Research On Complex Event Processing Techniques For Temporal Uncertainty Model

With the rapid development of the RFID and sensor network,the complex event processing(CEP)for RFID data has attracted wide attention in many fields,such as the event alerting and event monitoring.Existing work on CEP,however,assumes that the occurrence time of each atomic event is known precisely and the events from various sources can be merged into a single stream with a total or partial order.We observe that in real-world applications event occurrence times are often unknown or imprecise because raw RFID data information is known to be lossy and even misleading,that event occurrence times are subject to granularity mismatch,that events collected from a distributed system are subject to the clock synchronization problem.So we assign a time interval to each event to represent all of its possible occurrence times.In this thesis,we address pattern query evaluation in streams with uncertain occurrence times of events.Firstly,for character which the uncertain time event sequence meets,this thesis gives a temporal uncertainty model and formal pattern evaluation semantics under the model.Then we give some key concepts based on the model,such as uncertain events,probability events,signature sequences and so on.This model laid a foundation for solution of the problem.Secondly,this thesis gives the formal definition of nested CEP query over streams with uncertain timestamps.Then for this problem,this thesis puts forward the methods based on possible worlds and iteration to handle nested query on uncertain event.An iteration-based method improves the basic method based on the possible world model.Furthermore,two optimization methods,namely pruning based on the length of sub-queries and caching based on the sharing sub-expressions,are further explored.In addition,an efficient pruning technique to evaluate the result probability is studied based on the caching method.Finally,the experiments have verified the efficiency of the proposed methods when processing nested query over streams with uncertain timestamps.Finally,this thesis gives the formal definition of uncertain event query with event constraints.Then for this problem,this thesis puts forward the methods based on possible worlds first(possibleF_QP)and event constraints first(constraintsF_QP)to handle this kind of query.Considering that possibleF_QP and constraintsF_QP are affected by many factors,we evaluate them on synthetic data sets.Experiment results show that the performance of constraintsF_QP is better than that of possibleF_QP.In conclusion,this thesis studies two key issues of CEP query over streams with uncertain timestamps:nested query and event constraints,and puts forward efficient solutions to solve them.Extensive experiments validate the efficiency and accuracy of our methods.