Abstract

The Inherent parallelism of optics opens new frontiers for parallel processing.¹ This fact coupled with the difficulty of VLSI technology is handling the communication needs placed by the parallel computing requirements has generated great interest in digital optical computing. One parallel technique proposed for optical processing is symbolic substitution. We first present algorithms for performing optical digital arithmetic using the signed-digit representation.² This number system is well suited for optical symbolic processing due to its carry-free addition and subtraction. The addition/subtraction processing time is independent of word length; it depends only on the characteristics of the optical devices used. Second, we present the substitution rules to carry out addition, subtraction, multiplication, and division. Third, we give several approaches to implement such an optical processor as well as discuss their design trade-offs. The proposed processor is a carry-free highly parallel system aimed at exploiting the inherent and unique ability of optics to process information in two space dimensions. Finally, we present a speedup analysis and compare it to an electronic implementation.

© 1987 Optical Society of America