The Research On Complex Event Processing With Uncertainty For Lage-Scale Internet Of Things

In the 21 st century,with the rapid development of perceptual recognition technology,automatic information generation apparatus have been taken in a big way,such as intelligent identification terminal and sensor nodes.They accurately perceive,monitor and measure the physical world.On account of networking requirements of the physical world and expanding demands of the information world,the Internet of Things(IOT)as a new type of network emerges.The applications of the Internet of things almost cover a variety of industry.The events of the Internet of things have the features of heterogeneous,dispersive,massive and uncertain etc.At the same time,with the increasing of the scale of the global digital world,the era of big data is coming.Large-scale IOT is a combination of large data technology and Internet of things.As a key part of the IoT middleware,complex event processing technology can rapidly extract and identify meaningful information from the continuous data stream.The complex event processing for large-scale IOT environment faces many challenges.The first is how to efficiently handle the huge event flows.The other is how to accurately compute the uncertain results caused by uncertain event sources.In view of the challenges of complex event processing.This paper proposes to solve the uncertainty complex events using CEP2U(Complex Event Processing under Uncertainty)model.One this basis,we propose the method of distributed parallel processing based on rule partition.The main work of this paper includes:At first,research and analysis the problem of uncertainty in the application of Internet of things,CEP2 U model is proposed,considering the uncertainty of values and event occurs,using probability theory modles uncertainty problem.It uses concise and readable rules,and the system overhead introduced by uncertainty is relatively limited.Secondly,on the basis of CEP2 U model,to improve the performance further,we propose a distributed processing method based on rule partition.According to the delay condition between the processing nodes,it constructs the nodes into a shortest path tree.According to the actual situation of data flow and different queries,method patition query tasks into subtasks and deploy the subtasks one the corresponding nodes of the tree.This method realizes the true sense of the distributed parallel processing.