An error recovery technique for real-time distributed computer systems

Abstract

This thesis studies fault tolerant strategies for real-time distributed computer control systems so as to propose an error recovery technique that renders individual processors on the network resistant to soft failures. The technique is effective for soft failures which have as certain maximum duration, and does not require the use of specialised hardware. Attention is focused on achieving resistance to soft failures in environments which have demanding time constraints such as those found in computer systems for process control, materials handling and automated manufacturing. [Abbreviated Abstract. Open document to view full version]