Linear recursive networks.
Item
-
Title
-
Linear recursive networks.
-
Identifier
-
AAI9959232
-
identifier
-
9959232
-
Creator
-
Smires, Thami.
-
Contributor
-
Adviser: Seyed Ali Ghozati
-
Date
-
2000
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Computer Science | Engineering, Electronics and Electrical | Mathematics
-
Abstract
-
In recent years, the class of Linear Recursive Networks has been studied by many researchers. This thesis presents an extensive study of LRNs and their variances. It describes the static as well as dynamic properties of LRNs. In addition, the definition of the class of LRNs has been modified to cover a wider range of interconnection topologies. This has been accomplished by extending the alphabet over which the node labels of an LRN is generated to cover both numerals and non-numeral characters.;A hierarchy of LRNs has been defined and an important member of the class of LRNs, called the HLRN, has been studied exclusively. It is shown that the diameter of an n-dimensional HLRN is upper-bounded by O( n).;Several efficient routing strategies are developed for the class of LRNs. Some routing algorithms are applicable to all LRNs. Others only apply to some members of the class of LRNs.;Embedding properties of LRNs have been studied. Algorithms to embed an m-dimensional LRN into an n-dimensional one are presented. In particular, the embedding of an n-cube into an n-LRN is described. Partitioning of LRNs, which is essential in multiprocessing environment, is discussed. It is shown that the partitioning is process, which is controlled by the generator of the LRN, A.;Furthermore, a new cube-like interconnection topology, n-fastcube , is developed, The topological and routing properties of fast-cubes are investigated. The diameter of the fastcube if almost half of the diameter of the cube while its cost is the same as the cost of the cube. In the same context, a methodology to reduce the diameter of interconnection networks is introduced. Finally, fault-tolerance of LRNs is discussed and ways to handle node failures are presented.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
