摘要
目前基于QoS的服务选择方法中通常假设QoS属性值都是真实可信的,但该条件在实际应用中往往很难满足。为此,提出了一种保证QoS属性值可信性的方法。该方法将各属性视为未知参数,将服务提供者的历史运行数据以及服务使用者的反馈视为对未知参数的样本观测值,通过样本观测值和置信区间去检验属性值的可信性并进行修正。仿真实验表明,该方法的性能大幅优于现有方法,并特别适用于属性值偏差比较大的情况,平均均方误差能够下降90%以上。
Most current QoS-based service selection approaches always assume QoS attributes values are trustworthy,which is actually impossible in reality. For this problem,this paper proposed a method assuring the trustworthiness of QoS attributes, which considered QoS attributes as unknown parameters. And regarded the performance history from service providers and feedback from service users as sample observations. Through sample observations and confidence interval,checked and modified the values of QoS attributes. The simulation results demonstrate that compared to current methods,the performance of this method is much more effective,especially in environments of large deviation,which can reduce mean square error more than ninety percent.
出处
《计算机应用研究》
CSCD
北大核心
2010年第10期3658-3661,共4页
Application Research of Computers
基金
国家"863"计划资助项目(2008AA01A317)
关键词
服务质量
可信
服务选择
置信区间
QoS
trustworthiness
service selection
confidence interval
