Computational Power And Accuracy Trade-Offs In Optical Numerical Processors

Demetri Psaltis

doi:10.1117/12.960507

28 July 1986 Computational Power And Accuracy Trade-Offs In Optical Numerical Processors

Demetri Psaltis

Proceedings Volume 0614, Highly Parallel Signal Processing and Architectures; (1986) https://doi.org/10.1117/12.960507
Event: O-E/LASE'86 Symposium, 1986, Los Angeles, CA, United States

Abstract

The comparison between the capabilities of optical versus electronic signal processors is usually the first issue that is addressed when one considers using optics in a particular signal processing application. In general, it is necessary that optics offers a clear advantage over electronics in order to be able to justify opting for an optical approach. The comparison between optics and electronics is based on a number of related criteria whose relative importance depends on the application. These include processing power (e.g. the number of elementary operations performed per unit time), algorithmic flexibility, accuracy, power consumption, weight, cost. The primary advantage of analog optical systems is the extremely high processing power that can be achieved (in excessof 104 analog multiplication per second is possible with acoustooptic technology) with relatively small power and size requirements. The major disadvantages of conventional analog optical processors are lack of algorithmic flexibility (limited to linear transformations) and low accuracy. These two limitations along with immature device technologies have restricted the utilization of optical signal processors to only a small number of areas, such as synthetic aperture radar and acoustooptic spectrum analyzers. Thus there is strong motivation to find ways to improve the accuracy and extend the range of signal processing operations that can be optically performed in order to extend the applicability of optical techniques. In this paper we examine the method of digital multiplication by analog convolution (DMAC) [l] [4] which has received attention in recent years as a technique for building accurate optical processors. We will show that DMAC is fundamentally limited because it can provide accuracy only at the expense of using post-detection electronics that are more sophisticated than the electonics required to build the entire system electronically in the first place.

Citation Download Citation

Demetri Psaltis "Computational Power And Accuracy Trade-Offs In Optical Numerical Processors", Proc. SPIE 0614, Highly Parallel Signal Processing and Architectures, (28 July 1986); https://doi.org/10.1117/12.960507

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available