模糊PID控制算法和智能传感网络的水产养殖监控系统
作者:常波,张新荣
工
中国江苏省青浦区枚乘东路
摘要_根据水产养殖现代化发展的需求,将无线传感网络技术(WSN)引入到水产养殖现代化发展当中可以有效地解决水产养殖所面临的复杂工作环境.监控地点分散.高昂的铺线费用等问题.该系统以CCIIVIII0为核心来联接无线网络节点,这些节点服从zigbee协议.运用带有高精度传感器的数据采集终端来收集环境数据以获得实时的水产养殖环境参数.并且,运用模糊PID控制算法来处理环境参数以提高测试数据的准确性.各种传感器节点是嵌入式数据库来实现管理功能的.该系统实现了实时测量和控制养殖水体中的溶解氧,温度,pH和其它参数,降低了能源消耗,减少了时间等待,提高了水产养殖的智能控制水平和集约化程度.监控结果显示这个系统是高效实用的,并且也能够达到水产养殖环境参数监控系统的应用要求.
I..引言
在现代化水产养殖中,引进的生产与水产养殖安全方面的环境参数与水产养殖环境密切相关.对养殖水体的温度,溶解氧,PH值以及气体参数的自动监测为水产养殖的科学管理提供了准确的技术参数,这种科学管理对提高集约化养殖和实现水产养殖自动化程度具有重要意义[I.].现在,现有的监控系统经常采用分布式测量和控制,集中管理和采用微处理器作为核心的计算机配电监控模式,并且采用单片机.电气设备作为下位机和工业控制电脑作为上位机,上位机通过RSIVVIIIV和CAN总线与下位机相连接[II,III].但是这种控制模式具有布线问题,系统可靠性差,高昂的维护费用,较广的控制区域,散乱的监测节点,以及有限的远程遥控范围等缺点[IV,V].无线传感网络(WSN)拥有结构紧凑,易于布局,准确度高,成本低廉等优势,在工业监控领域具有广泛的应用前景,并且已经广泛引入到农业技术创新当中.例如 *51今日免费论文网|www.jxszl.com +Q: ^351916072* 
:环境监控,精确灌溉,农作物识别和可追溯性生产系统等.
这篇文章中,针对水产养殖生产的特点及其对监测系统的具体要求并且结合无线传感器网络(WSN)的优势,形成了I.个无线传感网络(WSN)的水产养殖控制系统.
II.系统整体设计
A_系统架构
传感器网络的实现是以zigbee树状拓扑为基础的.系统的整体架构如图I.所示.传感器节点的工作是收集水产养殖数据并将其传送到网关节点,网关节点的任务是接收数据信息并通过串行端口与监控中心进行交流,监控中心提供了I.个图形化的操作环境.
B_传感器网络节点的硬件结构
传感器节点是水产养殖监控系统的基础环节而且拥有对环境参数的收集,处理,无线通信以及其他功能.在水产养殖监控应用的背景下,对传感器节点的设计需要着重考虑这些因素:廉价,低能耗,稳定性和可靠性.该系统的无线通信模块采用射频芯片CCIIVIII0,CCIIVIII0支持II.IVGHz的IEEEVIII0II.I.V.IV协议,整合了ZigBee射频前端,存储器和微控制器在单个芯片上.CCIIVIII0利用AD转换器将传感器节点收集来的数据转换为数字信号,然后该数字信号通过无线多跳模式被传送到网关节点.网关节点同样可以使用CCIIVIII0,与传感器节点相似.终端节点的硬件框图如图II所示.
C_传感器测量电路的设计
考虑到节约成本,测量数据的准确性,以及其他要求即:不同的水生代谢,营养和生长以及其对池塘水温度,PH值,溶解氧的不同的要求.传感器的输出能够直接连接到CCIIVIII0的PO口,通过采用CCIIVIII0内部的ADC来完成的数据转换.
传感器及其在本设计中使用到的技术参数见表I..
III.模糊PID控制策略及其设计
A_模糊PID控制策略
着眼于水产养殖的特性即:多变量,大惯性,非线性,参数耦合,纯滞后性和较长的控制时间以及产生了显著的过冲,精确的数学模型无法建立,利用经典控制方法难以达到满意的控制精度.模糊控制模仿了人的思维来进行控制,拥有设计简单和强大的优势.因此,选择模糊控制规则就是控制系统的环境因素.模糊控制是根据控制量的实际变化来设计模糊PID控制器,并建立模糊控制规则表以达到良好的控制效果的控制.在这里仅以温度控制策略和设计来加以说明.
B_模糊PID控制的实现
模糊PID控制结合了模糊控制和PID控制,利用模糊数学的基本原理,应用模糊量来指示条件和行动规则,这些模糊控制规则都存储在计算机内,然后根据相应的实际情况,通过模糊推理,计算机控制系统自动调整PID参数.模糊PID控制器的结构见图III.
以误差和误差变化作为输入,温度模糊控制器的输入语言变量是温度误差e及其变化率ec,同时输出语言变量是控制参数的Kp,Ki,Kd的增量,也就是LJKp,LJKi,LJKd.在模糊规则下,条件语言变量构成输入空间,同时结束语言变量构成输出空间,对于每个语言变量,其值的模糊集具有相同的论域.根据实际工程经验,提取的Kp的模糊规则如表II所示.
PID控制是由规则,查表和计算的结果做出反应,然后进行模糊化处理,以实现温度控制.利用模糊规则进行模糊推理的过程是使用查表法来实现的.具体实施方法是分别存储LJKp,LJKi和LJKd的离线计算模糊控制表在存储器的数据块中,然后每隔III0毫秒周期性的采用中断方式调用查询模糊控制 *51今日免费论文网|www.jxszl.com +Q: ^351916072* 
表中的子程序,即按照定量e和ec的值来查询模糊控制表,并获得LJKp,LJKi和LJKd的定量值.程序流程图见图IV.
IV.系统软件设计
A_终端节点的软件设计
ZigBee设备具有能量检测和链路质量指示的功能.根据测试结果,ZigBee设备可以自动调整发射功率,可以在保证通信链路质量的前提下将能耗降至最低.在此设计中,由于监测在水产养殖环境数据的需要,WSN使用树形拓扑结构不仅比星形拓扑结构具有更好的功能和更大的覆盖面,而且还比网状网络更容易实现和维护.
在节点的软件设计中,通过调用ZigBee协议栈提供的API函数来完成设备初始化的同时,网络的配置和网络管理层网络的启动,无线adhoc网络包括I.些分布在养殖池塘的无线传感器节点成功启动.为了进I.步减少节点的功耗,该系统还具有常规的数据收集和定时的睡眠和唤醒功能,且该功能是灵活的,可动态配置的.传感器节点是主要负责收集环境数据,并把这些数据提供给路由节点.当没有数据发送或接收时,节点进入休眠模式,节点功耗跌入了最低.终端节点的软件流程图,见图V.
B_监控中心的软件设计
监控中心软件负责通过RSIIIIIII串口与网关节点通信.其目的是为了接收来自传感器节点的数据以实现人机交互,网络管理,传感器数据接收,数据处理和分析,以及数据库管理等功能.该设计Mscomn控制,采用VisualC++VI.0进行开发.Mscomn可以提供串行通信的所有功能,并且可以对串行端口进行数据读写.这个控件在沟通过程中封装了底层操作系统的程序,用户只需要设置监视器控件的属性和事件,用户和应用程序之间的异步串行通信可以轻松实现.控制中心的软件框图如图VI所示.
V.系统测试结果及分析
根据多项检测和控制,结合最新ZigBee无线网络技术来完成水产养殖环境监测系统的设计.根据设计要求,并考虑到成本,养殖池,选择I.0*I.0米,约I..VIII米水深的区域作为有效监测区,设置V个从传感节点(设置数量为NI.至NV)和I.个中心节点,监测环境参数.由V个传感器收集的实时数据通过无线通信发送到监控中心显示.因为是高精度的传感器,传输模块又对数据进行了优化,实际的数据传输质量得到了改善,与实测值的误差可以忽略不计.
根据多数鱼的生长和发展的需要,池塘水的最佳参数确定为水温IIV℃,PH值为VII,溶解氧为VIImg/L.I.旦检测参数被超越,电脑会显示检测地址和测量的参数值,并通常会在小于IIIs内发出警报.某节点的几个时间点所监视的数据平均值列于表III中.节点的部分时间点的监测数据见图VII.
从表III或图VII可以看出所测得的数据更准确,靠近最佳值.由实际的测量和计算,可以得出的结论是温度测量误差小于±0.II℃,PH值误差小于0.I.,溶解氧误差小于0.I.毫克/升.
VI.结论
水生环境监测系统的硬件和软件的设计结合了无线传感器网络技术和实时监测和管理温度,PH值,氧气和与水生生物生长相关的其他环境参数.该系统提供了I.个可行的,适用,成本相对较低的水产养殖自动化生产管理方案.该系统的网络结构是I.个简单的星形网络,每个传感器节点与汇聚节点进行有效的沟通,这简化了通信协议的设计.该系统具有低成本,低功耗,无需布线,组网灵活,界面友好等优点.实验结果表明,该系统能够及时发现池塘环境参数的变化,并能实现实时,准确地监测水生生物生长环境.
VII.致谢
这项工作是由中国国家自然科学基金(批准No.III0IXVIII.VIVIIIIX),II0I.0年中国淮安是科学和技术支持项目(批准号:SNI0IVV)和工II0I.0年度科技研究基金支持,中国(批准号:HGBIOIO).我们要感谢匿名审稿人明察秋毫的意见.
附件II:外文原文(复印件)
AquacultureMonitoringSystemBasedonFuzzy-PIDAlgorithmandIntelligentSensorNetworks
AquacultureMonitoringSystemBasedonFuzzy-PIDAlgorithmandIntelligentSensorNetworks
BoChang,XinrongZhang
HuaiyinInstituteofTechnology
MeiShengEastRoad,QingpuDistrict
Huaian,Jiangsu,China
Abstract-Accordingtothedevelopmentneedsofaquaculturemodernization,thewirelesssensornetwork(WSN)technologyisintroducedintothedevelopmentofaquaculturemonitoringsystem,whichcaneffectivelysolvetheproblemofcomplexworkenvironment,scatteredmonitoringlocations,andhighwiringcostintheaquaculture.ThesystemusesCCIIVIII0asthecoretodevelopwirelesssensornodeswhichfollowtheZigBeeprotocol,usesthedatacollectionterminalwithhigh-precisionsensortocollectdataoftheenvironmentforachievingreal-timemonitoringofaquacultureenvironmentandusesthefuzzyPIDcontrolalgorithmtoprocessenvironmentalparametersforimprovingtheaccuracyoftestdata.Themanagementfunctionofvarioussensornodesisachievedbasedonembeddeddatabase.Thesystemrealizedtherealtimemeasurementandcontrolforthedissolvedoxygen,temperature,pHandotherparametersoftheculturewater,reducedenergyconsumption,reducedlatencyandincreasedintensificationandintelligentlevelofaquaculture.Themonitoringresultshaveshownthatthissystemisusefulnessandefficiencyandcanmeettheapplicationrequirementofmonitoringenvironmentalparametersinaquaculture.
I.INTRODUCTION
Inmodernaquaculture,theenvironmentalparametersinaquacultureintroductionproductionandsafetyofaquaculturecloselyrelatestoenvironmentalconditions.Theautomaticmonitoringtothetemperature,dissolvedoxygen,PH,andotherparametersoftheculturewaterprovidesaccuratetechnicalparametersforthescientificmanagement,whichisveryimportanttoimprovethelevelofintensivefarmingandtoachieveaquacultureproductionautomation[I.].Atpresenttheexistingmonitoringsystemsoftenadoptsthedistributedmeasurementandcontrol,centralizedmanagementandcomputerdistributionmonitoringmodeusingmicroprocessorascore,andadoptssingle-chipmicrocomputer,electricalequipmentastheslavecomputerandindustrialcontrolcomputerasthemastercomputerviaRSIVVIIIVandCANbusconnectedwiththeslavecomputer[II,III].However,thiscontrolmodehasshortcomingsofwiringproblems,poorsystemreliability,highermaintenancecosts,widercontrolarea,scatteredmonitoringlocations,andlimitedremotecontrolrange,etc[IV,V].Wirelesssensornetwork(WSN)hastheadvantagesofcompactstructure,easylayout,highaccuracyandlowcost,etc,whichhasbroadapplicationprospectsinthefieldofindustrymonitoring[VI-VII],andisgraduallyintroducedintotheagriculturalproductiontechnologyinnovations,suchasenvironmentalmonitoring,precisionirrigation,cropidentificationandproductiontraceabilitysystems,etc[VIII-IX].
Inthispaper,aimingatthecharacteristicsofaquacultureproductionanditsspecificrequirementsformonitoringsystemsandcombiningadvantagesofWSNtechnology,aWSN-basedaquacultureenvironmentalmonitoringsystemwasdeveloped.
II.OVERALLDESIGNOFSYSTEM
A.SystemArchitecture
ThesensornetworkachievedisbasedonaZigBeetreetopology.TheoverallstructureofsystemisshowninFig.l.Thetaskofsensornodeistocollectaquacultureenvironmentaldataandtransmitthedatatothegatewaynode.Thegatewaynodeisresponsibleforreceivingdatainformationandcommunicatingwiththemonitoringcenterbytheserialport.Themonitoringcenterprovidesagraphicaloperatingenvironment.
B.TheHardwareStructureo/SensorNetworkNode
Sensornodeisthebasicunitofaquaculturemonitoringsystemandneedtohavetheenvironmentalparametersinformationcollection,processing,wirelesscommunications,andotherfunctions.Inthebackgroundofaquaculturemonitoringapplications,forthedesignofsensornodethesefactorssuchaslow-cost,lowpower,stability,andreliabilityarespeciallyconsidered.ThewirelesscommunicationmoduleofthissystemusesRFchipCCIIVIII0,whichsupportsII.IVGHzIEEEVIII0II.I.V.IVprotocol,integratesZigBeeRFfront-end,memoryandmicrocontrolleronasinglechip.TheCCIIVIII0convertsthecollecteddatafromthesensornodeintodigitalsignalsusingADconverter,thenthedataaretransmittedtothegatewaynodebywirelessmulti-hopmode.ThegatewaynodecanstillusetheCCIIVIII0,whichissimilartothesensornodes.ThehardwareblockdiagramoftenninalnodeisshowninFig.II.
C.Thedesignojsensormeasuringcircuit
Takingintoaccountsavingscost,dataaccuracyandotherrequirements,thatis,differentaquaticmetabolism,nutrition,andgrowthandthedifferentrequirementsfortemperature,pH,anddissolvedoxygenofthepondwater,theoutputofsensorcanbedirectlyconnectedtothePOportofCCIIVIII0,usinginternalADCofCCIIVIII0tocompletedataconversion.ThesensoranditstechnicalparametersusedinthisdesignareshowninTableI.
A.FuzzyPIDcontrolstrategy
Inviewofthecharacteristicsoftheaquaculturecontrol,thatis,multi-variable,largeinertia,nonlinear,couplingparameter,puredelayandalongertimespentinthecontrolandproducingsignificantovershoot,aprecisemathematicalmodelcannotbeestablishedandusingtheclassicalcontrolmethodisdifficulttoachievesatisfactorycontrolprecision.Fuzzycontrolstrategymimicsthehumanthoughttocontrol,withsimpledesignandrobustadvantages.Therefore,selectingafuzzycontrolalgorithmistocontroltheenvironmentalfactorsofthesystem.AccordingtoactualchangesofcontrolamounttodesignfuzzyPIDcontrollerandtoestablishthefuzzyruletable,agoodcontroleffectcanbeachieved.Hereonlythetemperaturecontrolstrategyanddesignaretobestated.
B.TheImplementationoJFuzzyPIDControl
ThefuzzyPIDcontrolisthatcombiningfuzzycontrolwithPIDcontrol,usingthebasicprinciplesoffuzzymathematics,applyingfuzzyamounttoindicateconditionsandactionsofrules,thesefuzzycontrolrulesarestoredincomputerknowledge,andthenaccordingtotheactualcorrespondingcircumstancesandusingfuzzyreasoning,thecomputercontrolsystemcanautomaticallyadjustthePIDparameters.ThestructureoffuzzyPIDcontrollerisshowninFigureIII.
Witherroranderrorchangeasinput,theinputlinguisticvariablesofthetemperaturefuzzycontrollerarethetemperatureerroreanditschangerateec,whiletheoutputlinguisticvariablearetheincrementsofcontrollingparametersKp,Ki,Kd,thatis,LJKp,LJKi,LJKd.Thepremiselinguisticvariablesinfuzzyrulesconstitutetheinputspace,whiletheconclusionlinguisticvariablesconstitutetheoutputspace.Foreachlinguisticvariable,thefuzzysetofitsvaluehasthesametheorydomain.Accordingtoactualprojectexperience,thefuzzyruleextractedofKpisshowninTableII.
PIDcontrolismadebythetreatingresultsoftherules,looking-uptableandcomputing,andthenmakingfuzzyprocessing,sothatcontroltemperatureisachieved.Theprocessofusingfuzzyrulesforfuzzyinferenceisimplementedbyusinglook-uptablemethod.TheconcreteimplementationmethodistostorerespectivelythefuzzycontroltableofLJKp,LJKiandLJKdcalculatedoff-lineinthememorydatablocks,theneveryIII0msperiodicinterruptcallsthesubroutineofqueryingfuzzycontroltable,thatis,accordingtothequantitativevalueofeandectoqueryfuzzycontroltableandtoobtainquantitativevaluesofLJKp,LJKiandLJKd.TheprogramflowchartisshowninFig.IV.
IV.SYSTEMSOFTWAREDESIGN
A.Thesoftwaredesignofterminalnode
ZigBeedeviceshavethefunctionofenergydetectionandlinkqualityindicator.Accordingtotestresults,ZigBeedevicecanautomaticallyadjustthetransmitpowerandcanminimumconsumedeviceenergyonthepremiseofensuringcommunicationlinkquality.Inthisdesign,fortheneedsofmonitoringenvironmentalparametersinaquaculture,WSNusingatreetopologystructurehastheadvantagesofbetterfunctionandlargercoveragethanthestarstructure,whileeasiertoimplementandmaintainthanthemeshnetwork.
Inthesoftwaredesignofnodes,bycallingtheAPIfunctionsprovidedZigBeeprotocolstacktocompletetheinitializationofdevice,theconfigurationofthenetworkandthestartingupnetworkofnetworkmanagementlayer,thewirelessadhocnetworkconsistedofwirelesssensornodesdistributedinanumberofbreedingpondareachieved.Toreducenodepowerconsumptionfurther,thesystemalsohasthefunctionoftheregulardatacollectionandregularsleepandwake-up,whichfunctionisflexibleanddynamicallyconfigurable.Sensornodeismainlyresponsibleforcollectingenvironmentaldataandsendsthesedatatotheroutingnode.Whenthereisnodatatosendorreceive,thenodeenterintosleepmode,thenodepowerconsumptionfellintothelowest.ThesoftwareflowchartoftheterminalnodesisshowninFig.V.
B.Thesoftwaredesignofthemonitoringcenter
ThemonitoringcentersoftwareisinchargeofcommunicatingwiththegatewaynodethroughtheRSIIIIIIIserialport.Itspurposeistoreceivedatafromsensornodestoachievethefunctionsofhuman-computerinteraction,networkmanagement,receivingsensorinformation,dataprocessingandanalysis,anddatabasemanagement.ThedesignisbasedonMscomncontrol,usingVisualC++VI.0fordevelopment.Mscomncanprovideallthefunctionalityoftheserialcommunicationandcanachievereaddatafromserialportorwritedatatotheserialport.Thiscontrolencapsulatestheunderlyingoperatingproceduresinthecommunicationprocessandusersonlyneedtosetthemonitorcontrolspropertiesandevents,theasynchronousserialcommunicationbetweentheuserandtheapplicationcanbeeasilyimplemented.ThesoftwareblockdiagramofcontrolcenterisshowninFig.VI.
V.SYSTEMTESTRESULTSANDANALYSIS
Uponcompletionofanumberdetectionandcontrol,combinedwiththelatestZigBeewirelessnetworkingtechnology,aquacultureenvironmentalmonitoringsystemisdesigned.Accordingtodesignrequirementsandtakingintoaccountthecost,thebreedingpond,effectiverangeofwhichisI.0mbyI.0m,isselectedasthedetectedarea,aboutI..VIIImwaterdepth,whereVslavesensingnodes(settingthenumberfortheNI.toNV)andacentralnodearearranged,theenvironmentalparametersweredetected.Thedatacollectedreal-timebyfivesensorsaresenttothemonitoringcenterbywirelesscommunicationanddisplayed.Becauseofhigh­precisionsensor,optimizedbytransmissionmodule,theactualdatatransmissionqualityhasimprovedandtheerrorwiththemeasuredvaluecanbeignored.
Accordingtotheneedsofmajorityoffishgrowthanddevelopment,theoptimumparameterofpondwaterdeterminedisthatthewatertemperatureisIIV°C,PHvalueisVII,thedissolvedoxygenisVIImglL.Oncedetectionparametersisoverrunning,computerwilldisplayaddressandgaugeparametervaluesandwillraisethealarmusuallyinlessthanIIIs.ThemonitoredaverageddataofonenodeatseveraltimepointsareshowninTableIII.ThemonitoringdataofpartofthenodeatseveraltimepointsisshowninFig.VII.
FromthetableIIIorFig.VII,itmaybeseenthatthemeasureddataaremoreaccurate,locatednearthebestvalue.Byactualmeasuringandcalculating,itcanbeconcludedthattemperaturemeasurementerrorislessthan±0.II°C,PHvalueerrorislessthan0.I.anddissolvedoxygenerroriswithin0.I.mg/L.
VI.CONCLUSIONS
Thehardwareandsoftwareofaquaticenvironmentmonitoringsystemaredesignedcombinedwithwirelesssensornetworktechnologyandthereal-timemonitoringandmanagementtotemperature,PHvalue,oxygenandotherparametersoftheaquaticcreaturesgrowthenvironmentwasrealized,whichprovidesafeasible,applicable,relativelylowcostsolutionsfortheaquacultureautomationproductionmanagement.Thesystemnetworkstructureisasimplestarnetworkandeachsensornodeisintheeffectivecommunicationscopeofconvergencenode,whichsimplifiedthedesignofcommunicationprotocol.Thesystemhastheadvantagesoflow-cost,lowpower,nowiring,flexiblenetworking,andfriendlyinterfaceandsoon.Theresultshaveshownthatthissystemcanpromptlydetectthechangesofenvironmentparameterinpondandcanrealizereal-timeandaccuratemonitoringtotheaquaticcreaturesgrowthofenvironment.
ACKNOWLEDGMENT
ThisworkwassupportedbytheNationalNaturalScienceFoundationofChina(GrantNo.III0IXVIII.VIVIIIIX),theII0I.0ScientificandTechnologicalSupportProjectsofHuaianCity,China(GrantNo.SNI0IVV)andtheII0I.0TechnologyResearchFundofHuaiyinInstituteofTechnology,China(GrantNo.HGBIOIO).Wewouldliketothanktheanonymousreviewersfortheirperspicaciouscomments.
REFERENCES
[I.]F.J.Pierce,T.V.Elliott.Regionalandon-farmwirelesssensornetworksforagriculturalsystemsnEasternWashington[J].ComputersandElectronicsinAgriculture,II00S,VII.(I):IIIII-IVIII.
[II]�engXZ,LiuG,ZhengDP,etal.StudyanddevelopmentofafieldmformatIonacqUIsItIonsystembasedonwirelesstechnique[C]IIActualTasksonAgriculturalEngineering,Opatija,CROATIA,II00VI:IIIVIII.-IIIVIIVII.
[III]HillJL,CullerDE.Mica:awirelessplatformfordeeplyembeddednetworks[J].TEEEMicro,II00II,IIII(VI):I.II-IIII.
[IV]ZigbeeAlliance.ZigbeeSpecificationVI..0[EB/OL].http://www.zigbee.org,II00V.
[V]CallawayEH.WirelessSensorNetwork:ArchitectureandProtocols[J].CRCPressLI.C,II00IV:IVI.-VIII.
[VI]DianeJCook,SajalKDas.Smartenvironments:technologies,protocolsandapphcatlOns[M].Hoboken:JohnWileyandSons.II00V:I.III-IVVI.
[VII]BogenaHR,HuismanJA,OberdrsterC,etal.Evaluationofalow-costsoilwatercontentsensorforwirelessnetworkapplications[J].JournalofHydrology,II00VII,IIIIVIV(I.-II):IIIII-IVII.
[VIII]PierceFJ,ElliottaTV.RegionalandOn-FarmWirelessSensorNetworksforAgriculturalSystemsinEasternWashington[J].ComputersandElectronicsinAgriculture,II00S,VII(l):IIIII-IVIII.
[IX]ProakisJG,SalehiM.Communicationsystemsengineering[M].UpperSaddleRiver:PrenticeHal,III00I..

原文链接：http://www.jxszl.com/lwqt/wxzs/24127.html