-10-21 | 1996-10-21 | Device and method for calculating FFT |
Publications (1)
Publication Number | Publication Date |
---|---|
WO1998018083A1 true WO1998018083A1 (en) | 1998-04-30 |
Family
ID=20404335
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/SE1997/001637 WO1998018083A1 (en) | 1996-10-21 | 1997-09-29 | A device and method for calculating fft |
Country Status (7)
Country | Link |
---|---|
US (1) | US6356926B1 (en) |
EP (1) | EP1008060B1 (en) |
AU (1) | AU4730497A (en) |
CA (1) | CA2269464A1 (en) |
DE (1) | DE69737699T2 (en) |
SE (1) | SE507529C2 (en) |
WO (1) | WO1998018083A1 (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB2391966A (en) * | 2002-08-15 | 2004-02-18 | Zarlink Semiconductor Ltd | A method and system for performing a fast-fourier transform |
CN1688104B (en) * | 2005-06-20 | 2010-11-10 | 北京中星微电子有限公司 | Digital signal processing method and apparatus |
Families Citing this family (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
IL131350A0 (en) * | 1999-08-11 | 2001-01-28 | Israel State | Data storage patterns for fast fourier transforms |
US6609140B1 (en) * | 1999-11-30 | 2003-08-19 | Mercury Computer Systems, Inc. | Methods and apparatus for fast fourier transforms |
EP1162547A3 (en) * | 2000-06-05 | 2003-09-03 | DSP Group Ltd. | In-Place Memory Management for FFT |
US6760741B1 (en) * | 2000-06-05 | 2004-07-06 | Corage Ltd. | FFT pointer mechanism for FFT memory management |
US7062523B1 (en) * | 2000-08-01 | 2006-06-13 | Analog Devices, Inc. | Method for efficiently computing a fast fourier transform |
US6963892B2 (en) * | 2001-12-26 | 2005-11-08 | Tropic Networks Inc. | Real-time method and apparatus for performing a large size fast fourier transform |
CA2377623C (en) * | 2002-03-20 | 2008-04-22 | Dongxing Jin | Method and apparatus for computation reduction for tone detection |
KR20040032683A (en) * | 2002-10-10 | 2004-04-17 | 엘지전자 주식회사 | Fast fourier transform apparatus for wireless lan system |
KR100577005B1 (en) * | 2003-12-27 | 2006-05-10 | 한국전자통신연구원 | Apparatus and method of selectively performing fast hadamard transform and fast fourier transform |
KR100518797B1 (en) * | 2004-01-07 | 2005-10-05 | 삼성전자주식회사 | Fast Fourier Transform device capable of improving a processing speed and a method processing thereof |
KR100825771B1 (en) * | 2004-02-11 | 2008-04-28 | 삼성전자주식회사 | Fast fourier transformation processor and method using half-sized memory |
US20050198092A1 (en) * | 2004-03-02 | 2005-09-08 | Jia-Pei Shen | Fast fourier transform circuit having partitioned memory for minimal latency during in-place computation |
US7296045B2 (en) * | 2004-06-10 | 2007-11-13 | Hasan Sehitoglu | Matrix-valued methods and apparatus for signal processing |
US20060075010A1 (en) * | 2004-10-05 | 2006-04-06 | Wadleigh Kevin R | Fast fourier transform method and apparatus |
KR100618889B1 (en) * | 2005-02-12 | 2006-09-01 | 삼성전자주식회사 | Fast Fourier Transform processor capable of reducing the size of memories |
TWI298448B (en) * | 2005-05-05 | 2008-07-01 | Ind Tech Res Inst | Memory-based fast fourier transformer (fft) |
KR20070075946A (en) * | 2006-01-17 | 2007-07-24 | 장영범 | Method and apparatus for low-power fast fourier transform and broadcasting terminal using the same |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US3721812A (en) * | 1971-03-29 | 1973-03-20 | Interstate Electronics Corp | Fast fourier transform computer and method for simultaneously processing two independent sets of data |
GB2006485A (en) * | 1977-10-07 | 1979-05-02 | Secr Defence | Improvements in or relating to Spectrum Analysers |
GB1546173A (en) * | 1977-03-11 | 1979-05-16 | Secr Defence | Signal processing equipment |
US5038311A (en) * | 1990-08-10 | 1991-08-06 | General Electric Company | Pipelined fast fourier transform processor |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US3617720A (en) * | 1967-09-12 | 1971-11-02 | Bell Telephone Labor Inc | Fast fourier transform using hierarchical store |
BE757750A (en) * | 1969-12-31 | 1971-04-01 | Thomson Csf | IMPROVEMENTS TO REAL-TIME ELECTRIC SIGNAL PROCESSING DEVICES |
US3673399A (en) * | 1970-05-28 | 1972-06-27 | Ibm | Fft processor with unique addressing |
JPS62175866A (en) * | 1986-01-30 | 1987-08-01 | Nec Corp | Signal processor |
JPH0668123A (en) * | 1992-05-22 | 1994-03-11 | Nec Corp | Signal processing circuit |
JP3675537B2 (en) * | 1995-11-29 | 2005-07-27 | 富士通株式会社 | Memory distributed parallel computer performing fast Fourier transform and method thereof |
-
1996
- 1996-10-21 SE SE9603862A patent/SE507529C2/en not_active IP Right Cessation
-
1997
- 1997-09-29 AU AU47304/97A patent/AU4730497A/en not_active Abandoned
- 1997-09-29 EP EP97909781A patent/EP1008060B1/en not_active Expired - Lifetime
- 1997-09-29 US US09/284,752 patent/US6356926B1/en not_active Expired - Lifetime
- 1997-09-29 WO PCT/SE1997/001637 patent/WO1998018083A1/en active IP Right Grant
- 1997-09-29 CA CA002269464A patent/CA2269464A1/en not_active Abandoned
- 1997-09-29 DE DE69737699T patent/DE69737699T2/en not_active Expired - Lifetime
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US3721812A (en) * | 1971-03-29 | 1973-03-20 | Interstate Electronics Corp | Fast fourier transform computer and method for simultaneously processing two independent sets of data |
GB1546173A (en) * | 1977-03-11 | 1979-05-16 | Secr Defence | Signal processing equipment |
GB2006485A (en) * | 1977-10-07 | 1979-05-02 | Secr Defence | Improvements in or relating to Spectrum Analysers |
US5038311A (en) * | 1990-08-10 | 1991-08-06 | General Electric Company | Pipelined fast fourier transform processor |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB2391966A (en) * | 2002-08-15 | 2004-02-18 | Zarlink Semiconductor Ltd | A method and system for performing a fast-fourier transform |
GB2391966B (en) * | 2002-08-15 | 2005-08-31 | Zarlink Semiconductor Ltd | A method and system for performing a fast-fourier transform |
US7024443B2 (en) | 2002-08-15 | 2006-04-04 | 1021 Technologies Kk | Method and system for performing a fast-Fourier transform |
CN1688104B (en) * | 2005-06-20 | 2010-11-10 | 北京中星微电子有限公司 | Digital signal processing method and apparatus |
Also Published As
Publication number | Publication date |
---|---|
DE69737699D1 (en) | 2007-06-14 |
SE507529C2 (en) | 1998-06-15 |
EP1008060A1 (en) | 2000-06-14 |
EP1008060B1 (en) | 2007-05-02 |
SE9603862L (en) | 1998-04-22 |
US6356926B1 (en) | 2002-03-12 |
AU4730497A (en) | 1998-05-15 |
DE69737699T2 (en) | 2008-01-10 |
CA2269464A1 (en) | 1998-04-30 |
SE9603862D0 (en) | 1996-10-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP1008060A1 (en) | A device and method for calculating fft | |
US7164723B2 (en) | Modulation apparatus using mixed-radix fast fourier transform | |
US6061705A (en) | Power and area efficient fast fourier transform processor | |
US7233968B2 (en) | Fast fourier transform apparatus | |
US7702712B2 (en) | FFT architecture and method | |
US6366936B1 (en) | Pipelined fast fourier transform (FFT) processor having convergent block floating point (CBFP) algorithm | |
US20080208944A1 (en) | Digital signal processor structure for performing length-scalable fast fourier transformation | |
Martens | Recursive cyclotomic factorization--A new algorithm for calculating the discrete Fourier transform | |
US20080071848A1 (en) | In-Place Radix-2 Butterfly Processor and Method | |
US8917588B2 (en) | Fast Fourier transform and inverse fast Fourier transform (FFT/IFFT) operating core | |
US7415584B2 (en) | Interleaving input sequences to memory | |
US20050177608A1 (en) | Fast Fourier transform processor and method using half-sized memory | |
US7428564B2 (en) | Pipelined FFT processor with memory address interleaving | |
US20060200513A1 (en) | Fast Fourier transform processor and method capable of reducing size of memories | |
US20050289207A1 (en) | Fast fourier transform processor, dynamic scaling method and fast Fourier transform with radix-8 algorithm | |
US6157938A (en) | Fast fourier transform device with parallel lattice architecture | |
US6728742B1 (en) | Data storage patterns for fast fourier transforms | |
Lee et al. | Modified sdf architecture for mixed dif/dit fft | |
KR20060066869A (en) | Apparatus and method of fft for the high data rate | |
KR19990077845A (en) | Pipelined fast fourier transform processor | |
US20060089822A1 (en) | FFT/IFFT processing system employing a real-complex mapping architecture | |
KR100890768B1 (en) | Data transforming processor and ofdm receiver having the same | |
CN118051709A (en) | FFT processor and operation method | |
WO2005052808A1 (en) | Pipelined fft processor with memory address interleaving | |
WO2003041010A2 (en) | Method and system for performing fast fourier transforms and inverse fast fourier transforms |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): AL AM AT AU AZ BA BB BG BR BY CA CH CN CU CZ DE DK EE ES FI GB GE GH HU ID IL IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MD MG MK MN MW MX NO NZ PL PT RO RU SD SE SG SI SK SL TJ TM TR TT UA UG US UZ VN YU ZW AM AZ BY KG KZ MD RU TJ TM |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): GH KE LS MW SD SZ UG ZW AT BE CH DE DK ES FI FR GB GR IE IT LU MC |
|
DFPE | Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101) | ||
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
ENP | Entry into the national phase |
Ref document number: 2269464 Country of ref document: CA Ref country code: CA Ref document number: 2269464 Kind code of ref document: A Format of ref document f/p: F |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1997909781 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 09284752 Country of ref document: US |
|
REG | Reference to national code |
Ref country code: DE Ref legal event code: 8642 |
|
WWP | Wipo information: published in national office |
Ref document number: 1997909781 Country of ref document: EP |
|
WWG | Wipo information: grant in national office |
Ref document number: 1997909781 Country of ref document: EP |