Research On Mapping And Simulation Of Multi-core Real-Time System Resource Reservation

With real-time system more and more been used in the multi-core processor,there is growing emphasis on the study of real-time task scheduling problem.The current study focuses on the problem of how to ensure the task scheduling can be,real-time systems,the efficiency of the system and how to save resources and reduce the energy consumption of the processor.Therefore,this paper will focus on the task scheduling algorithms to improve the overall efficiency of the system.Real-time applications on multi-core processors by way of the operation of resource reservation,which can be divided into multiple tasks and each task to deploy to the processor core.In order to save computing resource usage and reduce energy consumption of processors,the processor core should be use less as far as possible when the deployment.This paper proposes the task deployment issues about the use of Packing problem(Bin Packing)in the real-time system resources reserve method.In order to solve this problem,the paper propose to replace the population based on random genetic algorithm(I_GA)to achieve the basic fish algorithm(Genetic Algorithm,hereinafter referred to GA)improved algorithm based on fish and fish invasion mode(I_AFSA)to achieve the improvement of the fish swarm(Artificial Fish Swarm Algorithm,hereinafter referred to AFSA).Finally,using the Matlab software implements a real-time scheduling integrated simulation platform.The main work of this thesis is:(1)In chapter ?,this paper mainly introduces the application background of real-time system,then introduces the real-time system development and research status at home and abroad,finally,the paper sum up the main content and structure arrangement.(2)In chapter ?,this paper mainly introduces the relevant theoretical knowledge in real-time systems,including knowledge of real-time systems,real-time scheduling,real-time system model and other knowledge.(3)In chapter ?,this paper analyzes and studies the problem of task mapping in multi core real time system.Assign tasks to the processor core set and requires a minimum of processor cores occupy this issue is mapped to a one-dimensional packing problem.Also it proposes task scheduling model and done some research about scheduling algorithms for real-time tasks.(4)In chapter ?,this paper presents that replace genetic algorithm based on random population(I_GA)to solve multi-task scheduling problem tracking system.In this chapter,this paper propose an improved algorithm(I_GA)based on genetic algorithm(GA),which is based onrandom population replacement.Then the performance of GA algorithm and I_GA algorithm are compared from several aspects.The experimental results show that the I_GA algorithm is better than the GA algorithm in terms of saving processor core resources,improving the performance and convergence speed of the algorithm and also meets the requirements of real-time to a certain extent.(5)In chapter ?,the paper presents the problem of task scheduling in multi core real time system based on the fish swarm intrusion pattern swarm algorithm(I_AFSA).In this chapter,we propose a fish swarm algorithm(I_AFSA)based on the artificial fish swarm algorithm(AFSA).Then,the AFSA algorithm and I_AFSA algorithm are compared from several aspects.The experimental results show that the I_AFSA algorithm is better than AFSA in terms of saving processor core resources,improving the efficiency of the algorithm and convergence rate and also meets the requirements of real-time to a certain extent.(6)In chapter ?,a real-time task scheduling simulation platform is designed.The simulation platform is integrated with the GA algorithm,I_GA algorithm,AFSA algorithm and I_AFSA algorithm.Through this integrated simulation platform,we can get the relationship between the number of iterations and adaptive value/the fitness value,but also can get the number of processor cores and the time used by the algorithm.In order to verify the performance of I_GA algorithm and I_AFSA algorithm,this paper compares the two kinds of improved algorithm and the non improved algorithm in the R2012 b MATLAB software,which is based on the efficiency of the algorithm,the processor resource and the selection of the initial parameters.Finally,the experimental results show that the proposed I_GA algorithm,I_AFSA algorithm can effectively solve this problem and use less processor core resources,and also they are better than the non improved algorithm in the efficiency and accuracy of the search for the optimal solution.