Research On Scheduling Algorithm For Energy Consumption Optimization Of Computing System Of Connected Vehicles

People’s demand for comfortable,intelligent,and safe automobiles has reshaped the global traditional automobile industry and promoted the development of automobiles in the direction of the new four modernization(electrification,networking,intelligence,and sharing).In the era of Intelligent Connected Vehicle(ICV),new energy vehicles are an inevitable trend in the development of ICV.At present,new energy vehicles are mainly pure electric vehicles,so energy consumption has become a new limitation in ICV,and energy consumption optimization in computing system structure has become a new research hotspot.The essence of the ICV is a real-time heterogeneous distributed embedded system that is unchanged,so real-time is also a factor that cannot be ignored.This paper mainly starts from the computing system structure and designs the task scheduling algorithm under the premise of satisfying the real-time performance of the vehicle to realize the optimization of energy consumption.The main work and contributions of this paper are as follows:The introduction of the new four modernization of automobiles have a huge impact on the current automobile industry.The traditional automotive distributed electronic/electrical(E/E)architecture has many defects and cannot meet the development needs of intelligent networked vehicles.Major companies are committed to proposing a new generation of automotive E/E architecture.According to the future automotive E/E architecture proposed by current car companies,parts suppliers,and solution providers,this paper presents an E/E architecture for ICV.Based on this architecture,a corresponding scheduling algorithm is designed according to the computing system structure,aiming at optimizing the energy consumption of the processor scheduling task under the premise of satisfying the real-time performance of the computing system structure.There are various types of functional applications in automobiles,and this paper mainly considers two of them.One of them is single-task(functional application)scheduling,which is also the task model studied in most articles,and the other is hybrid task scheduling.For single-task scheduling,this paper first model the system,functional applications,and energy consumption.Then an Integer Linear Programming Scheduling(ILPS)algorithm is proposed to optimize the energy consumption algorithm under the deadline limit while the frequency is fixed.In the actual scheduling,each processor supports Dynamic Voltage/Frequency Scaling(DVFS)technology and the frequency is variable,so this paper further proposed the DVFS-Available Energy Saving(DAES)algorithm and Energy Further Saving(EFS)scheduling algorithm,combing table scheduling and integer linear programming method to solve the problem of optimizing energy consumption for single task scheduling under real-time constraints.For hybrid task scheduling,since the start time of many functional applications in ICV is random and unpredictable,it will increase the difficulty of scheduling.When a random functional application arrives,it will compete with the running functional application for computing resources.How to design a scheduling algorithm to optimize energy consumption while all functional applications can be completed within their respective deadlines is one of the research focus of this paper.To solve this problem,this paper first modeled the application of hybrid task scheduling,and then proposed Hybrid Task Integer Linear Programming Scheduling(HTILPS)and Hybrid Task DVFS Energy Saving(HTDAES)scheduling algorithms respectively to solve the energy consumption optimization problem in the case of frequency-invariant and frequencyavailable.Finally,taking diamond graph,linear algebra,Gaussian elimination,and random application model as examples,the superiority of the proposed algorithm is verified by comparing it with the state-of-the-art algorithms through experiments.According to the experimental results,the algorithm proposed in this paper outperforms the state-of-the-art scheduling algorithm in both scheduling length and energy consumption.