基于层次自动机模型的复杂事件层次实现研究

复杂事件处理技术从数据流中提取满足特定模式的事件序列,具有实时、海量、智能的特点,近年来引起了学术界和商业界的广泛关注。但是,之前的工作侧重于对单层复杂事件检测的研究。事实上,由于业务系统对信息有不同层次的需求,需要对事件进行分层处理,单层复杂事件检测并不能充分支持事件分层的需求。针对这种情况,在事件层次概念以及传统NFA模型的基础上,定义了分层复杂事件检测模型层次自动机NHA,基于NHA模型设计了更为直观高效的EH-Tree结构,并给出了分层复杂事件检测HCED算法和代价模型。最后以吞吐量和内存占用为指标,进行了大量的实验,对比并分析了HCED算法与传统基于NFA模型的SASE算法的时间性能和空间性能。实验结果表明,HCED算法能有效且高效地实现分层复杂事件检测,填补了CEP不支持分层复杂事件检测的空白,为下一步研究提供了基础。

Study of Complex Event Hierarchies Realization Based on Hierarchical Automata Model

Complex event processing technology extracts event sequences which satisfy the specific patterns (called complex events) from continuous event streams.It processes a large sum of data in real time intelligently and has attracted the attention from both academia and industry in recent years.However,the state-of-the-art researches focus on single-layer complex event detection.In fact,because business systems have different hierarchical needs of information,events in system should be processed according to their hierarchies.Single-layer complex event detection cannot afford to meet the needs of hierarchical information in different managerial hierarchies.To deal with such problem,in this work,a hierarchical complex event detection model,hierarchical automata,was defined based on the concept of event hierarchies and traditional nondeterministic finite automaton (NFA) model.Then we designed a more intuitive and effective EH-Tree structure and proposed hierarchical complex event detection (HCED) algorithm and its cost model.Finally,taking throughput and memory consumption as indexes,massive experiments were performed to compare the spatial and temporal performance of HCED algorithm based on EH-Tree and SASE algorithm based on NFA.The results testify that our HCED algorithm can perform hierarchical complex event detection effectively and efficiently,which fills in the blanks of hierarchical complex event detection and indicates that our work can be a basis for further research of complex event processing.