This thesis will present a new topology independent network architecture for highly fault tolerant, mesh interconnected Local Area Networks(LAN). This architecture was developed within the IEEE 802 Local Computer Reference Model framework and a LAN utilising this architecture will act as a direct replacement for the Media Access Control (MAC) and Physical (PHY) layers of that reference model. The only current MAC protocol that provides a high level of fault tolerance and topological independence is the flooding protocol. The problems with this protocol are highlighted and as a result, a new protocol that combines redundant routing with learning automata techniques has been developed. The learning automata provide the protocol with the capability to detect and adapt to changes in the network connectivity while redundant routing is utilised to cope with the immediate effects of failures. Detailed analysis of the performance of the new learning automata protocol is given. The steady state or long term performance analysis has shown that this protocol offers considerable improvement when compared to the flooding protocol and the protocol's non-learning equivalent. The learning automata protocol is capable of combining the advantages of the other protocols whilst minimising their disadvantages. When considering the performance of the learning automata protocol under failure conditions, the advantages of its learning capability are clear. Even under high numbers of failures, the protocol is capable of adapting promptly to the resulting changes in network connectivity and is able to optimise the use of the remaining communications capacity.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

The currently prevalent forms of display controller hardware are discussed in terms of their architecture, and their relationship to the software which forms the graphics packages with which they are to be used. New architectures are proposed utilising both conventional, commercially available components and special purpose Large Scale Integrated circuits. The concept of a "hardware graphics package" Is developed and implementation details presented. The conclusions drawn from this work are that in order to provide an appropriate environment for the type of interactive graphics capabilities which will form the hub of much future software, more emphasis must be placed on intelligent display systems. This distributed approach requires that a host computer provides a display controller with a high-level scene description which is subsequently rendered to constituent polygonal facets by the controller. Substantial benefits accrue from this reduced dependence upon a single Central Processing element.