Execution-Time Properties of a Hybrid Controller

We are interested in gaining more knowledge about the execution-time requirements of real-time control tasks, and this paper presents a system for gathering statistics of the execution times of real-time tasks. The system allows real-time code to be instrumented in order to gather, analyze, and present execution-time statistics. It has been used to instrument an existing hybrid controller to gather execution-time statistics. These statistics indicate that the worst-case execution times of the controller occur in conjunction with reference and mode changes, and that a scheduling based on that worst-case execution time leads to poor CPU utilization.