Model-based energy optimization of automotive control systems

Reducing the energy consumption of controllers in vehicles requires sophisticated regulation mechanisms. Better power management can be enabled by allowing the controller to shut down sensors, actuators or embedded control units in a way that keeps the car safe and comfortable for the user, with the goal of optimizing the (average or maximal) energy consumption. This paper proposes an approach to systematically explore the design space of SW/HW mappings to determine energy-optimal deployments. It employs constraint-solving techniques for generating deployment candidates and probabilistic analyses for computing the expected energy consumption of the respective deployment. The feasibility and scalability of the method is demonstrated by several case studies.