Page 490 - Introduction to Information Optics

Chapter 9 Computing with Optics

2
^uoqiang Li and Mohammad S. Alam
SHANGHAI INSTITUTE OF OPTICS AND FINE MECHANICS,
CHINESE ACADEMY OF SCIENCES, CHINA
2
THE UNIVERSITY OF SOUTH ALABAMA, MOBILE, ALABAMA

In this chapter, we review the recent advancements in parallel optoelectronic
signed-digit computing. We begin by discussing signed-digit number systems,
including modified signed-digit, trinary signed-digit, quaternary signed-digit,
and negabinary signed-digit representations. Then, we cover the parallel arith-
metic algorithms and architectures, fast-conversion algorithms between the
signed and unsigned number systems, encoding, and sample experimental veri-
fication via optoelectronic implementation. The parallelism of the algorithms
and the architectures are well mapped by efficiently utilizing the inherent
parallelism of optics. By exploiting the redundancy of the signed-digit number
systems, addition and subtraction can be completed in fixed steps independent
of the length of the operands. The table lookup, content-addressable-memory,
symbolic substitution, and parallel logic array approaches to optical comput-
ing can utilize the benefit of optics in parallel information processing and may
lead to powerful and general purpose optoelectronic computing systems.
Digital optical systems can be built by cascading two-dimensional planar
arrays interconnected in free space. The proposed algorithms and architectures
show great promise, especially with the development of optical interconnection,
optoelectronic devices, and optical storage technology.

475