Determining Reliable and Precise Execution Time Bounds of Real-Time Software

Abstract

Embedded-control systems in transportation, medical instruments, and manufacturing are often time critical, that is, they have to fi nish execution within their period or they have to respond to sensor input within a given deadline. Both period and deadline are dictated by the physics of the system. Examples for sensor-actuator control are the airbag and the automatic brake control in cars. Clearly, if the airbag controller fails to inflate the airbag in time, the driver will hit the steering wheel. Flight control and guidance, collision-avoidance, or ground proximity warning systems in airplanes are examples from the aviation domain. Deadlines in airplanes are often in the order of milliseconds, while crankshaft-synchronous tasks in cars often have periods on the order of microseconds. A rigid verification of time-critical software should include a static timing verification. Measurement-based methods cannot guarantee to hit the worst case and to determine its execution time (WCET) because the number of cases to consider is too large to allow exhaustive methods. Certification rules and practice of the European Union Aviation Safety Agency (EASA) requires the use of static timing analysis in the certification of time-critical plane components.