Variable word length DSP using serial-by-modulus residue arithmetic

  • Authors:

  • W. Kenneth Jenkins;Andrew J. Mansen

  • Affiliations:

  • Department of Electrical and Computer Engineering and The Coordinated Science Laboratory, Urbana, Illinois;Department of Electrical and Computer Engineering and The Coordinated Science Laboratory, Urbana, Illinois

  • Venue:
  • ICASSP'93 Proceedings of the 1993 IEEE international conference on Acoustics, speech, and signal processing: digital speech processing - Volume III
  • Year:
  • 1993

Quantified Score

Hi-index 0.00

Visualization

Abstract

The concept of a variable word length sum-of-products signal processing kemel is developed based on a serial-bymodulus residue number system (RNS) architecture. Due to the fact that the RNS is not a weighted number representation, if the instantaneous dynamic range requirement can be estimated it is possible to perform the computation with only enough residue digits to provide the necessary dynamic range. Although it is difficult to estimate instantaneous dynamic range requirements for arbitrary computational processes encountered in general purpose computers, in well defined signal processing tasks, such as the sum-of-products kemel found in digital filters and vector processors, it may be possible to take advantage of a variable word length design to improve computational efficiency for special-purpose processors.