QoS Constrained Grid Workflow Scheduling Optimization Based on a Novel PSO Algorithm
Qian Tao,Huiyou Chang,Yang Yi,Chunqin Gu,Yang Yu
DOI: https://doi.org/10.1109/gcc.2009.39
2009-01-01
Abstract:Currently, computational grid, service grid and data grid are becoming richer and more complex. Grid infrastructure composed of heterogeneous resource is widely distributed, workflow scheduling problem in grid environment described by directed acrylic graph (DAG) becomes an important and difficult problem. Meanwhile, the research on workflow scheduling problem in grid environment mainly focuses on the time and cost constrained optimization, but the key problems about stability, flexibility, security and load balancing arenpsilat considered adequately. Aiming at these problems, we redefine parameters of quality of service (QoS) and the model of grid workflow scheduling, and put forward a rotary hybrid discrete particle swarm optimization (RHCurrently, computational grid, service grid and data grid are becoming richer and more complex. Grid infrastructure composed of heterogeneous resource is widely distributed, workflow scheduling problem in grid environment described by directed acrylic graph (DAG) becomes an important and difficult problem. Meanwhile, the research on workflow scheduling problem in grid environment mainly focuses on the time and cost constrained optimization, but the key problems about stability, flexibility, security and load balancing arenpsilat considered adequately. Aiming at these problems, we redefine parameters of quality of service (QoS) and the model of grid workflow scheduling, and put forward a rotary hybrid discrete particle swarm optimization (RHDPSO) algorithm, in which double extremums are disturbed by the method of random time sequence based on rotation discretization, to overcome premature convergence and local optimum. The simulation results show that the RHDPSO algorithm has fast convergence, high precision and strong robustness, and can effectively restrain premature convergence, compared with DPSO. The performance of our algorithm is very promising, scheduling, and put forward a rotary hybrid discrete particle swarm optimization- (RHDPSO) algorithm, in which double extremums are disturbed by the method of random time sequence based on rotation discretization, to overcome premature convergence and local optimum. The simulation results show that the RHDPSO algorithm has fast convergence, high precision and strong robustness, and can effectively restrain premature convergence, compared with DPSO. The performance of our algorithm is very promising.DPSO) algorithm, in which double extremums are disturbed by the method of random time sequence based on rotation discretization, to overcome premature convergence and local optimum. The simulation results show that the RHDPSO algorithm has fast convergence, high precision and strong robustness, and can effectively restrain premature convergence, compared with DPSO. The performance of our algorithm is very promising, scheduling, and put forward a rotary hybrid discrete particle swarm optimization (RHDPSO) algorithm, in which double extremums are disturbed by the method of random time sequence based on rotation discretization, to overcome premature convergence and local optimum. The simulation results show that the RHDPSO algorithm has fast convergence, high precision and strong robustness, and can effectively restrain premature convergence, compared with DPSO. The performance of our algorithm is very promising.