Abstract

Modified signed-digit (MSD) number representation with its inherent weak interdigit dependence is capable of performing parallel addition and subtraction operations. A MSD-based multiplication scheme that contains an array of parallel additions has proved to be a competitive multiplication method. Optical implementation of an MSD arithmetic processor has recently been studied. A single-step optical content-addressable memory-based MSD adder has been proposed. Since to generate a single-step 1-bit MSD addition result, as many as fifty-six reduced logic product terms have to be stored in a CAM, direct parallel stacking of N + 1 such 1-bit MSD adders for an TV-bit MSD addition operation will occupy a huge volume. In this paper, a novel scheme is proposed to make a space-multiplexed optical CAM for an N-bit MSD addition. The N-bit input data are first permuted to form a data matrix. By modifying an existing triple-matrix multiplier, a space-multiplexed CAM is formed in a compact and efficient way.

© 1991 Optical Society of America