一种改进的分布式服务QoS预测方法

An Improved Method of Distributed Services QoS Prediction

随着Web服务数量的增长,选择高质量的服务成为了用户的重要需求。服务的QoS预测为用户提供了了解服务质量的手段。在目前主流的QoS预测方法中,基于网络定位的方法具有不依赖历史数据,更灵活的优点,但同时也有准确率较低,不易于扩展等方面的问题。本文针对这些问题提出了一种改进的基于分布式网络坐标系统的服务QoS预测方法,实验结果表明,本文方法不仅提高了预测的准确率,保证了预测的稳定性,还提高了定位计算的收敛速度。

With the increase number of Web services, choosing the service with high quality becomes users＇ important requirement. Service QoS prediction provides effective way for users to acquire the quality of services. In comparison with other methods, the network coordinate - based methods do not need the historical data and are more flexible. But they often generate relatively low prediction accu- racies and have scalability problem. This paper proposes an improved method to deal with the problems discussed above. The experi- mental results show that our method can effectively improve the prediction accuracy, ensure the stability of prediction and speed up the convergence of allocation computing.