摘要
服务质量(quality of service,QoS)作为Web服务非功能属性的代表,已被广泛作为重要的服务选择依据。现有QoS预测方法存在着难以兼顾运行效率与预测准确度的问题,且普遍忽略了服务器端的QoS预测。针对该问题,提出一种适用于服务器端环境的轻量级QoS预测机制(LPM)。LPM基于卡尔曼滤波算法构建QoS状态转换模型来实现QoS预测,并借助预测准确度优化预测周期。实验结果显示,在存在显著量测噪声的应用环境中,LPM的预测准确度明显优于常规方法。LPM的QoS预测结果可为用户选择Web服务提供首要的客观依据。
QoS as the representative of the Web service non-functional properties has been widely used as an important basis in service selection. However, the existing QoS prediction methods can hardly guarantee the efficiency and prediction accuracy at the same time, and generally ignore the server-side QoS prediction. This paper proposed a lightweight prediction mechanism （LPM） for server-side environment. Based on Kalman filtering, LPM built QoS state transition model to periodically generate QoS predicted value, and measured prediction accuracy to optimize the prediction period. Simulations show that LPM has bet- ter prediction accuracy compared to conventional method in the environment with significant measurement noise. The output of the QoS predicted value can provide primary objective reference basis for users to select Web services.
出处
《计算机应用研究》
CSCD
北大核心
2016年第11期3311-3314,3333,共5页
Application Research of Computers
基金
国家"863"计划资助项目(2012AA012704)
信息保障技术重点实验室开放基金课题(KJ-13-110)
