Study On Data Processing Techniques For Distributed RFID Systems

RFID technology has achieved rapid development recent years,and has been widely used in many industries.The distributed RFID technology that appeared recently has opened up a new technical space for RFID technology.The rapid development of RFID and the diversity of application scenarios put forward higher requirements for RFID data processing technology.How to clean and classify the data of RFID becomes an important part of the further development.To solve the problem of RFID data processing for RFID data cleansing and classification problems,algorithm and the corresponding system is designed to deal with large-scale RFID data efficiently and accurately.According to the new characteristics of distributed RFID data structure,the data processing system is designed to process the data sets,including the data's cleansing and classification.When the number of antennas is more than ten,the overall performance of the system is significantly improved compared with the traditional processing system.The system is designed with the structure of MVC,the coupling degree of each module is relatively low,which is convenient for the development and expansion of the function,and the stability of the system is increased.This paper improves the traditional SNM algorithm to solve the problem of cleansing RFID data,by analyzing the influence factors of SNM algorithm.Also the ideal condition of the experiment and the condition of poor reading performance are discussed,and the optimal window size of the two cases is obtained.As a result of this,the processing performance of the algorithm is improved by 10%to 15%in the corresponding region.In this paper,the matching algorithm is used to improve the matching process,and the performance of the algorithm is improved by about 35%compared with the automaton algorithm in solving the RFID data classification problem.The introduction of pruning algorithm for tree matching,with a certain probability threshold,the performance of the algorithm is further optimized.The function of each module was tested to verify the algorithm performance and effect of each model in this paper.Through the performance test,it is proved that the improved algorithm has significant effect in performance compared to the original algorithm,also the research value of RFID data processing is proved sufficiently.At the end of this paper,the RFID data processing technology is summarized and prospected,the possibility of further improvement using the system with machine learning is descuesed.It also makes clear to develop a more universal data processing system for defferent application scenarios to realize the complete data storage.