Claims
- 1. A charge domain computing device for performing a Fourier transform by simultaneous matrix-vector multiplication operations in respective plural charge coupled device/charge injection device arrays comprising:
- a plurality of charge coupled device/charge injection device arrays storing charge coupled device charge packets in respective charge coupled device cells arrayed in rows and column, said charge packets having amounts of charge corresponding to the values of corresponding elements of respective matrix operators;
- a Hartley transform processor for each of said arrays having a signal processing operation in a matrix vector multiplication operation for producing real and imaginary parts of said Fourier transform using a Hartley transform; and
- a device for simultaneously applying an incoming vector to said plurality of charge coupled device/charge injection device arrays;
- wherein each matrix-vector multiplication operation in respective plural charge coupled device/charge injection device arrays simultaneously produces said real and imaginary parts of said Fourier transform.
- 2. The device of claim 1 wherein each of said charge coupled device/charge injection device arrays comprises:
- an array of n rows and M columns of charge coupled device matrix cell groups corresponding to a matrix of N rows and M columns of matrix elements, each of said matrix elements representable with b binary bits of precision, each of said matrix cell groups comprising a column of b charge coupled device cells storing b charge coupled device charge packets representing the b binary bits of the corresponding matrix element, the amount of charge in each packet corresponding to one of two predetermined amounts of charge;
- each of said charge coupled device cells comprising a holding site and a charge sensing site, each charge packet initially residing at the respective holding site;
- a device for sensing, for each row, an analog signal corresponding to a total amount of charge residing under all charge sensing sites of the charge coupled device cells in the row;
- an array of c rows and M columns of charge coupled device vector cells corresponding to a vector of M elements representable with c binary bits of precision, each one of said M columns of charge coupled device vector cells storing a plurality of c charge packets representing the c binary bits of the corresponding vector element, the amount of charge in each packet corresponding to one of two predetermined amounts of charge; and
- a multiplying device operative for each one of said c rows of said charge coupled device vector cells for temporarily transferring to said charge sensing site the charge packet in each one of said m columns of matrix cells for which the charge packet in the corresponding one of said m columns and said one row of said charge coupled device vector cells has an amount of charge corresponding to a predetermined binary value.
- 3. The device of claim 2 further comprising arithmetic means operative in synchronism with said multiplying device, comprising:
- means for receiving, for each row, the signal sensed by said sensing device, whereby to receive N.times.b signals in each one of c operations of said multiplying means;
- means for converting each of said signals to a corresponding byte of output binary bits; and
- means for combining the output binary bits of all of said signals in accordance with appropriate powers of two to generate bits representing an N-element vector corresponding to the product of said vector and said matrix.
- 4. The device of claim 1 wherein said array of matrix charge coupled device cells is distributed among a plurality of b integrated circuits containing sub-arrays of said M columns and N rows of said matrix charge coupled device cells, each of said sub-arrays corresponding to a bit-plane of matrix cells representing bits of the same power of two for all of said matrix elements.
- 5. The device as set forth in claim 1, wherein the Hartley transform is a combination of fast Hartley transform and a discrete Hartley transform.
- 6. A charge domain computing device for performing a Fourier transform by simultaneous matrix vector multiplication operations in respective plural charge coupled device/charge injection device arrays comprising:
- a first charge coupled device/charge injection device array storing charge packets;
- a first Hartley transform processor for computing a first matrix operator represented by said charge packets derived from a first permutation of a Hartley transform and wherein said first array having a first matrix vector multiplication operation which produces the real part of the Fourier transform of an incoming vector;
- a second charge coupled device/charge injection device array storing charge packets;
- a second Hartley transform processor for computing a second matrix operator represented by charge packets derived from a second permutation of said Hartley transform and having a second matrix vector multiplication operation which produces the imaginary part of the Fourier transform of said incoming vector; and
- a device for simultaneously applying said incoming vector to said first and second charge coupled device/charge injection device arrays;
- wherein said first and second matrix-vector multiplication operations in respective plural charge coupled device/charge injection device arrays simultaneously produce said respective real and said imaginary parts of said Fourier transform.
- 7. The device of claim 6 wherein said first permutation corresponds to a sum of permutated and unpermutated version of said Hartley transform and said second permutation corresponds to a difference between said permutated and unpermutated versions of said Hartley transform.
- 8. The device of claim 6 wherein said Hartley transforms and said charge coupled device/charge injection device arrays are each of a size m which is a fraction of the size N of the Fourier transform and wherein said incoming vector is divisible into smaller vectors combinable into respective sums and differences and wherein said device for applying applies respective ones of said sums and differences to respective ones of said charge coupled device/charge injection device arrays.
- 9. The device of claim 8 wherein said incoming vector is divisible into two equal portions, each one of said first and second charge coupled device/charge injection device arrays comprising a pair of individual charge coupled device/charge injection device arrays of size M, four different permutations of said Hartley transform being embedded in the respective M-size charge coupled device/charge injection device arrays.
- 10. The device of claim 6 wherein each of said charge coupled device/charge injection device arrays comprises:
- an array of N rows and M columns of charge coupled device matrix cell groups corresponding to a matrix of N rows and M columns of matrix elements, each of said matrix elements representable with b binary bits of precision, each of said matrix cell groups comprising a column of b charge coupled device cells storing b charge coupled device charge packets representing the b binary bits of the corresponding matrix element, the amount of charge in each packet corresponding to one of two predetermined amounts of charge;
- each of said charge coupled device cells comprising a holding site and a charge sensing site, each charge packet initially residing at the respective holding site;
- a device for sensing, for each row, an analog signal corresponding to a total amount of charge residing under all charge sensing sites of the charge coupled device cells in the row;
- an array of c rows and M columns charge coupled device vector cells corresponding to a vector of M elements representable with c binary bits of precision, each one of M columns of charge coupled device vector cells storing a plurality of c charge packets representing c binary bits of the corresponding vector element, the amount of charge in each packet corresponding to one of two predetermined amounts of charge; and
- a multiplying device operative on each one of said c rows of said charge coupled device vector cells for temporarily transferring to said charge sensing site the charge packet in each one of said M columns of matrix cells for which the charge packet in the corresponding one of said M columns and said one row of said charge coupled device vector cells has an amount of charge corresponding to a predetermined binary value.
- 11. The device of claim 10 further comprising arithmetic means operative in synchronism with said multiplying device, comprising:
- means for receiving, for each row, the signal based by said sensing device, whereby to receive N.times.b signals in each one of c operations of said multiplying means;
- means for converting each of said signals to a corresponding byte of output binary bits; and
- means for combining the output binary bits of all of said signals in accordance with appropriate powers of two to generate bits representing an N-element vector corresponding to the product of said vector and said matrix.
- 12. The device as set forth in claim 6, wherein the Hartley transform is a combination of a fast Hartley transform and a discrete Hartley transform.
CROSS-REFERENCE TO RELATED APPLICATIONS
This is a division, of application Ser. No. 08/161,908, filed Nov. 30, 1993, now U.S. Pat. No. 5,508,538, which is a continuation-in-part of application Ser. No. 08/049,829, filed Apr. 19, 1993, now U.S. Pat. No. 5,491,650.
ORIGIN OF THE INVENTION
The invention described herein was made in the performance of work under a NASA contract, and is subject to the provisions of Public Law 96-517 (35 USC 202) in which the contractor has elected to retain title.
Divisions (1)
|
Number |
Date |
Country |
Parent |
161908 |
Nov 1993 |
|
Continuation in Parts (1)
|
Number |
Date |
Country |
Parent |
049829 |
Apr 1993 |
|