Research On Low-energy Parallel Scheduling Algorithms For Automotive CPS And Industrial IoT Devices

The development of information technology promotes the evolution of embedded systems into Cyber-Physical Systems(CPS)and the Internet of Things(Io T).Automotive CPS and industrial Io T devices are typical applications of embedded systems in the CPS and Io T fields.Automotive CPS is a high safety-critical and energy sensitive cyber-physical system,so high safety and energy efficiency are the main goals of automotive CPS development and design considerations;industrial Io T devices are performance-critical and energy constrained Io T systems,so high performance and energy-aware are the main goals of industrial Io T device development and design considerations.In recent years,automotive CPS and industrial Io T devices have been designed to take advantage of multi-core and multi-processor performance to execute parallel applications with data dependencies and precedence constraints.Researches on scheduling algorithms based on parallel applications have emerged.However,existing researches on energy-efficient scheduling for automotive CPS only involve real-time and reliability-aware scheduling algorithms,and there are no related researches on energy-efficient scheduling algorithms that satisfy functional safety requirements;while existing researches on energy-aware scheduling for industrial Io T devices only consider dynamic energy constraint,and there are no energy-aware scheduling algorithms that satisfy the combination of dynamic and static energy constraints.Therefore,this paper aims to fill this gap,and proposes the following main works for the research of low-energy parallel scheduling algorithms using Dynamic Voltage and Frequency Scaling(DVFS)technology in automotive CPS and industrial Io T devices:1.This paper proposes an energy-efficient scheduling algorithm with two verifications and one optimization.This algorithm decomposes the CPS high-efficiency functional safety problem into three sub-problems,namely 1)response time requirement verification;2)functional safety requirement verification;3)energy consumption optimization under functional safety constraint;and based on automotive safety integrity level(ASIL)Decomposition to achieve energy-efficient functional safety scheduling algorithm design.Experiments conducted through real and synthetic automotive CPS applications show that the proposed algorithm can effectively reduce system energy consumption under the premise of meeting functional safety requirements,and solves the problem of automotive CPS energy-efficient functional safety scheduling.2.This paper proposes a model-measurement-design-optimization energy-aware scheduling framework ESF to satisfy the joint constraints on dynamic and static energy.The framework has four deeply integrated design flows,namely 1)modeling of power and energy constraints;2)measurement of power parameters based on industrial Io T devices;3)basic framework design including energy pre-assignment;4)framework optimization.The theoretical and practical co-verification experiments show that ESF can indeed guarantee the joint constraints of dynamic and static energy and optimize the system response time.ESF is a well-designed energy-aware scheduling framework.