Interval-based Rfid Complex Event Processing

RFID(Radio Frenquency Identification) is an automatic technology to collect data and identify objects fleetly. It has been widely deployed in various applications, such as manufacture, logistics, transportation, medical infrastructure, assets management and so on.As the RFID technology being more and more popularized, the enterprises have already deployed massive RFID devices. However, a problem comes along. With the RFID technology being gradually adopted, it generates massive RFID data, the data is real-time, large volumn and contains abundant semantics. So how to collect and manage this raw data is an emergent issue. What's more, this kind of raw data means nothing to the end users. Beacause the raw RFID data is no more than a RFID identifer, an EPC code and the time it is generated at. It is low level semantic. For most of the end users, the data they care about is more high level semantic which can directly indicate how the business is running. RFID Complex Event Processing is a technology to solve this problem, that is to extract valuable, meaningful information from the raw RFID data.In this paper, we analyze the status quo of the RFID Complex Event Processing, and propose a interval-based method for RFID Complex Event Processing. First of all, after analyzing the complicate event relations under the interval-based event model, we propose a complex event model which can allow user to define the complex event, this event model unifies the data for the processing which will be done at next step. Then, according to the character of the RFID Complex Event Processing, we extend the traditional Petri-Net and propose a Petri-Net based event detecting model. Finally, a series of optimization strategies which are business related are propsed to reduce the memory cost during the processing. Our main contribution are as follows:First, after analyzing the possible relations between any two events under a Interval-based event model, we propose a interval-based event model. With this model, the complicated event relations between RFID events can be easily expressed. And user can use this model to define the complex events they need. At the same time, This model unifies the data for the following processing.Second, because the relation between RFID event is rather complicated, we extend the Petri net which has a good cucurrency to defect RFID event. The extension includes: colored token, body function and guard fuction. Third, after analyzing the characters of RFID Complex Event in practical applications, this paper propose a series of optimization strategies which can reduce the memory cost during the processing. Also, these strategies gives a good example on how to improve RFID Complex Event Processing according to the practical application.Finally, based on the contribution mentioned before, we design and implement a RFID Event Processing system. Wth this system, user can define, detect and manage RFID complex events. What's more, the businese steps can be welly monitored with this system.