基于基区间的随机滑动窗口聚集

提出了一个基于基区间的实时随机滑动窗口聚集算法。首先，按照规则将窗口中的数据项划分成一系列基区间，然后分别对这些基区间进行聚集计算，窗口中数据项的聚集等于这些基区间聚集和。窗口滑动后，窗口中数据项的聚集可以部分地利用上一次窗口聚集的结果。模拟实验表明，与对窗口中的数据整体进行聚集相比，基于基区间的聚集算法可以有效地降低窗口聚集的时间，提高数据流处理的实时性。

Random sliding window aggregates based on base intervals

An algorithm for random slidden window aggregate based on base intervals was proposed. According to rules, the data set among sliding window were partitioned into a serial of sets called base intervals, then the base intervals were aggregated separately. The aggregate result on whole window was equal to the sum of base intervals. After window had been slidden, previous partial results could be used to compute aggregate on data sets belonging to current window. Compared with aggregate on whole data set within window, the simulative experimental results Show that the approach based on base intervals can reduce efficiently time and improve the real time performance of data stream processing.