Skip to Main Content
Frequently Asked Questions
Submit an ETD
Global Search Box
Need Help?
Keyword Search
Participating Institutions
Advanced Search
School Logo
Files
File List
29112.pdf (3.42 MB)
ETD Abstract Container
Abstract Header
A Comparative Study on Methods for Stochastic Number Generation
Author Info
Shenoi, Sangeetha Chandra
Permalink:
http://rave.ohiolink.edu/etdc/view?acc_num=ucin1511881394773194
Abstract Details
Year and Degree
2017, MS, University of Cincinnati, Engineering and Applied Science: Computer Engineering.
Abstract
Stochastic computing is a re-emerging method of approximate computing. The main advantages of stochastic computing - low hardware requirements for computing elements and error tolerant properties has made it an attractive method for machine learning applications. However, these advantages of stochastic computing are offset by the hardware requirements and low speed of generation of stochastic numbers from a conventional binary bit stream. There are many methods of generating stochastic numbers that have been proposed. We conduct a comparative study on various methods with regards to their statistical properties and hardware implementation cost in terms of area overhead. As a part of this work, we explore the behavior of cellular automata (CA) when used for pseudo random number generation in stochastic computing (SC). We compare the accuracy and statistical properties with a conventionally used linear feedback shift register (LFSR). In addition, several variations of the basic stochastic number generator are explored. For example, we consider two stochastic number generators sharing one random number generator, where bit permutation is used to reduce correlation. We also compare the statistical and hardware properties of other designs in the literature, including those employing an S-Box or circular shifting. In addition, we also explore and compare the properties when the bits are inverted to generate a new random number.
Committee
Carla Purdy, Ph.D. (Committee Chair)
Wen-Ben Jone, Ph.D. (Committee Member)
George Purdy, Ph.D. (Committee Member)
Pages
59 p.
Subject Headings
Computer Engineering
Keywords
stochastic computing
;
cellular automata
;
hardware design
;
statistical properties
;
stochastic number generation
Recommended Citations
Refworks
EndNote
RIS
Mendeley
Citations
Shenoi, S. C. (2017).
A Comparative Study on Methods for Stochastic Number Generation
[Master's thesis, University of Cincinnati]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1511881394773194
APA Style (7th edition)
Shenoi, Sangeetha Chandra.
A Comparative Study on Methods for Stochastic Number Generation.
2017. University of Cincinnati, Master's thesis.
OhioLINK Electronic Theses and Dissertations Center
, http://rave.ohiolink.edu/etdc/view?acc_num=ucin1511881394773194.
MLA Style (8th edition)
Shenoi, Sangeetha Chandra. "A Comparative Study on Methods for Stochastic Number Generation." Master's thesis, University of Cincinnati, 2017. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1511881394773194
Chicago Manual of Style (17th edition)
Abstract Footer
Document number:
ucin1511881394773194
Download Count:
780
Copyright Info
© 2017, some rights reserved.
A Comparative Study on Methods for Stochastic Number Generation by Sangeetha Chandra Shenoi is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License. Based on a work at etd.ohiolink.edu.
This open access ETD is published by University of Cincinnati and OhioLINK.