An Improved Fault Tolerance Optimization Method Of Real-time Embedded System

Real-time embedded systems play a more and more important role in the economic,military,science,etc.At the same time,the requirement of the real-time and reliability gets higher.Fault tolerance is an important technology that supports reliability of system.This paper improves the real time and reliability of embedded systems as the main line.Consider it from mining advantages and disadvantages of fault-tolerant technology,the optimization problem on task mapping processor has been carried on the thorough research.For the hardware redundancy fault-tolerant technology tends to produce the defect of high design and high production costs,this paper puts forward an improved fault tolerance optimization method of real-time embedded system.The optimization method is based on checkpoint fault-tolerant technology,and it comprehensively analyzes system failure performance,hard real-time tasks' time constraints,and the utility function value of the soft real-time tasks.Based on the design of the application model and processor model,the improved optimization method calculates the failure probability of the system to ensure it within the maximum failure probability value,and calculates the finish time of hard tasks to determine the schedulability.System fault analysis on checkpoint rollback recovery,respectively,has carried on the simulation research with no fault tolerance analysis and that on re-execution in the existing literature.The schedulability analysis has carried on the comparative study with that on re-execution in the existing literature.The results verified the effectiveness of system fault analysis and the schedulability analysis on the checkpoint rollback recovery.For the performance of tabu search algorithm has certain limitation when avoiding local optimal solution in the existing optimization literature,an improved tabu search algorithm is given.The heuristic optimization algorithm obtains the best utility function value of soft tasks.There are two kinds of simple structure on neighboring nodes.This paper uses a greedy algorithm to obtain the initial solution,the rule of tabu follows the tabu of adjacent nodes through exchange,the pardon guideline follows that the maximum target value from these solutions is setted by the current best solution when a certain number of solutions are the tabu.The algorithm can obtain the final solution that has higher quality,faster convergence speed and higher calculation efficiency.It also can improve the robustness of optimization method.The simulation results verify the effectiveness of the proposed algorithm.On this basis,this paper designs the improved fault tolerance optimization method based on checkpoint technology.The fault tolerance optimization method can carry on the fault analysis and the schedulability analysis.That integrating the failure analysis and the schedulability analysis into the improved tabu search algorithm can quickly get the best utility function value.In order to evaluate the pros and cons of the fault tolerance optimization method comprehensively,the evaluation index of that is given,and this paper carrys on the experimental verification accordingly.The experimental results show that the improved fault tolerance optimization method has the validity.The above research work for fault tolerance optimization on real-time embedded system provides a new solution.