Impact analysis in real-time control systems

Abstract

Real time control systems typically receive streams of sampled data signals, process them, and generate output to control devices via actuators. Control system components make assumptions about sampling rate and other characteristics, such as filtering of data streams. These application-specific semantic dependencies and time-sensitive semantic properties are usually not explicit in the source code implementation. The paper discusses an approach to modeling the system architecture of real time applications, capturing semantic dependencies, and recording time-sensitive properties. Analysis of the model can identify inconsistencies in a system design and can determine the impact of a system change. This architectural dependency model can evolve incrementally to record hidden side effects, as they are discovered during maintenance, and prevent such errors in the future.