Precise Measurement-Based Worst-Case Execution Time Estimation

A real-time computer system is a computer system in which the correctness of the system behavior depends not only on the logic results of the computations, but also on the physical instant at which these results are produced. Today, most real-time computer systems are implemented as collections of software tasks that are executed concurrently on a suitable hardware platform consisting of one or multiple microprocessors or microprocessor cores. Obtaining the worst-case execution time (WCET) of each real-time task is an essential step in ensuring the correctness of such a system. Measurement-based timing analysis (MBTA) is an easily retargetable analysis approach for estimating the WCET of a given task running on a particular target platform. In this approach, the execution times of individual task fragments are measured while the task is being executed on the intended target hardware. The observed local execution times of the individual task fragments are subsequently combined into a global WCET estimate for the task as a whole. Industry demandsWCET estimates that are close to the actualWCET. To fulfill this requirement, we must limit two opposed influences on the closeness of WCET estimates obtained by MBTA: (1) conservative WCET estimate calculation, otherwise known as pessimism, and (2) incomplete measurement coverage, also known as optimism