TECHNIQUES FOR THE DESIGN AND MANAGEMENT OF DYNAMIC COMPUTER NETWORKS

A dynamic computer network is a computer network in which the sites of the nodes (host computers), the physical connections among the nodes, and the communication paths among the nodes can all be varied with respect to time and space. The study of dynamic computer networks is motivated by the decreasing cost and size of hardware components which have made the installation of a computer network in multiple mobile units (such as aircrafts, human beings, and automobiles) feasible. The design and management of such computer networks is complicated by their inherently dynamic natures. Adaptability, reconfigurability and efficiency are the key concepts in the design of dynamic networks. Research has involved investigation of the design and management issues of dynamic computer networks, examining the evolution of dynamic computer networks, the characteristics of dynamic computer networks, routing control, and reconfiguration in such systems.;The characteristics of dynamic computer networks are analyzed based on a classification of communication subnetworks for distributed computer systems. Since the main difference between dynamic computer networks and traditional distributed computer systems is that the communication subnetwork of the former is much more complicated, the objective of the classification is to pinpoint the similarities and differences of dynamic computer networks with respect to traditional distributed computer systems.;In the area of relay management, a fast 2-node shortest path algorithm is designed, which can be used as the backbone of the routing algorithm. Under certain assumptions, the shortest path algorithm has an average execution time.;(DIAGRAM, TABLE OR GRAPHIC OMITTED...PLEASE SEE DAI).;Studies of the evolution of dynamic computer networks include a survey of the "predecessors" (prototypes) of dynamic computer networks, and investigations of the state-of-the-art of packet radio technology and VLSI technology. The recent advances of packet radio technology and VLSI technology has constituted driving force for the development of dynamic computer networks.;and worst-case execution time O((VBAR)E(VBAR)log(VBAR)E(VBAR)), where (VBAR)V(VBAR) is the number of nodes, and (VBAR)E(VBAR) is the number of edges in the network. This algorithm has the fastest average execution time among all the existing 2-node shortest path algorithms, and the same worst-case execution time as the best existing 2-node shortest path algorithm. The efficiency of the algorithm makes it suitable to be used in a dynamic environment. . . . (Author's abstract exceeds stipulated maximum length. Discontinued here with permission of school.) UMI.