A Dynamic Price Model with Demand Prediction and Task Classification in Grid

In the market-oriented grid systems, price management is one of the significant parts, which is used to balance the resource load and adjust the profits of resource providers. This paper proposes a dynamic price model based on demand prediction and task classification. We apply Markov chain to predict the future demands of grid resources and propose a resource price-adjusting mechanism based on the future demands, which takes into the interdependence of price and demand into consideration. Therefore, the mechanism can balance resource loads and guarantee the profits of resource providers at the same time. In addition, we classified the tasks into two categories: exclusive tasks and shared tasks. According to the differences between them, we provide two different task-pricing strategies. Based on these, we introduce a novel task-pricing mechanism, which takes serving time, the workload and the type of the task into consideration at the same time. Compared with time-based pricing model, it is more reasonable and more feasible. Simulation results show the applicability and effectiveness of our work.