754|信息安全研究JournalofInformationSecurityResearch第9卷第8期2023年8月Vol.9No.8Aug.2023DOI:10.12379/j.issn.2096-1057.2023.08.06收稿日期:2022-11-28基金项目:海南省自然科学基金青年项目(620QN287,621QN0901);国家重点研发计划项目(2020YFB0505700);海南省自然科学基金项目(619MS076);海南省自然科学基金高层次人才项目(621RC602);海南省大学生创新创业项目(S202013892102);三亚市高校及医疗机构专项科技计划项目(2021GXYL53)引用格式:江荣旺,魏爽,龙草芳,等.基于联邦学习的车联网虚假位置攻击检测研究[J].信息安全研究,2023,9(8):754761基于联邦学习的车联网虚假位置攻击检测研究江荣旺魏爽龙草芳杨明(三亚学院海南三亚572022)(容淳铭院士工作站海南三亚572022)(rongwangjiang@sanyau.edu.cn)ResearchonMaliciousLocationAttackDetectionofVANETBasedonFederatedLearningJiangRongwang,WeiShuang,LongCaofang,andYangMing(SanyaCollege,Sanya,Hainan572022)(AcademicianRongChunmingWorkstation,Sanya,Hainan572022)AbstractMaliciousbehaviordetectionisanimportantpartofthesecurityneedsoftheInternetofvehicles.IntheInternetofvehicles,maliciousvehiclescanachievemaliciouslocationattackbyforgingfalsebasicsecurityinformation(BSM)information.Atpresent,thetraditionalsolutiontothemaliciouslocationattackontheInternetofvehiclesistodetectthemaliciousbehaviorofvehiclesthroughmachinelearningordeeplearning.Thesemethodsrequiredatacollecting,causingprivacyproblems.Inordertosolvethisproblems,thispaperproposedadetectionschemeofmaliciouslocationattacksontheInternetofvehiclesbasedonFederatedlearning.Theschemedoesnotneedtocollectuserdata,andthedetectionmodeluseslocaldataandsimulateddataforlocaltraining,whichensurestheprivacyofvehicleusers,reducesdatatransmissionandsavesbandwidth.ThemaliciouslocationattackdetectionmodelbasedonFederatedlearningwastrainedandtestedusingthepublicVeReMidataset,andtheperformanceofthedatacentricmaliciouslocationattackdetectionschemewascompared.Throughcomparison,theperformanceofmaliciouslocationattackdetectionbasedonFederatedlearningissimilartothatoftraditionaldatacentricmaliciouslocationattackdetectionscheme,butthemaliciousloca...
