Affiliations:
Department of Computer Technology and Applications, State Key Laboratory of Plateau Ecology and Agriculture, Qinghai University, Xining, China
Correspondence:
[*]
Corresponding author. Xiaoying Wang, State Key Laboratory of Plateau Ecology and Agriculture, Department of Computer Technology and Applications, Qinghai University, Xining 810016, China. E-mail: wxy_cta@qhu.edu.cn.
Abstract: With the construction and application of large-scale datacenters, the issue of resource allocation in cloud computing becomes a serious concern. Although the current static allocation method can make applications get corresponding resources, there still exist some shortcomings such as resource surpluses or shortages. This kind of problem is more crucial in real-time requirements of mobile cloud computing service. Therefore, it is necessary to establish a forecasting model to predict the future resource demands, and then perform on-demand distribution, which can effectively reduce the unnecessary daily network management fees and address the issues mentioned above. This paper focuses on CPU resource forecasting, establishing three forecasting models including Markov chain, weighted Markov chain and stacking weighted Markov chain. By comparing and analyzing the experiment results, the most reasonable forecasting model is found and explained.
Keywords: Mobile cloud computing, resource allocation, forecasting model, markov
chain, CPU
