An Investigation into Cost Modelling for Design of Distributed Automotive Electrical Architectures

The emerging automotive electrical architecture is a distributed real-time control system whose functions are integrated by a combination of shared memory, hardwired connections and in-vehicle control networks such as CAN, LIN and FlexRay. The uptake of AUTOSAR for the integration of application software, with common embedded device driver components, offers the opportunity for the partitioning and business model of automotive electrical architecture design to change. Currently control of a system is centralised into a single function specific ECU requiring hard wiring to all sensors and actuators. Situations such as this lead to an over complication in system partitioning, unnecessary wiring and connectors, and often a large number of medium performance ECUs integrated by a number of CAN buses. A neater, simpler and more cost effective solution to the design and system partitioning problem may be to reduce the number of specialist ECUs integrated by CAN and replace with generic ECUs integrated with a faster network such as FlexRay. The problem then becomes how to determine the trade off between all of the possible design parameters which may include cost, electronic control system functionality, sensor/actuator placement. This paper describes an investigation into cost modelling of a vehicle electrical architecture to support decisions on system partitioning and in-vehicle control network choice. A LIN bus cost model is developed and it is shown how this can in turn be developed into a cost model for the entire vehicle electrical architecture based on hardware and software costs.