A Cloud Resource Allocation Mechanism Based on Mean-Variance Optimization and Double Multi-Attribution Auction

Abstract

Abstract. As a new kind of commercial model, cloud computing can integrate various kinds of resources in the network. Resource providers offer these resources to users in the form of service and receive corresponding profits. To make more rational use of the cloud resources, an effective mechanism is necessary for allocating the resources. In this paper, the price attribution and non-price attributions of both traders are analyzed. The support vector machine algorithm is utilized to predict the price, further determining the quote and bid. Then, the BP neural network algorithm is used to transfer the non-price attributions to the quality index. Finally, to maximize the total satisfaction of resource providers and resource consumers, the mean-variance optimization algorithm is adopted to obtain the optimized cloud resource allocation scheme. Simulation results have shown that the proposed mechanism is feasible and effective.