Talks and Poster Presentations (with Proceedings-Entry):
W. Elmenreich, P. Peti:
"Achieving Dependability in Time-Triggered Networks by Sensor Fusion";
Talk: IEEE International Conference on Intelligent Engineering Systems,
- 2002-05-28; in: "Proceedings of the 6th IEEE International Conference on Intelligent Engineering Systems (INES)",
This paper presents a system architecture that supports the development of dependable control applications by sensor fusion algorithms.
Each measurement is represented as a compound of a name, a measurement instant, the measured value, and a confidence marker indicating the reliability and preciseness. Such sensor observations are processed by a network of fusion nodes in order to get data which is more dependable. The sensor fusion uses a probability model of sensor readings where the expected variance of a measurement corresponds directly to its confidence. Besides the fusion of different values the paper presents also approaches for fusing observations taken at different instants.
The presented approach has several advantages: First, it supports a modular system development, because the fusion algorithms are implemented only with respect to the interface specification and thus independent of the actual control application. This enables separate development and testing of subsystems that finally join together to form the whole application. Second, it supports the extension of existing applications in order to transform a real-time system that does not tolerate sensor faults into one that does. We use an analysis of the worst-case execution time of the resulting application. Given that the necessary timing constraints can be satisfied, the modified application will show the same temporal behavior as the original application.
Electronic version of the publication:
Created from the Publication Database of the Vienna University of Technology.