Research On Energy-aware Scheduling In Automotive Cyber-Physical Systems

Automotive Cyber Physical Systems(ACPS)is a typical application of Cyber Physical Systems in the automotive field.In essence,ACPS is a heterogeneous distributed embedded system,which interacts with the physical world in real-time,and integrate computing and network deeply.The new energy automobile such as blade electric vehicles significantly oriented the development of future automotive industry,so energy resourses will become a particularly important factor in the design phase of ACPS.Therefore,how to achieve low power consumption and ensure the real-time and reliability of applications in ACPS is a new research hotspot.For the perspective of energy aware scheduling,this paper studies the real-time and reliability of distributed function modeled by DAG in ACPS under the condition of limited system energy,and adopts Dynamic Voltage and Frequency Scaling(DVFS)technology to achieve the purpose of energy conservation.The main work and contributions are as follows:(1)This paper first focuses on the research of schedule length optimization under energy constraint in ACPS.The problem is decomposed into two sub-problems,namely,energy allocation and scheduling length optimization.We propose an algorithm named Efficient Scheduling with Energy Consumption Constraint(ESECC)to solve the problem of minimizing the scheduling length of parallel application under energy constraint.The proposed algorithm realize the transfer of energy constraint of application to that of each task by a pre-allocation approach,and introduce a concept of available energy for energy pre-allocation of unscheduled tasks,and improves the pessimistic results caused by the unreasonable energy allocation of existing algorithms.(2)This paper then focuses on the research of reliability optimization under energy constraint in ACPS.The problem is decomposed into three sub-problems.namely,energy allocation meet the deadline and reliability optimization.We propose an algorithm named Reliability Enhancement with Response time and Energy Constraints(REREC)to solve the problem of reliability maximization of the parallel applications under energy constraints.REREC algorithm reallocates the energy based on the ESECC,and transfers the deadline of application to the time constraint of each task,then selects the processor with high reliability under both energy and time constraint.This study covers the shortage of current research which merely considering the constraint of energy or deadline.(3)Extensive comparative experiments were do to verify the algorithms proposed in this paper.Experiments on the real parallel applications validate the proposed ESECC algorithm can effectively reduce the schedule length of applications while satisfying the energy constraints.Experiments on the real-life automotive function and synthetic randomly generated automotive functions show that the proposed REREC algorithm can still achieve very high reliability while satisfying energy constraints and not exceeding application deadlines.Therefore,our research could facilitates a part of energy-aware scheduling during the design phase of ACPS.