Research On Techniques Of Large-scale RFID Tag Data Collection

Thanks to the national strategy of "Made in China 2025",the manufacturing industry has set off a trend of digitalization and intelligent upgrading.Under this trend,radio frequency identification(RFID)technology,which is the core of the Internet of Things perception lay-er,has attracted wide attention and development.Compared with today 's mainstream barcode technology,RFID provides faster identification speed,longer identification distance and non-line-of-sight identification ability.These advantages makes RFID technology has been used by more and more productions and retailers to track materials and commodities in real time.With the extensive deployment of RFID technology,the RFID system scales up day by day,which brings great challenges to data collection in RFID systems.A large number of tags collided with each other,which leads to the inefficiency of data collection.To solve this problem,this thesis focuses on large-scale RFID tag data collection technique.On the one hand,it aims to coordinate the tag transmissions,so as to reduce the signal conflict between tags and improve channel utilization.On the other hand,it aims to improve the channel utilization and achieve on-demand data collection by deactivating non-target tags,so as to speed up the information collection process through reducing the size of RFID tags.The main contributions of this thesis are summarized as follows:For the problem of dynamic tag identification,this thesis studies how to avoid redundant transmission from identified tags.First,this thesis proposes to efficiently deactivate identified tags for avoiding redundant tag transmission,so as to reduce the signal conflicts between tags.Secondly,this thesis proposes to fast detect new tags and missing tags with a novel indicator vector,so as to speed up the tag identification process.By deactivating identified tags and iden-tifying dynamic tags,the conflicts between tags decreases significantly,which greatly improves the time efficiency of tag identification.For the problem of tag data polling problem,this thesis studies how to reduce the broadcast overhead caused by polling operation.First,this thesis proposes to build a one-to-one mapping between target tags and slots in the time,which ensures that each target tag can be assigned to an exclusive slot,so as to avoid signal conflicts between target tags.Secondly,this thesis proposes a signature-based filtering scheme to deactivate non-target tags,so as to avoid other tags interfering with the target tag data collection.By assigning target tags the corresponding slots and filtering other non-target tags based on tag signature,the polling overhead significantly reduces,which enables readers to collect information from specific target tags more quickly.For the problem of multi-category tag management,this thesis studies the implementations and applications differential tag sampling.First,this thesis proposes to assign different sampling probabilities to various types of tags by using standardized C1G2 commands,so that system managers can realize differential tag sampling according to the importance of the tagged items.Second,this thesis proposes to analyze how to use the differential sampling method to realize the personalized management of tags,and designs two time-efficient tag tracking methods based on differential sampling.By setting an sampling probability for each tag with differential tag sampling method,the channel resources can be allocated according to the importance of the tag,which significantly improve the efficiency of large-scale tag monitoring.