一种全局QoS感知服务组合的权重设计方法

A synthetic weight for global QoS-aware service composition

基于全局QoS感知的服务组合可以通过加权聚合的方式构建为一个单目标优化问题,此时需要一个合适的方法去估算权重,为此提出了一种基于非均匀空间理想加权距离的权重合成方法.该方法考虑用户的全局定性需求,首先运用G1法生成主观权重,然后使用熵权法计算各组合点上QoS属性的内在客观权重;最后给出非均匀空间距离计算方法,结合逼近理想法设计合成权重.构建于QWS数据集上的实验,阐述了合成权重的计算过程和评估了相关参数对权重的影响.结果表明,合成权重能够综合反映用户的偏好和QoS属性的内在特征.

The global QoS-aware service composition problems can be addressed by modeling them as single objective programming problem through utilizing weighted aggregation method.However there is no proper solution for estimating the weights.So the paper proposes a synthetic approach based on weighted distance in nonuniform space.Firstly,the paper defines a mixed integer programming to describe the issue,which considers the coexistence of global quantitative and qualitative constraints.Secondly,the paper utilizes G1 method to yield the subjective weights.Then the paper exploits entropy-weight method to calculate a series of objective weights.And then the paper provides the calculate method of weighted distance and utilizes TOPSIS（the technique for order preference by similarity to ideal solution） to design the synthetic weights.Finally,a set of experiments working on QWS real dataset is used to demonstrate the calculation of weights and evaluate the influence of relevant parameters.The results indicate that the synthetic weights can reflect user ＇ s preferences and intrinsic preferences of QoS attributes in various composition points.