摘要
随着RFID和传感器等数据采集设备的广泛使用及物联网的发展,产生了大量的事件类型的数据,原始的事件数据必须经过复杂事件处理(Complex Event Processing,CEP),才能变成具有丰富语意并对用户有价值的信息,复杂事件处理作为物联网智能处理层的重要组成部分,越来越受到重视.在实际应用中,许多事件流具有长过程的特点,要求相应的复杂事件处理需设置大时间窗口,相对于有限的内存,复杂事件处理面临新的挑战.现有的复杂事件处理均局限于内存进行,均未涉及外存的事件存储和检测.因此,现有的模型和系统均不能用于长过程复杂事件处理.为此,本文提出基于时间片划分的HTF(Hash structure by object ID in memory and Timeslice File in disk)事件实例存储策略和基于实例映射表的大时间窗口复杂事件检测方法,形成了面向长过程的复杂事件处理模型LPCEP(Complex Event Processing for Long Process).相关实验验证了模型用于长过程复杂事件处理的有效性和高效性.
With the wide use of sensors and RFID devices and the development of the Internet of things ,large quantifies of event data are generated. Original event data must be processed by CEP in order to become valuable information with rich semantics. CEP has be- come research hotspot as an important part of the intelligent processing layer of the Intemet of things. In practical applications, large numbers of event streams have the characteristics of long process, which demands the corresponding CEP to set large time windows and makes CEP face new challenges in the limited main memory. In all the current CEP methods ,complex events are processed in the main memory without events stored and detected in the auxiliary storage. So,the current methods can not be applied in CEP for long process. Therefore, the paper proposes time slice based event instance storage strategy and instance mapping table based complex event detection method for large time windows, forming the CEP model for long process. The correlative experiments verify the effectiveness and efficiency of the model.
出处
《小型微型计算机系统》
CSCD
北大核心
2015年第1期71-76,共6页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(61272177)资助
关键词
长过程复杂事件处理
HTF事件存储策略
实例映射表
候选实例转换
事件实例匹配
complex event processing for long process
HTF event storage strategy
instance mapping table
candidate instance trans- formation
event instance matching
