Physical and Technological Limitations of Optical Information Processing and Computing
- PDF / 776,923 Bytes
- 6 Pages / 604.8 x 806.4 pts Page_size
- 66 Downloads / 213 Views
36
ing elements. A wide ränge of computational problems exist that lend thems e l v e s q u i t e n a t u r a l l y to o p t i c a l processing architectures, including pattern recognition, earth resources data acquisition and analysis, texture discrimination, synthetic aperture radar (SAR) image formation, radar ambiguity function generation, spread spectrum identification a n d analysis, systolic array processing, phased array beam steering, and artificial (robotic) vision. In addirion, many neural network processes that inherently rely on intricate interconnection patterns have been or can be implemented optically. These and other applications are treated in more detail in accompanying articles in this special issue of the MRS Bulletin1-2 and in special issues of IEEE Proceedings3 and Optical Engineering* on optical Computing. A generalized optical processor or Computer can be depicted schematically as shown in Figure 1. The physical cons t i t u e n t e l e m e n t s of such a System include a central processing unit (CPU) that performs the essential implementable function, a data management processor that orchestrates the flow of data and sequence of Operations (usually considered part of the CPU in a traditional electronic Computer), several types of m e m o r y elements for both short-term and long-term data storage
and buffering, format devices to spatially organize input data fields, input devices to convert data input types to a form amenable to subsequent processing, Output devices to convert p r o cessed results to detectable and interp r e t a b l e f o r m s , a n d d e t e c t o r s to produce externally addressable results. In Figure 1, feedback interconnects are explicitly s h o w n as separate components to emphasize their crucial role in implementing parallel iterative a l g o r i t h m s a n d complex w e i g h t i n g functions. A wide variety of optical components are required to implement processors based on the generalized architecture shown in Figure l. 5 These include oneand two-dimensional spatial light modulators, volume holographic optical elements, threshold arrays, optical memory elements, sources, source arrays, detectors, and detector arrays. The State of the technology is such that while demonstration devices and prototypes are proliferating, with the exception of sources, detectors and detector arrays, few (if any) such components have as yet achieved significant commercial success or even demonstrated technological viability. In large part, this is due to the fact that each of the candidate technologies has placed rather severe demands on the state-of-the-art of the materials employed regardless of the nature of the optical effect utilized (e.g., electrooptic, magnetooptic, photorefractive, or electroabsorptive). In other words, the magnitudes of observable optical perturbations per unit excitation are just not large enough with readily available materials to allow flexible device engineering. As such, the answer to whether optical processing and Computing will come of age may
Data Loading...
