基于直方图的数据流降载策略研究
发布时间：2019-05-16 20:16
【摘要】：提出了基于直方图的降载策略,能很好地减缓在过载发生时系统性能的下降.降载的目标在于删除过载数据的同时尽可能地保持数据流的特征.为了处理大量被延迟的数据,构建了一种塔形矩阵的数据存储结构,利用其对过载数据分桶,每桶提取一个代表数据并删除该桶中其余数据,将每个桶的代表数据组成新的数据流参与查询操作.实验结果表明:这种降载方法能有效减少系统负担,生成的新数据流参与数据流查询后所得查询结果错误率较低,其性能优于其他已有算法.
[Abstract]:A load shedding strategy based on histogram is proposed, which can slow down the degradation of system performance when overload occurs. The goal of load shedding is to delete overloaded data while maintaining the characteristics of the data stream as much as possible. In order to deal with a large number of delayed data, a data storage structure of tower matrix is constructed, which is used to extract one representative data per barrel and delete the rest of the data in the bucket. The representative data of each bucket is formed into a new data stream to participate in the query operation. The experimental results show that the load shedding method can effectively reduce the burden of the system, and the error rate of the query results obtained by the generated new data stream participating in the data stream query is low, and its performance is better than that of other existing algorithms.
【作者单位】： 华中科技大学计算机科学与技术学院;湖北大学计算机与信息工程学院;
【基金】：国家自然科学基金资助项目(61173049)
【分类号】：TP311.13;TP333
，
本文编号：2478531
[Abstract]:A load shedding strategy based on histogram is proposed, which can slow down the degradation of system performance when overload occurs. The goal of load shedding is to delete overloaded data while maintaining the characteristics of the data stream as much as possible. In order to deal with a large number of delayed data, a data storage structure of tower matrix is constructed, which is used to extract one representative data per barrel and delete the rest of the data in the bucket. The representative data of each bucket is formed into a new data stream to participate in the query operation. The experimental results show that the load shedding method can effectively reduce the burden of the system, and the error rate of the query results obtained by the generated new data stream participating in the data stream query is low, and its performance is better than that of other existing algorithms.
【作者单位】： 华中科技大学计算机科学与技术学院;湖北大学计算机与信息工程学院;
【基金】：国家自然科学基金资助项目(61173049)
【分类号】：TP311.13;TP333
，
本文编号：2478531
本文链接：https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/2478531.html
