跳到主要內容

簡易檢索 / 詳目顯示

研究生: 鄭郁全
Yu-Quan Zheng
論文名稱: 混合深度強化學習用於多功能RIS裝載於無人機之流體天線輔助之全雙工網路
Hybrid Deep Reinforcement Learning for Multi-Functional RIS-Mounted UAVs in Fluid Antenna-Assisted Full-Duplex Networks
指導教授: 沈立翔
Li-Hsiang Shen
口試委員:
學位類別: 碩士
Master
系所名稱: 資訊電機學院 - 通訊工程學系
Department of Communication Engineering
論文出版年: 2025
畢業學年度: 113
語文別: 中文
論文頁數: 69
中文關鍵詞: 無人機通訊全雙工多功能可重構智慧表面深度強化學習流體天線
外文關鍵詞: UAV, Full-Duplex, MF-RIS, Deep Reinforcement Learning, Fluid Antenna
相關次數: 點閱:114下載:0
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報
  • 為了滿足第六代(6G)行動通訊系統的高流量需求,已經為未
    來的 6G 無線網路提出了各種新興技術。本論文提出了一種全雙工
    (FD)通訊架構,透過部署無人機(UAV)並輔以多功能可重構智慧
    表面(MF-RIS)和流體天線(FA)來支援上行(UL)和下行(DL)
    用戶,以最大限度地提高系統能源效率(EE)。MF-RIS 具有同時反
    射、折射、放大和收集能量的能力。系統模型包括無人機到基地台、
    無人機到使用者的雙向通道,以及 MF-RIS 放大係數、相位矩陣等控
    制變數。為了實現 EE 優化,我們設計了一種具有自註意力機制的混
    合深度強化學習 (DRL) 和超參數優化演算法 (HIPPO),該演算法將
    深度 Q 網路 (DQN) 與近端策略優化 (PPO) 相結合,以優化無人機的
    位置、發射波束成形、FA 位置和 MF-RIS 參數。模擬結果表明,與
    傳統 RIS、固定天線陣列、半雙工和 FD 或其他現有的 DRL 方法相
    比,所提出的 HIPPO 方案在不同場景下均實現了最高的 EE 性能。


    Tomeetthehightrafficdemandinsixthgeneration(6G)mobilecommunica-
    tion systems,variousemergingtechnologieshavebeenproposedforfuture6G
    wireless networks.Thisthesisproposesafull-duplex(FD)communication
    architectureinsupportofuplink(UL)anddownlink(DL)usersbydeploying
    unmanned aerialvehicles(UAVs)complementedbyamulti-functionalrecon-
    figurable smartsurface(MF-RIS)andfluidantennas(FA)tomaximizethe
    system energyefficiency(EE).MF-RIShastheabilitytoreflect,refract,am-
    plify andharvestenergysimultaneously.Thesystemmodelincludesbidirec-
    tional channelsfromdronetobasestationandfromdronetouser,aswellas
    MF-RIS amplificationfactor,phasematrix,andothercontrolvariables.To
    achievetheEEoptimization,wedesignahybriddeepreinforcementlearning
    (DRL) withself-attentionmechanismandhyperparameteroptimizationalgo-
    rithm (HIPPO)thatcombinesdeepQ-network(DQN)withproximalpolicy
    optimization (PPO)tooptimizeUAVs’positions,transmitbeamforming,FA
    positionsandMF-RISparameters.Simulationresultsshowthatcompared
    to traditionalRIS,fixedantennaarrays,half-duplexandFDorotherex-
    isting DRLmethods,theproposedHIPPOschemeaccomplishesthehighest
    EE performanceunderdifferentscenarios.

    Contents Chinese Abstract i English Abstract ii Contents iii List ofFigures v List ofTables vii 1 Introduction 1 2 SystemModelandProblemFormulation 7 2.1 SystemModel........................... 7 2.2 ChannelModeling........................ 10 2.3 FluidAntennaSystem...................... 12 2.4 Multi-HopMF-RISChannelModel............... 13 2.5 PowerDissipationModel..................... 18 2.5.1 PowerModelofUAV................... 18 2.5.2 PowerModelofMF-RIS................. 19 2.6 ProblemFormulation....................... 21 3 ProposedAlgorithm 23 3.1 AttentionMechanism....................... 23 3.2 DeepReinforcementLearning.................. 27 3.2.1 State,Action,andRewardSpace............ 28 3.3 HyperparameterOptimizationbasedonPPO......... 32 4 PerformanceEvaluation 37 4.1 SystemParameters........................ 37 4.2 SimulationResult......................... 38 4.3 ComputationalComplexityAnalyze............... 51 5 Conclusions 54 Bibliography 55

    Bibliography
    [1] C.-J. Ku,L.-H.Shen,andK.-T.Feng,“Reconfigurableintelligentsurfaceassisted
    interferencemitigationfor6Gfull-duplexMIMOcommunicationsystems,”in Proc.
    IEEE AnnualInternationalSymposiumonPersonal,IndoorandMobileRadioCom-
    munications(PIMRC), September2022,pp.327–332.
    [2] Y. Zeng,R.Zhang,andT.J.Lim,“Wirelesscommunicationswithunmannedaerial
    vehicles:Opportunitiesandchallenges,”vol.54,no.5,pp.36–42,May2016.
    [3] R. I.Bor-Yaliniz,A.El-Keyi,andH.Yanikomeroglu,“Efficient3-Dplacementofan
    aerial basestationinnextgenerationcellularnetworks,”in Proc.IEEEInternational
    ConferenceonCommunications(ICC), May2016.
    [4] M. Mozaffari,W.Saad,M.Bennis et al., “Efficientdeploymentofmultipleunmanned
    aerial vehiclesforoptimalwirelesscoverage,”vol.20,no.8,pp.1647–1650,August
    2016.
    [5] A. Al-Hourani,S.Kandeepan,andS.Lardner,“OptimalLAPaltitudeformaximum
    coverage,”vol.3,no.6,pp.569–572,July2014.
    [6] A. MerwadayandI.Guvenc,“UAVassistedheterogeneousnetworksforpublicsafety
    communications,”in Proc.IEEEWirelessCommunicationsandNetworkingConfer-
    enceWorkshops(WCNCW), March2015,pp.329–334.
    [7] A. Osseiran,F.Boccardi,V.Braun et al., “Scenariosfor5Gmobileandwireless
    communications:thevisionofthemetisproject,”vol.52,no.5,pp.26–35,May
    2014.
    [8] Y. Liu,S.Wang,L.Ma et al., “Energy-efficientRIS-UAVrelaywithtrajectoryopti-
    mization andfaircommunication,”in Proc.IEEEInternationalConferenceonCom-
    puter andCommunications(ICCC), December2022,pp.339–343.
    55
    [9] Y. Yao,K.Lv,S.Huang et al., “3Ddeploymentandenergyefficiencyoptimiza-
    tion basedonDRLforRIS-assistedair-to-groundcommunicationsnetworks,” IEEE
    TransactionsonVehicularTechnology, vol.73,no.10,pp.14988–15003,October
    2024.
    [10] Q. WuandR.Zhang,“Towardssmartandreconfigurableenvironment:Intelligent
    reflecting surfaceaidedwirelessnetwork,”vol.58,no.1,pp.106–112,January2020.
    [11] C. Huang,A.Zappone,G.C.Alexandropoulos et al., “Reconfigurableintelligent
    surfaces forenergyefficiencyinwirelesscommunication,”vol.18,no.8,pp.4157–
    4170, August2019.
    [12] W. Wang,W.Ni,H.Tian et al., “Multi-functionalreconfigurableintelligentsurface:
    System modelingandperformanceoptimization,”vol.23,no.4,pp.3025–3041,April
    2024.
    [13] Y.-T. Li,L.-H.Shen,K.-T.Feng et al., “Geneticmulti-agentreinforcementlearning
    for multipledouble-sidedSTAR-RISsinfull-duplexmimonetworks,”in Proc.IEEE
    International ConferenceonCommunications(ICC), June2024,pp.5003–5008.
    [14] D. Zhang,S.Ye,M.Xiao et al., “Fluidantennaarrayenhancedover-the-aircomputa-
    tion,” IEEE WirelessCommunicationsLetters, vol.13,no.6,pp.1541–1545,March
    2024.
    [15] K.-K. Wong,W.K.New,X.Hao et al., “Fluidantennasystem—part i:Preliminar-
    ies,” IEEE CommunicationsLetters, vol.27,no.8,pp.1919–1923,August2023.
    [16] L. Zhu,W.Ma,andR.Zhang,“Movableantennasforwirelesscommunication:Op-
    portunitiesandchallenges,” IEEE CommunicationsMagazine, vol.62,no.6,pp.
    114–120, June2024.
    [17] K.-K. Wong,A.Shojaeifard,K.-F.Tong et al., “Fluidantennasystems,” IEEE Trans-
    actions onWirelessCommunications, vol.20,no.3,pp.1950–1962,March2021.
    [18] X. Liu,Y.Liu,andY.Chen,“Machinelearningempoweredtrajectoryandpassive
    beamformingdesigninUAV-RISwirelessnetworks,” IEEE JournalonSelectedAreas
    in Communications, vol.39,no.7,pp.2042–2055,July2021.
    [19] W. Xue,P.Kolaric,J.Fan et al., “Inversereinforcementlearningintrackingcon-
    trol basedoninverseoptimalcontrol,” IEEE TransactionsonNeuralNetworksand
    LearningSystems, vol.52,no.10,pp.10570–10581,October2022.
    56
    [20] K. Doya,“Reinforcementlearningincontinuoustimeandspace,” NeuralComputa-
    tion, vol.12,no.1,pp.219–245,January2000.
    [21] H. ModaresandF.L.Lewis,“Optimaltrackingcontrolofnonlinearpartially-
    unknownconstrained-inputsystemsusingintegralreinforcementlearning,” Automat-
    ica, vol.50,no.7,pp.1780–1792,July2014.
    [22] J. Zhang,Z.Wang,andH.Zhang,“Data-basedoptimalcontrolofmultiagentsys-
    tems: Areinforcementlearningdesignapproach,” IEEE TransactionsonCybernetics,
    vol.49,no.12,pp.4441–4449,December2019.
    [23] B. Kiumarsi,F.L.Lewis,H.Modares et al., “ReinforcementQ-learningforoptimal
    trackingcontroloflineardiscrete-timesystemswithunknowndynamics,” Automatica,
    vol.50,no.4,pp.1167–1175,April2014.
    [24] Y. Jiang,J.Fan,T.Chai et al., “Trackingcontrolforlineardiscrete-timenetworked
    controlsystemswithunknowndynamicsanddropout,” IEEE TransactionsonNeural
    Networks andLearningSystems, vol.29,no.10,pp.4607–4620,October2018.
    [25] H. Mei,K.Yang,Q.Liu et al., “3D-trajectoryandphaseshiftdesignforRIS-assisted
    UAVsystemsusingdeepreinforcementlearning,” IEEE TransactionsonVehicular
    Technology, vol.71,no.3,pp.3020–3029,March2022.
    [26] P.-C.Wu,L.-H.Shen,K.-T.Feng et al., “Federatedreinforcementlearningformulti-
    dual-STAR-RISassistedDFRC-enabledmulti-BSinISACsystems,”in Proc.IEEE
    International ConferenceonCommunications(ICC), June2024,pp.2986–2991.
    [27] X. Li,L.Dong,L.Xue,andC.Sun,“Hybridreinforcementlearningforoptimal
    controlofnon-linearswitchingsystem,” IEEE TransactionsonNeuralNetworksand
    LearningSystems, vol.34,no.11,pp.9161–9170,November2023.
    [28] R. Zhong,Y.Liu,X.Mu et al., “HybridreinforcementlearningforSTAR-RISs:a
    coupled phase-shiftmodelbasedbeamformer,” IEEE JournalonSelectedAreasin
    Communications, vol.40,no.9,pp.2556–2569,September2022.
    [29] L.-H. ShenandY.-H.Chiu,“RIS-aidedfluidantennaarray-mountedUAVnetworks,”
    IEEE WirelessCommunicationsLetters, vol.14,no.4,pp.1049–1053,January2025.
    [30] A. Faisal,I.Al-Nahhal,O.A.Dobre et al., “Deepreinforcementlearningforoptimiz-
    ing RIS-assistedHD-FDwirelesssystems,” IEEE CommunicationsLetters, vol.25,
    no. 12,pp.3893–3897,December2021.
    57
    [31] Y. Cui,T.Lv,andY.Cao,“DRL-basedresourcemanagementinRIS-assisteduplink
    cell-free network,”in Proc.IEEEGlobecomWorkshops(GCWkshps), December2022,
    pp. 1495–1500.
    [32] J. Chen,Y.Xu,D.Yang et al., “UAV-assistedISCCnetworks:jointresourceand
    trajectoryoptimization,” IEEE WirelessCommunicationsLetters, vol.13,no.9,pp.
    2372–2376, September2024.
    [33] Y. Wan,Z.Zhao,J.Tang et al., “Multi-UAVformationobstaclesavoidancepath
    planning basedonPPOalgorithm,”in Proc.InternationalConferenceonBigData
    and InformationAnalytics(BigDIA), December2023,pp.55–62.
    [34] K. Chikhaoui,H.Ghazzai,andY.Massoud,“PPO-basedreinforcementlearningfor
    UAVnavigationinurbanenvironments,”in Proc.IEEEInternationalMidwestSym-
    posiumonCircuitsandSystems(MWSCAS), August2022.
    [35] A. Al-Hourani,S.Kandeepan,andS.Lardner,“Optimallapaltitudeformaximum
    coverage,” IEEE WirelessCommunicationsLetters, vol.3,no.6,pp.569–572,De-
    cember2014.
    [36] Z. Li,J.Zhang,J.Zhu et al., “Risenergyefficiencyoptimizationwithpracticalpower
    models,”in Proc.InternationalWirelessCommunicationsandMobileComputing
    (IWCMC), June2023,pp.1172–1177.
    [37] K. Ntontin,A.A.Boulogeorgos,E.Björnson et al., “Wirelessenergyharvesting
    for autonomousreconfigurableintelligentsurfaces,” IEEE TransactionsonGreen
    CommunicationsandNetworking, vol.7,no.1,pp.2473–2400,March2023.
    [38] K. Ntontin,A.-A.A.Boulogeorgos,E.Björnson et al., “Millimeterwavevs.THz
    energy harvestingforautonomousreconfigurableintelligentsurfaces,”in Proc.IEEE
    International ConferenceonCommunicationsWorkshops(ICCWorkshops), May
    2022, pp.1213–1218.
    [39] H. T.Le,T.V.Nguyen,H.T.T.Pham et al., “Harvestedenergyevaluationoffree-
    space opticsris-assistedground-hap-uavsystemovercompositechannels,”in Proc.
    IEEE VehicularTechnologyConference(VTC2024-Spring), June2024.
    [40] Q. WuandR.Zhang,“Intelligentreflectingsurfaceenhancedwirelessnetworkvia
    jointactiveandpassivebeamforming,” IEEE TransactionsonWirelessCommunica-
    tions, vol.18,no.11,pp.5394–5409,November2019.
    58
    [41] A. Vaswani,N.Shazeer,N.Parmar et al., “Attentionisallyouneed,”in Proc.Ad-
    vancesinNeuralInformationProcessingSystems, I.Guyon,U.V.Luxburg,S.Bengio
    et al., Eds.,vol.30,2017,pp.1–11.
    [42] C. Meng,K.Xiong,W.Chena et al., “Sum-ratemaximizationinSTAR-RIS-assisted
    RSMA networks:APPO-basedalgorithm,” IEEE InternetofThingsJourna, vol.11,
    no. 4,pp.5667–5680,February2024.
    [43] T. M.Hoang,L.T.Dung,,B.C.Nguyen et al., “Outageprobabilityandthroughput
    of mobilemultiantennaUAV-assistedFD-NOMArelaysystemwithimperfectCSI,”
    IEEE SystemsJournal, vol.17,no.1,pp.1477–1488,June2023.
    [44] P.Raut,K.Singh,C.-P.Li et al., “NonlinearEH-basedUAV-assistedFDIoTnet-
    works:Infiniteandfiniteblocklengthanalysis,” IEEE InternetofThingsJournal,
    vol.8,no.24,pp.17655–17668,December2021.
    [45] 3GPP,“Studyonchannelmodelforfrequenciesfrom0.5to100GHz(v16.1.0,release
    16),” 3GPP, SophiaAntipolis,France,Tech.Rep.TR38.901, December2019.
    [46] H. Yan,Y.Chen,andS.-H.Yang,“UAV-enabledwirelesspowertransferwithbase
    station chargingandUAVpowerconsumption,” IEEE TransactionsonVehicular
    Technology, vol.69,no.11,pp.12883–12896,November2020.
    [47] L. Jiao,K.Yu,J.Chen et al., “Performanceanalysisofuplink/downlinkdecoupled
    access incellular-V2Xnetworks,” IEEE TransactionsonMobileComputing, vol.23,
    no. 5,pp.5516–5630,May2024.
    [48] M. Li,H.Li,P.Ma,andH.Wang,“EnergymaximizationforgroundnodesinUAV-
    enabled wirelesspowertransfersystems,” IEEE InternetofThingsJournal, vol.10,
    no. 19,pp.17096–17109,October2023.

    QR CODE
    :::