The Research Of Technologies For Uncertainty Complex Event Processing

With the rapid development of Internet of Things(IoT),a large amount of data streams are being generated every day.Due to noise,network communication technology,sensor error,natural factors and other reasons,the raw event might be uncertain.However,the existing technology of data processing is not effective enough in dealing with these uncertain original events.More than that,on account of the limitation of the current knowledge,the uncertainty caused by setting event recognition rules according to experience may also lead to wrong results.The technology of event processing of uncertain complex events is confronted with many challenges.The first is how to recognize the complex events among vast amount of uncertain event streams in real-time.The second is how to calculate the probability of complex events among massive relevant uncertain events.The third is how to apply complex event processing under the circumstance of the rules being uncertain.In this paper,we take a deep analysis of the current researching conditions and challenges of complex event processing in Internet of Things.In order to improve the efficiency of dealing with uncertain complex events,we propose a method based on mNFA and probability to address the issue,and try to realize more improvement.Besides,considering the uncertainty of the rules,we bring forward a new method based on Bayesian networks to deal with uncertain complex events.The paper is mainly elaborated from the following two aspects:(1)In the research and analysis of realistic large-scale IoT,due to noise,network communication technology,sensor error,natural factors and other reasons,the raw events may be produced with uncertainty,whose types will be categorized in the paper.Considering the uncertainty of existence and attributes of events,on the basis of mNFA and probability theory,we propose a method to deal with the uncertain event streams.It can not only process massive uncertain event streams efficiently,but also provide new perspective in query processing of probabilistic event streams.Furthermore,we try to optimize this method by applying dynamic probability calculation algorithm and filtering irrelevant events in advance.The result of the experiment shows that our method is more efficient and practical than the traditional ones.(2)Complex events are generated from detecting the event rules of massive original event streams.These high-level complex events are more significant for users.The rules are usually set by the experts in this field after empirical analysis.However,the limitation and incompleteness of current knowledge and experience lead to the uncertainty of rules.So,we bring forward a new method based on Bayesian networks to deal with uncertain complex events in this paper.The event rules are translated into Bayesian network node,and URCEP(Uncertainty Rule Complex Event Processing)model is used to calculate complex events.At the same time,users can add custom rule nodes to simulate the effects of uncertain factors that might be ignored.The experiment result show that the method is very effective in dealing with uncertain problem.