Validation of Networked Automotive Control Systems Using Global Predicates

Vehicles consist of many connected networks of electronic control units (ECUs). Automotive application software (e.g. traction control system, climate control, engine management) is distributed across many separate ECUs, making application testing and system integration very difficult. The main difficulty is in constructing a global application state, due to the asynchronous independent operation of each ECU. The aim of this research is to make the system integration more efficient by creating a global application state based on analysis of application test results. To achieve this goal, a prototype program was developed to construct global states and to analyse these global states. The prototype collects ECU state and network interaction data using the Vector CANoe tool. From these data a lattice of consistent global application states is constructed. The global state lattice can then be used as the basis for analysing ECU signal consistency across ECUs and identifying the potential for erroneous system states to be entered. Test results from the prototype demonstrate the validity of the theoretical approach despite the disadvantage of the state space explosion associated with large distributed systems.