面向RFID数据处理的复杂事件模式匹配方法

Pattern Matching Method of Complex Event for RFID Data Processing

RFID数据具有不确定性,复杂事件处理技术将RFID数据看作不同类型的事件,从事件流中检测符合特定匹配模式的复杂事件。概率事件流分为多项概率事件流和单项概率事件流;针对多项概率事件流,提出NFA-MMG模式匹配方法,亦即使用多个有向无环图结合自动机实现模式匹配。针对单项概率事件流,提出NFA-Tree模式匹配方法,亦即使用匹配树结合自动机实现模式匹配;并提出改进的NFA-Tree方法,即基于概率阈值进行过滤,提高结果过滤效率。实验结果验证了上述模式匹配方法的性能优势。

RFID data is generally uncertain.Complex event processing（CEP） treats the data as different types of events,queries sequence of events in which match specific patterns of sequence are defined by high-level application from the event stream.Event stream is divided into multiple alternative event stream and single alternatives event stream.NFA-MMG pattern matching method for multiple alternatives event stream was proposed.The method uses combination of directed acyclic graph and automatic machines to achieve complex event pattern matching on the uncertain data.NFA-Tree pattern matching method for single alternatives event stream with the use of matching tree and automatic machines on uncertain data was proposed.The NFA-Tree algorithm was improved by pruning the matching tree to improve the efficiency of query optimization,which filters the results of the match situation based on probability threshold.The complex event processing system prototype uncertain data was developmented to realize the above algorithm,and the experiment examines the validation and performance of the algorithms.