The automotive industry is one of Europe’s driving economic forces. A huge variety of cars from premium cars, to e-mobile and low-cost cars, are developed and sold. Unseen, deep down into numerous electronic control units (ECU) engine control functions, comfort functions, driving assistance functions, as well as security functions, are implemented. In fact, this means the programme on a variety of processors, namely microcontrollers.
The amazing variety of functions implemented in software
In the automotive domain raises the need for specialised design methodologies. With automotive software engineering (ASE) a new design methodology has been founded, so fundamental that we set up a complete master program on this topic. Challenges in ASE are to handle efficient interaction with the sensoric connected to the peripherals, guarantee real-time requirements and separate hardware basis from software implementation. Last is crucial due to the huge variety of product versions, e.g. identic functions for different model lines implemented on different hardware platforms.
All hardware dependencies increase costs in terms of system complexity, testability and portability. We have evaluated the well-known AUTOSAR approach. This separates three levels: RTE, BasisSW and the microcontroller core. This approach allows the separation of hardware and software, not only in the final design stage but also in the development state. Unfortunately, the separation levels increase the complexity of virtual and final test processes. We developed a test method adapted to the level approach for so-called built-in-self-test.
A test module which has access to all levels individually is inserted into the design. Thus, communication can be tested for all levels separately and correct functionality can be proved. Extending our approach we added a module to detect in detail the timing for individual communication. Thus overall communication time can be quantised as well as the delay introduced by each level. Finally, our test module can be exchanged by a monitoring module. In the same way, critical system parameters can be evaluated and monitored. This allows us to integrate health monitoring within the automotive application. In the development of our test and monitoring approach for automotive software systems students from the automotive software engineering master program are involved.
Professor Dr Wolfram Hardt
Technische Universität Chemnitz
Tel: +49 371 531 25550
hardt@cs.tu-chemnitz.de
