Method and apparatus for computing the discrete Fourier transform recursively

• US 4,023,028 A
• Filed: 08/20/1974
• Issued: 05/10/1977
• Est. Priority Date: 08/20/1974
• Status: Expired due to Term

• Current Assignee: the united states of america as represented by the secretary of the navy
• Original Assignee: the united states of america as represented by the secretary of the navy
• Inventors: Dillard, George M.
• Primary Examiner(s): Malzahn, David H.
• Application Number:
  US05/498,888
• Time in Patent Office: 994 Days
• Field of Search: 235/156, 235/152, 324/77 B, 324/77 D
• US Class Current: 708/405
• CPC Class Codes: G06F 17/141 Discrete Fourier transforms

Abstract

A wholly digital system for computing the discrete Fourier transform of sequentially received data in a recursive fashion. Two parallel shift registers store and shift the real and imaginary components of the complex number X_k + iY_k. The data in the parallel registers are successively shifted one bit per strobe in response to receipt of new data. Additional logic operates recursively on successive data inputs to compute the discrete Fourier transform.

Claims

1. An apparatus for computing the discrete Fourier transform of a waveform comprising:

input register means for receiving a series of digital data samples representative of said waveform;

subtract and hold means connected to said input register means for subtracting the oldest data sample stored in said input register from the newest data sample stored in said input register means;

an adder network connected to the output of said subtract and hold means;

a multiplier network having a first input connected to the output of said adder network and having a second input;

constants storage means connected to said second multiplier network input for outputting to said multiplier network a series of predetermined digital constants;

output register means having an input connected to said multiplier network and an output connected to said adder network second input;

timing means for gating said input and output register means, said subtract and hold means, said adder network, said multiplier network and said constants storage means at predetermined intervals.

2. The apparatus of claim 1 wherein:

said input register means comprises first and second input register means for receiving the real and imaginary components respectively of said digital data samples;

said subtract and hold means comprises first and second subtract and hold means;

said adder network comprises first and second adders;

said constants storage means comprises first and second constants storage devices; and

said output register means comprises first and second output register means.

3. The apparatus of claim 2 wherein said first constants storage device holds the real part of e².sup.π

ij/N, j = 0, 1 . . . , N-1 and said second constants storage device holds the imaginary part of e².sup.π

ij/N, j = 0, 1 . . . , N-1 where i = √

-1, j = an index indicating a particular Fourier coefficient and N = the number of digital samples stored in said input register means minus 1.

4. The apparatus of claim 3 wherein said timing means simultaneously gates said first and second input registers in response to each digital data sample;

said timing means simultaneously gates said first and second subtract and hold means subsequent to each gate to said input registers;

said timing means simultaneously gates said first and second adders and said multiplier network with a first series of pulses subsequent to each gate to said first and second subtract and hold means; and

said timing means gates said constants storage means and said output registers with a second series of pulses, each pulse of said second series of pulses following a pulse of said first series of pulses;

whereby in response to the receipt of each digital data sample, said first output register contains a series of the real part of Fourier coefficients and said second output register contains a series of the imaginary part of Fourier coefficients.

5. A method of computing the discrete Fourier transform of a waveform comprising the steps of:

receiving a series of digital data representative of said waveform;

storing sequentially said series of digital data representative of said waveform;

subtracting the oldest stored data from the newest stored data to obtain a difference each time a data is received;

generating a series of predetermined constants;

providing an output shift storage means;

adding sequentially each of the data stored in said output shift storage means to said difference to produce a series of addition signals;

multiplying each addition signal by a predetermined constant from said series of predetermined constants to obtain a series of products;

sequentially storing said series of products in said output shift storage means;

whereby said output shift storage means contains a series of Fourier coefficients.