Abstract
Keywords
Control, real-time implementation, distributed emmbedded systems, simulation, design lifecycle.
Abstract
The challenge in embedded systems is to design more and more complex applications while taking into account strong cost constraints such as financial costs, consumption and footprint. To reduce hardware cost, the automotive industry has chosen to use distributed architectures made of ECUs (Electronic Control Unit) linked together by networks such as CAN, FlexRay, ... [8]. Processors are placed near sensors and actuators, sensors data may be used by many functions : the number of sensors and the wiring may be reduced. This increasing hardware complexity impact on software complexity design : computations have to be placed and scheduled on ECUs and communications must also be scheduled on the network taking into account the real-time constraints. The design of a real time control embedded system usually starts by a modelling step where control engineers describe mathematically the behaviour of the system to be controlled (control law synthesis). The behaviour is validated by simulation, then the control laws are implemented on the hardware architecture by computer engineers.