Research On RFID Tag Classification And Sorting Algorithm

With the emerging development of the Internet of Things technology,the trend of items being connected to the network has become more and more obvious.To further reduce the labor cost in the storage environment,people began to use the Radio Frequency Identification(RFID)techonology in the warehousing environment.The RFID system completes the inbound and outbound configuration of the goods and the destination matching of the luggage by monitoring the RFID tags on the conveyor belt.How to quickly distinguish the order of these group reading tags on the conveyor belt has become a new topic.In view of the above situation,this paper proposes an Abnormal Data Detection and Error Tolerant algorithm(ADD-ET)based on the warehouse-conveyor scenario,which effectively improves the sorting accuracy of the RFID tags on the conveyor belt.In addition,for the off-site tags existing in the sorting area of the conveyor belt,an RFID tag classification method based on kNN machine learning algorithm is proposed,which can effectively provide an accuracy of nearly 99% to identify whether the tags are on the conveyor belt or not in the warehouse environment.The algorithm provides a guarantee for the tag sorting algorithm on the conveyor belt.The main contributions of the paper are as follows:(1)Based on the warehouse-conveyor scenario,a set of RFID system simulation platform has been built.The platform first calculates the theoretical values of all multidimensional information of the RFID tags in this scenario.The multidimensional measurements are Recived Signal Strength Indicator(RSSI),phase information,and Doppler Frequency Shift(DOPPLER).Combining the theoretical value with the Response Reception Ratio principle(RRR)of the tags,and adding the reading ability of the RFID reader into consideration,a simulation test platform which is closer to the RFID real system is realized.(2)An ADD-ET algorithm that uses both RSSI and DOPPLER information is proposed.This algorithm performs reliability detection on RSSI measurement values.When RSSI information is reliable,RSSI measurement values will be used for direct sorting.When the RSSI information is unreliable,the error tolerant algorithm will be used as an auxiliary to analyze the DOPPLER measurements and give the relative position ranking results.This method shows a very impressive sorting effect in the real test environment.(3)Combining the SARFID method of tag classification,an RFID tag classification method based on kNN machine learning algorithm is proposed.This method uses the RSSI and DOPPLER measurements returned by the RFID tags to calculate their statistical characteristics and matching characteristics.Combined with the kNN machine learning algorithm,in the simulation test experiment,the classification accuracy rate of "whether the tags are on the conveyor belt or not" is 98.8%,and the correct rate of the static and dynamic classification of the RFID tag is as high as 99.99%.