Research On The Detection Techniques About Containment Reationship Of Moving Objects In RFID Space

Recently, with the rapid development of networking technology, RFID as one of the key technologies, have attracted widespread attention to academia and industry at home and abroad. Location Service as one of the most valuable application of RFID technology has broad application prospects, but because of RFID itself is not a precise positioning equipment, RF interference, object movement, monitoring scenario and other reasons, generating large uncertainty spatial and temporal information stream. Meanwhile requires of location-based service in RFID space becoming much more complex and diverse than before, bring many new challenges to spatio temporal query. Researching on the relationship between moving objects, which based on uncertainty RFID data, is particular important for providing deep level location service.This thesis sets out to research the technologies for detecting containmentship between moving objects.Firstly, for the characteristic of RFID data, this thesis does pre-processing for the redundant portion of the RFID data efficiently, building a dynamic graph mode driving by data stream, which can reflect the information about monitoring objects in real time. This thesis uses the motion information from the graph model to analysis historical correlation of the objects, which provides a reliable basis for containment relationship reasoning subsequently.Secondly, considering possible relationship and space conatraints of monitored object in reality, this thesis builds possible relationship model, probabilistic rule model and number limited model assiting containment relationship detection. This thesis uses rule of possible relationship model and correlation information of the monitoring objects to calculate the probability of containment relationship. For the case of a fully loaded container, the strategy finding objects from container (UTOD) is adopted, which selects top-k objects with the maximum probability as the result, according to the capacity of the container. For the case of a container without fully loaded, the strategy finding container from object (DTOU) is adopted, which selects container with the maximum probability value and not reach its maximum capacity as the container of the object. In addition, adding mutually exclusive relationship detection in the above method, selecting the winner as the object contained in the current container. This thesis uses priori rule set to reduce the possible containment relationship between monitored objects, thereby improving the accuracy and efficiency for containment relationship reasoning.Thirdly, aiming at the problem of storage cost caused by the large-scale RFID data, this thesis uses containment relationship for data compression and storage. This method makes use of the characteristic that large amount of container and its corresponding contains have the same movement feature, using motion information of the container representing objects contained in it, so as to achieve the purpose of data compression. In addition, considering the feature that time information, location information appeared in large numbers in the data, this thesis uses complex coding method for secondary compress.This method makes the motion information of an item can be represented by a complex coding which is consisted of time sequence encoding and path encoding.Finally, after tremendous experiments and analysis on RFID dataset, the result shows: Compared to using the history information alone, the new model and the new algorithm in this thesis have a better performance at computation time and accuracy. As for the problem of data storage, the method of the data compression provided in this thesis, have a good performance on time costing and greatly reducing the data size under the condition of esuring the integrity of the data.In conclusion, this thesis studies on the containment relationship detection and data compression technologies for RFID data, and proposes new solutions. Theoretical analysis and experimental results show that, compared to the existing approach, our approach doesn’t need complex iterative probabilistic reasoning and has obvious advantages in performance and practicability.