Measurement-based worst-case execution time analysis

Traditional worst-case execution time (WCET) analysis methods based on static program analysis re-quire a precise timing model of a target processor. The construction of such a timing model is expensive and time consuming. In this paper we present a hybrid WCET analysis framework using runtime measurements together with static program analysis. The novel aspect of this frame-work is that it uses automatic generation of test data to derive the instruction timing of code sequences. Pro-gram paths are decomposed into subpaths to make run-time measurements feasible.