电子测量技术ELECTRONICMEASUREMENTTECHNOLOGY第46卷第1期2023年1月DOI:10.19651/j.cnki.emt.2210264伪RFID标签信号的谱熵特征提取与选择*高威1吴海锋1,2曾玉1,2普崇荣1(1.云南民族大学电气信息工程学院昆明650504;2.云南省高校智能传感器网络及信息系统创新昆明650504)摘要:为解决RFID通信安全问题,本文提出一种新的物理层标签防伪技术,该技术对原始信号进行信号分离得到期望、噪音和标准化信号,然后对3种信号的谱熵特征进行特征提取,最后利用特征选择实现真伪标签分类。另外,本文还提出了一种新的交叉验证来客观测试物理层方法的性能。结果表明,本文方法准确率比传统方法提升近4%,在新交叉验证下,物理层方法的分类准确率会下降8%~10%,由此本文得到重要结论,在物理层识别方法中利用谱熵特征实现标签防伪有着重要意义。关键词:无线射频识别;防伪;安全;特征选择;交叉验证中图分类号:TN9文献标识码:A国家标准学科分类代码:510.4SpectralentropyfeatureextractionandselectionofpseudoRFIDtagsignalsGaoWei1WuHaifeng1,2ZengYu1,2PuChongrong1(1.SchoolofElectricalandInformationEngineering,YunnanMinzuUniversity,Kunming650504,China;2.InnovationofIntelligentSensorNetworkandInformationSysteminYunnanProvinceUniversity,Kunming650504,China)Abstract:InordertosolvetheproblemofRFIDcommunicationsecurity,thispaperproposesanewphysicallayertaganti-counterfeitingtechnology.Thistechnologyseparatestheoriginalsignaltoobtaintheexpected,noiseandnormalizedsignals,andextractsthespectralentropyfeaturesofthethreesignals.Finally,featureselectionisusedtoachievetrueandfalselabelclassification.Inaddition,thispaperproposesanewcross-validationtoobjectivelytesttheperformanceofthephysicallayermethod.Theresultsshowthattheaccuracyofthemethodinthispaperisnearly4%higherthanthatofthetraditionalmethod.Underthenewcross-validation,theclassificationaccuracyofthephysicallayermethodwilldropby8~10percentagepoints.Fromthis,wegetanimportantconclusion,itisofgreatsignificancetousethespectralentropyfeatureinthephysicallayeridentificationmethodtorealizethelabelanti-counterfeiting.Keywords:RFID;anti-counterfeiting;security;cross-validation;featureselection收稿日期:2022-06-09*基金项目:国家自然科...
