Automotive applications classed as safety-related or safety-critical are now important differentiating technologies in the automotive industry. The emergence of safety standard ISO 26262 underlines the increasing importance of safety in automotive software. As well as functional requirements, hard real-time requirements are of crucial importance to safety-related software as there is a need to prove that the system functionality is fulfilled, even in worst-case scenarios.
Measurement-based WCET (Worst-Case Execution Time) analysis combines on-target timing measurements with static analysis of program structure to calculate predicted worst-case paths and times. This is in contrast to traditional end-to-end timing measurements, which give no confidence that the worst-case path is actually tested and no insight into the location of any timing problems that do emerge.
This paper presents the principles of WCET analysis alongside the results of using a commercially available WCET toolkit within the development of a commercial, safety-related software project. The observed benefits include greatly increased confidence in timing correctness, even in worst-case scenarios, understanding of timing variability, location of timing hot-spots and greatly reduced costs of optimization.