RFID Counterfeit Tag Identification Under Signal Statistical Characteristics

Radio Frequency Identification(RFID)is an automatic identification technology that exchanges data information with electronic tags with unique identification(ID)information through wireless communication.If combined with computer network technology,an Internet of Things that can connect the virtual world and the real world can be built.The widespread use of the Internet of Things can provide strong technical support for the development of the world economy,and can greatly reduce the operating costs of world logistics,which has attracted great attention from all countries.The UHF RFID system of passive tags has the advantages of low cost,small antenna,good confidentiality,and long reading and writing distance.It is considered to be one of the best front-end technologies of the Internet of Things in the 21st century.In recent years,the scientific community has been emphasizing the security of information exchange,and information security issues are very important to ensure reliable interaction between devices.An RFID system has two main components,one or more readers(or interrogators)and tags.The purpose of the reader is to send commands to the tags over a wireless channel.After receiving the reader’s command,the tag decodes the transmission and responds with the appropriate data.However,the tag can respond to signals from different RFID readers,and any person or organization with a reader can freely access the RFID reader.Access control labels,information security issues are difficult to guarantee.Since radio frequency identification(RFID)tags have the disadvantage of being easily cloned,anti-counterfeiting of tags is an important issue in RFID communication security.Physical layer label recognition is a new technology for label anti-counterfeiting,which introduces machine learning ideas and achieves anti-counterfeiting by learning label signal features.Due to its low cost and no need to add additional hardware,it has received widespread attention in recent years.However,there are still some issues with this technology that have not been fully discussed,such as the characteristics of label signals and cross validation testing.In this regard,this article aims to study the anti-counterfeiting performance of RFID communication by simulating pseudo RFID tag attack scenarios to verify the recognition rate of fake RFID tags.At the same time,a new feature learning method is proposed to separate the extracted signals,extracting time-frequency statistical features not only from the original tag signal,but also from its expected,noise,and standardized signals,and using feature selection to achieve true and false tag classification.At the same time,this article also proposes a new cross validation method to objectively test the performance of physical layer methods.This new cross validation method can more accurately simulate real-world attack scenarios.The experiment used software radio equipment to classify 140 labels from 3 manufacturers and 7 categories.The results showed that the classification accuracy of our method improved by 3-4 percentage points compared to traditional methods.In addition,under the new cross validation,the classification accuracy of the physical layer method will decrease by 8-10 percentage points.Therefore,this article draws an important conclusion that the performance of the physical layer recognition method will depend on whether the training library has pseudo label class data.