基于时间序列的感知QoS的云服务组合

QoS-aware cloud service composition based on time series

研究基于时间序列的感知QoS的云服务组合,将服务的QoS偏好随时间不断变化的过程纳入云服务组合的研究范围,将云服务组合建模成时间序列的相似度对比问题。分别用欧几里得距离和扩展Frobenius范数距离度量二维时间序列的相似度,继而用基于主成分分析的扩展Frobenius范数距离和欧几里得距离、Brute-Force等方法度量多维时间序列的相似度,通过实验对比验证扩展Frobenius范数距离度量相似度在时间和准确性上的优越性。

In this paper,we study QoS-aware cloud service composition based on time series,taking the sustaining change process of service QoS preference into research scope of cloud service composition,and transform the composition of cloud services into similarity comparison problems between time series.Euclidean distance and the extend Frobenius norm distance are used to measure similarity between two-dimensional time series respectively.In the next,the extend Frobenius norm distance or Euclidean distance based on principal component analysis and Brute-Force method are adopted to measure the similarity between multidimensional time series.Experiments show that the extend Frobenius norm distance has better performance on similarity measurement in time and accuracy.