基于DRL的SME-ISAC无人机语义安全系统资源分配算法

打开文本图片集

中图分类号：TN925 文献标志码：A文章编号：1003-3114（2025）06-1138-11开放科学（资源服务）标识码（OSID）：

Abstract：AsemanticsecuritysystemforUnmannedAerial Vehicle（UAV）edgecomputing，denoted Semantic-enhancedMobile Edge-ISC（SME-ISAC），isproposedbasedonDeepReinforcementLearing（DRL）andIntellgentReflectingSurface（IRS）withnan IntegratedSensingandCommuication （ISAC）framework.ArtficialNoise（AN）isintroducedtosuppressteeavesdropper'teptionrate，andultipleISuitsareleveragedtoasistothcommunicationandssingeamforing，terebyablngeficintintegationofcommunication，sensing，andcomputationresoucestoenhancethesystemscurerate.JointoptimizationofIRSphaseshifts， UAVtrajectosaskdgatios，maticsolalcandrifalseowerisedut，eiglingo convexoptimizatioproblemcharacteriedbysevereonconvexityandouplingamongvariables.Toaddressthesedificulties，SAISACbasedonSoftAtor-Critic（SAC）isdeveloped：thisagorithmemployingdiscreteSACfortrajectoryandsybolcountoptization andcontinuousSACforIRSphaseshiftandofloadingratioadjustment.Simulationstudiesdemonstratethatrapidconvergenceis achievedwithinashorttraininghoronandthattheproposdalgoritmsignificantlyoutpefostraditionalbaselinsinbothpase design and UAV trajectory optimization，thereby evidencing itssuperiority and practical applicability.

Keywords：UAV;ISAC;semantic communication;IRS

0引言

随着算力设备和人工智能算法的发展，语义通信被认为是一种新的通信范式，因此被广泛关注。（剩余18215字）