Research On The Probability Calculation Method Of Region-based Non-deterministic RFID Event

In recent years, radio frequency identification (RFID) technology has becomethe focus of research attention. RFID data management system can be quick and easyto access to the object information through the readers and the tags. In manyapplications, we use RFID technology for object locating, tracking, etc.Non-deterministic is a significant feature of the RFID data, which runs throughthe RFID application data from creation to the demise of the entire life cycle. In RFIDapplications, manifestations of non-deterministic events are random events.Therefore,we usually use probability values to represent the non-deterministic. Thefocus of this paper is how the RFID raw data stream probability fitting event ofnon-deterministic and how to extract from the complex composite events. At the sametime，to calculate the probability of the basic events and composite events.The paper presents a new basic event probability calculation method，in whichMarkov chain method is applied. In this paper, the concept of Region was advancedaccording to the reader coverage. And the entire application space is discreet intomultiple locations. Then it introduces the event probability calculation in singleregion. Proposed location probability table and transition probability table to describethe probability of the location and status of the label object transfer transitionprobability, by which the calculation method is given for different locations.This paper, starting from non-deterministic the RFID event model’s specificlocation probability calculation method, analyzes the regional relations in the case ofmultiple readers. In complex environments，it provides an effective method tocalculate the probability of simple events and composite events under multi-region，which is including that the location is adjacent，separate and tangent. It proposes thecomposite event detection methods in the complex environment of the single-regionand multi-region. And the event semantics can be extracted by this method in acomplex environment.In this complex environment, event recall/accuracy validation experimentswere carried out for MarkovTech-method. Beside, in the memory utilization and overall response time, as well as the results of the calculation of the probability ofextracting valid events, MarkovTech-method compared PEEX and particle filtermethod prove the effectiveness and give the corresponding analysis of theenvironment results.