Task Scheduling Problem Based On The Availability Of Research And Implementation Of The Particle Swarm Algorithm

With the development of computer technology, distributed heterogeneous systemsis increasingly used in processing of solving solution with large and complexproblems. It is the network of computers, workstations and other computing resourcesinto together, achieveing the resource sharing, which can make the system reach thehigh function through efficient resource management and technology of adjusting task.However, the resources are limited, in order to allow limited resources to meet theneeds best, there must be a reasonable use of resources to reduce system overhead asmuch as possible. Therefore, the system task scheduling technique is effective up toone of the key technologies. In general, the task scheduling problem is NP completeproblem, it is difficult to explore the exact solution within a reasonable time, so farmainly with the Opportunistic Load Balancing (OLB), the Minimum Completion Time(MCT), Simulated Annealing (SA), Tabu Search (Tabu), Genetic Algorithm (GA),Neural Networks, Min-min algorithm (Min-min), the max-min algorithm (Max-min)and other algorithms to seek the approximate optimal solution of the problem.PSO as a heuristic optimizing algorithms is obtained from birds’ flock foragingbehavior,and based on artificial life and evolutionary computation theory whileretaining and using the position and velocity information. In most cases, it can quicklyconverge to the optimal value. Since PSO is very intuitive thinking and realize it isalso simple and has high efficiency in the implementation, it has been as anindependent branch, compared with the genetic algorithms, evolutionary programmingand other evolutionary algorithms. Availability is a basic distributed heterogeneoussystem performance parameters. It refers to the node computing time effectivelythroughout their share of run time. In this paper, the design of the existing frameworkof PSO, the combination of genetic algorithm in the roulette wheel selection toimprove particle swarm algorithm, but also on the particle swarm algorithm in theencoding, fitness function to study, to form the whole idea of the algorithm to taskscheduling problem of the availability issues in depth. First, we assume that tasks aremutually independent, the availability of different nodes with different computingpower, and different tasks in different nodes of the availability requirements are alsodifferent. And then propose a PSO-based task scheduling algorithm for distributedheterogeneous systems to efficiently improve the usability. Finally, by using experimental simulation tools SimGrid, do the simulations test between the algorithmand some conventional algorithms, and compare the algorithm with the traditionalmethod of experimental data, It is prove that the thesis fefered in article can do a goodeffect, which is the scheduling algorithm can achieve a better experimental results andcompared to the traditional algorithm, the improved availability of the results would bemore pronounced.