复杂事件模式检测与CEP测试数据生成算法研究

发布时间：2018-04-05 17:26

本文选题：复杂事件处理　切入点：事件模式检测　出处：《计算机学报》2017年01期

【摘要】：随着信息技术的快速发展和广泛应用,大数据正以不可阻挡的气势向我们走来.大数据源于信息技术,同时又向信息技术提出挑战.如何征服大数据给信息技术处理能力上带来的挑战是一个广泛关注的课题.由于采用面向流式数据的处理策略,复杂事件处理(Complex Event Processing,CEP)技术被认为是一种有希望征服大数据挑战的技术之一.然而,当事件流成为大数据时,目前的CEP模型、事件模式检测和CEP系统测试数据生成的方法和技术都存在很多不足.文章针对这一问题讨论了复杂事件建模、模式检测与测试数据的生成方法.提出一种CEP代数模型,用多种事件算子来表达事件之间的关系和事件流模型.在这里,一个创新的思想是把CEP代数模型表达式解释成算术文法产生式,从而可以用词法分析技术解决复杂事件模式检测问题.为了有效地测试复杂事件模式检测的算法,基于CEP代数模型,提出了一个用于支持上述CEP模式检测的大数据事件集的生成算法.并由此生成了不同量级的测试数据,测试了一个RFID物联网中CEP模式检测引擎.实验结果表明了提出的事件模式检测算法和CEP系统测试大数据事件集生成算法的有效性.
[Abstract]:With the rapid development and wide application of information technology, big data is coming to us with irresistible momentum.Big data originates from the information technology, at the same time brings forward the challenge to the information technology at the same time.How to conquer the challenge that big data brings to IT processing ability is a widely concerned topic.Because of the flow oriented data processing strategy, complex Event processing technique is considered to be one of the promising techniques to overcome big data's challenge.However, when the event flow becomes big data, the current methods and techniques of CEP model, event pattern detection and CEP system test data generation have many shortcomings.In this paper, the methods of complex event modeling, pattern detection and test data generation are discussed.A CEP algebraic model is proposed to express the relationship between events and the event flow model by a variety of event operators.An innovative idea here is to interpret CEP algebraic model expressions as arithmetic grammars, so that lexical analysis can be used to solve the problem of complex event pattern detection.In order to effectively test the algorithm of complex event pattern detection, based on the CEP algebraic model, an algorithm for generating big data event set which supports the above CEP pattern detection is proposed.The test data of different orders are generated and a CEP mode detection engine in the RFID Internet of things is tested.Experimental results show the effectiveness of the proposed event pattern detection algorithm and the CEP system test big data event set generation algorithm.
【作者单位】：北方工业大学计算机学院;
【基金】：国家自然科学基金(61370051)资助~~
【分类号】：TP311.13

【相似文献】