Studies On Optimization Of Grid Resource Market Based On Optimal Competition Winner

Grid Computing is the most potential next generation computing platform to solve various kinds of large-scale computing. With its help,we can integrate all the resources that scattered different locations in the world,to achieve the access for all for resources.The key of Grid Computing is how to coordinate and schedule the resources effectively,to make the grid achieve the optimal performance.This paper is set in the Grid Computing,to research how to select to involve in grid computing resources and tasks using the ecomomy model. The optimal competition for winner problem in the combinatorial double auction of Econet model.Its problem is that both parties are required to choose the good buyers and sellers on the condition that resources supply not more than demand for,before they enter the auction process.This paper introduces a new integrative algorithm of particle swarm optimization and genetic algorithm to determine the optimal competition for winner in combinatorial double auction.And make the benefit of market maximization.This algorithm introduces the genetic algorithm into the particle swarm optimization algorithm,cross-reference the idea of cross,mutation and selection in the genetic algorithm.To get the new particle,by making corresponding operation on the particles in particle swarm, individual and global optimal particles.It easily overcome PSO into a local optimun value of this inherent flaw,and find the global optimum again in the serach space.Meanwhile,it introduces the conception of trust value into the improved particle swarm algorithm,to select buyers and sellers with a high trust value that can enter into the combinatorial double auction and market transaction,and to prevent malicious sellers and buyers to bid.In this way,it greatly gets rid of the possibility of making the whole network into disorder and paralytic,when implementing the tasks or resources with low trust value. In turn, can increase the success rate of grid resource scheduling to optimize the overall performance of grid system.Experiments simulate the achieve of combinatorial double auction,using GridSim simulation software.Experimental results show that compared with improved genetic algorithm and discrete PSO,this improved algorithm can make the system returns best when using it in the combinatorial double auction model.And ensure the participation of each auction get more benefit as muc as possible.