复杂事件模式检测与CEP测试数据生成算法研究

Study on Algorithm of Checking Event Patten and Generating Test Data for Complex Event Processing

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.

Big Data is coming irresistibly with fast development and wide application of Information technology.The Big Data is created by IT and it challenges IT＇s technology contrarily.How to conquer the challenges is wildly focused problem.The Complex Event Processing（CEP in short）sounds to be one of promising technologies because its technical strategy on event flow analysis.However,there is much insufficient in current CEP model,event pattern checking and CEP engine test because the event flow has become bigger.In this paper,a method for modeling CEP,a model based on algorithm for checking event pattern and for generating test data of CEP engine are discussed.This model is proposed for describing relationship between complex events,and it defines number event operators.A novel idea is that the algebra model of CEP can be explained as event grammar products.Therefore,the technology of lexical analysis can be employed for checking event pattern.In order to test the algorithm of checking event pattern,a kind of CEP engine is proposed for sowing different probably event patterns in big test data with respect to specific probability distributions.Various volumes of Big Data are generated using proposed algorithm and an event pattern check program of RFID system is tested,the empirical results strong support the feasibility of proposed algorithm.