|Shahriar, Md. Sumon
|This thesis report is submitted in partial fulfillment of the requirements for the degree of Bachelor of Science in Computer Science and Engineering, 2006.
|Cataloged from PDF version of thesis report.
|Includes bibliographical references (page 38).
|This paper proposes an asymmetric high-radix signed-digital (AHSD) adder for addition
on the basis of neural network (NN) and shows that by using NN the AHSD number
system supports carry-free(CF) addition. Besides, the advantages of the NN are the
simple construction in high speed operation. Emphasis is placed on the NN to perform
the function of addition based on the novel algorithm in the AHSD number system.
Since the signed-digit number system represent the binary numbers that uses only one
redundant digit for any radix r 2, the high-speed adder in the processor can be realized
in the signed-digit system without a delay of the carry propagation. A Novel NN design
has been constructed for CF adder based on the AHSD (4) number system is also
presented. Moreover, if the radix is specified as r = 2m, where m is any positive integer,
the binary-to-AHSD(r) conversion can be done in constant time regardless of the wordlength.
Hence, the AHSD-to-binary conversion dominates the performance of an AHSD
based arithmetic system.
In order to investigate how NN design based on the AHSD number system achieves its
functions, computer simulations for key circuits of conversion from binary to AHSD (4)
based arithmetic systems are made. The result shows the proposed NN design can
perform the operations in higher speed than existing CF addition for AHSD.
|BRAC University thesis are protected by copyright. They may be viewed from this source for any purpose, but reproduction or distribution in any format is prohibited without written permission.
|Computer science and engineering
|On the realization of asymmetric high radix signed digital adder using neural network
|Department of Computer Science and Engineering, BRAC University
|B. Computer Science and Engineering