Title :
A systolic SBNR adaptive signal processor
Author :
Andrews, Michael
fDate :
2/1/1986 12:00:00 AM
Abstract :
A new realization for adaptive signal processing units is proposed which uses a special subset of signed digit number representations (SDNR´s). This signed binary number representation (SBNR) captures all of the efficiencies of SDNR arithmetic but also makes circuit realizations less complex. Furthermore, a natural interface between analog and digital numbers is provided. The serial on-line processing nature of SBNR utilizes the MSB first. An area/time complexity for VLSI implementations in comparable systolic array architectures contrasts the effectiveness of five different primitive VLSI cells and organizations.
Keywords :
Adaptive signal processing; MOS integrated circuits; Multivalued logic circuits; Redundant number systems; Systolic arrays; VLSI; Very large-scale integration (VLSI); Adaptive signal processing; Arithmetic; Integrated circuit interconnections; Least squares approximation; Multivalued logic; Signal processing; Signal processing algorithms; Silicon; Systolic arrays; Very large scale integration;
Journal_Title :
Circuits and Systems, IEEE Transactions on
DOI :
10.1109/TCS.1986.1085883