Predicting execution-time for variable behaviour embedded real-time components

Embedded systems for vehicle control critically depend on efficient and reliable control software, together with practical methods for their production. Component-based software engineering for embedded systems is currently gaining ground since variability, reusability, and maintainability are supported. However, existing tools and methods do not guarantee efficient resource usage in these systems. In this paper we present a method, which increases the accuracy of execution time predictions for embedded components without lowering reusability of the components. For assessing the correct timing behaviour, the method classifies run types by their time in addition to their probability