Large-grain, dynamic control system architectures

Most previous work in distributed real-time systems has focused on the task level of abstraction, and has assumed that all tasks have behavioral characteristics that can be determined statically. When attempting to apply these techniques to some applications (such as shipboard AAW systems), problems arise. In response to this, the dynamic path paradigm has been developed. Novel aspects of the dynamic path paradigm include its large granularity and its ability to accommodate systems that have dynamic variability. This paper provides an introduction to the dynamic path paradigm. Three classes of dynamic path are defined: the transient path, the continuous path, and the quasi-continuous path. To illustrate these classes of paths, an air defense application is modeled using dynamic paths.