資源描述：
《基于遺傳算法的網(wǎng)絡(luò)編碼優(yōu)化new》由會(huì)員上傳分享，免費(fèi)在線閱讀，更多相關(guān)內(nèi)容在教育資源-天天文庫。
1、ISSN1000-9825,CODENRUXUEWE-mail:jos@iscas.ac.cnJournalofSoftware,Vol.20,No.8,August2009,pp.2269?2279http://www.jos.org.cndoi:10.3724/SP.J.1001.2009.03370Tel/Fax:+86-10-62562563?byInstituteofSoftware,theChineseAcademyofSciences.Allrightsreserved.?基于遺傳算法的
2、網(wǎng)絡(luò)編碼優(yōu)化+鄧亮,趙進(jìn),王新(復(fù)旦大學(xué)計(jì)算機(jī)科學(xué)技術(shù)學(xué)院,上海200433)GeneticAlgorithmSolutionofNetworkCodingOptimization+DENGLiang,ZHAOJin,WANGXin(SchoolofComputerScience,FudanUniversity,Shanghai200433,China)+Correspondingauthor:E-mail:xinw@fudan.edu.cnDengL,ZhaoJ,WangX.Geneticalgor
3、ithmsolutionofnetworkcodingoptimization.JournalofSoftware,2009,20(8):2269?2279.http://www.jos.org.cn/1000-9825/3370.htmAbstract:Afterthebestoptimizingapproachofnetworkcodingisbeingstudied,somemethodsareproposedbasedonthecharacteristicsofthenetworkcoding
4、overheadoptimizationproblem.First,twomodificationsareaddedtothepreprocessingphase:1)Howtogenerateafitnessvaluetoanetworkcodingschemeunderacertainnetworkcodingoptimizationrequestispresented.Thismakesdifferentnetworkcodingoptimizationproblemsbesolvedwitht
5、hesamegeneticalgorithm.2)Anadditionalexamprocessingofthemulti-inoutgoinglinksisimportedtoreducethesolutionspace.Second,experimentalresultsshowthattherandomgeneratedsolutionofnetworkcodingoptimizationproblemcanhardlyachievethemulticastrate,threenewstepsa
6、resuggestedbetakenwiththecommongeneticalgorithm:1)usemoredelicatemembergeneratingfunctiontogenerateinitialmembers;2)addnewmembersatthebeginningofeachroundofthegeneticalgorithmtoavoidlocalizedoptimization;3)assignafitnessvaluebasedoneachreceiver’sdatarat
7、eratherthan?1tothosenetworkcodingsolutionswhichcannotachievethemaxmulticastrate.Experimentalresultsshowdramaticimprovementsintermsofbothefficiencyandresult.Keywords:multicast;networkcoding;optimization;heuristic;geneticalgorithm摘要:在前人優(yōu)化研究方法的基礎(chǔ)上,結(jié)合網(wǎng)絡(luò)編碼優(yōu)化
8、問題自身的特點(diǎn)提出了新的解決方案.首先是算法的預(yù)處理部分:1)給出了統(tǒng)一的方法由不同的資源描述函數(shù)生成遺傳算法所必須的適應(yīng)值函數(shù),使得各種不同的網(wǎng)絡(luò)編碼資源優(yōu)化問題都能利用同樣的遺傳算法模型;2)通過檢驗(yàn)有多條輸入鏈路的輸出鏈路進(jìn)一步縮小優(yōu)化算法的搜索范圍.其次,針對(duì)網(wǎng)絡(luò)編碼資源優(yōu)化問題隨機(jī)解幾乎不能讓所有接收者都達(dá)到組播速率的特點(diǎn),在一般的遺傳算法中加入以下新的處理:1)在初始化階段使用更為精細(xì)的算法產(chǎn)生更高質(zhì)量的初始成員.2)在遺傳算法每次循環(huán)開始時(shí)
