Date of Graduation

2005

Document Type

Dissertation/Thesis

Abstract

With the proliferation of wireless networked computing devices in recent years, the importance of communication between nodes in these networks has increased greatly. The mobility of these nodes complicates the problem because a given node's connections to the network are likely to change over time with the result that communication paths are established ad hoc. As the speed of the nodes increases or the computing power available to them is reduced, many standard techniques for predicting available message pathways no longer work. In these highly mobile ad hoc networks, the links between nodes can change between the time they are detected and the time they are to be used. Another development in recent years has been the study of swarm intelligence which offers new insights to many problems involving autonomous agents. In a swarm intelligent system, the individual actors operate independently according to some simple set of rules and overall behavior patterns emerge. This work looks at various aspects of the problem of message broadcast in highly mobile ad hoc networks. It begins by examining broadcast in fixed grids. It then applies observations pertaining to grids to a series of simulations of highly mobile ad hoc networks using cellular automata. Methods for determining the efficiency of broadcast algorithms are developed. The effects of collisions on broadcast and the effects of toggling between different types of transmission neighborhoods are then studied. Formulas are developed for predicting the point at which an individual node can safely stop re-transmitting the message. Next, a radio-based simulator is developed and used to gauge the effectiveness and promise of the results of the earlier simulations. Finally, a swarm intelligence inspired protocol for message transmission is given.

Share

COinS