摘要
复杂事件处理是RFID数据管理的关键技术,由于受到内存的限制,海量实时的RFID原始流数据处理的中间结果部分只能存储在外存中,会产生内存瓶颈,严重限制了大规模RFID的部署。为此,提出了B+-树分时优化索引(B IOT)的复杂事件处理算法。在内存受限的情况下,将数据流按时序进行分割,且用B+-树进行区间分块索引,之后利用RFID数据流统计分布特性进行复杂事件查找与匹配,避免了频繁搜索外存,极大地降低了I/O开销并提高了吞吐量。此外,进行了相关的对比实验,验证了算法的有效性。
Complex event processing （CEP） is a key technology of RFID data management. But the massive stream of realtime data may cause memory bottlenecks, data placed in external storage and restrictions on large-scale deployment of RFID. So this paper proposed a BlOT （B Plus tree indexing optimization of time-slicing） algorithm. With memory constraint, according to timing sequence, the data stream would be divided into segments which would be indexed by B ＋ - tree, then to avoid frequently searching pre-event at external storage matched complex event with the guide of statistical law. This algorithm has greatly reduced the I/O overhead and increased throughput. In addition, correlated comparative experiments are carried out to verify the effectiveness of the algorithm.
出处
《计算机应用研究》
CSCD
北大核心
2009年第8期2864-2867,共4页
Application Research of Computers
基金
NSFC-JST重大国际(地区)合作项目(60720106001)
国家自然科学基金资助项目(60803043)
关键词
分布差
内存瓶颈
复杂事件处理
priority
memory bottleneck
complex event processing
