1.一(yi)種多智(zhi)(zhi)能(neng)(neng)超(chao)(chao)(chao)(chao)(chao)(chao)表(biao)(biao)(biao)(biao)面(mian)(mian)輔助多播通(tong)信(xin)(xin)系(xi)(xi)(xi)統(tong)聯合波(bo)(bo)(bo)(bo)束(shu)(shu)成形(xing)方(fang)法(fa),其(qi)特征在(zai)于,基(ji)(ji)(ji)(ji)于下(xia)(xia)行多組(zu)(zu)多播通(tong)信(xin)(xin)系(xi)(xi)(xi)統(tong),所述系(xi)(xi)(xi)統(tong)包括基(ji)(ji)(ji)(ji)站(zhan)、用戶(hu)(hu)和(he)智(zhi)(zhi)能(neng)(neng)超(chao)(chao)(chao)(chao)(chao)(chao)表(biao)(biao)(biao)(biao)面(mian)(mian),其(qi)中,基(ji)(ji)(ji)(ji)站(zhan)包含(han)n個(ge)(ge)天線(xian)陣元(yuan),用戶(hu)(hu)有(you)(you)(you)k個(ge)(ge),k個(ge)(ge)用戶(hu)(hu)被分為(wei)(wei)(wei)g組(zu)(zu),第(di)(di)g組(zu)(zu)用戶(hu)(hu)集合為(wei)(wei)(wei)基(ji)(ji)(ji)(ji)站(zhan)向同一(yi)組(zu)(zu)用戶(hu)(hu)發送(song)同樣的(de)(de)廣播信(xin)(xin)號;智(zhi)(zhi)能(neng)(neng)超(chao)(chao)(chao)(chao)(chao)(chao)表(biao)(biao)(biao)(biao)面(mian)(mian)有(you)(you)(you)l塊(kuai),第(di)(di)l塊(kuai)配備(bei)ml個(ge)(ge)反(fan)射單元(yuan),其(qi)反(fan)射系(xi)(xi)(xi)數(shu)矩(ju)陣為(wei)(wei)(wei)對角(jiao)陣θl,m為(wei)(wei)(wei)第(di)(di)l塊(kuai)智(zhi)(zhi)能(neng)(neng)超(chao)(chao)(chao)(chao)(chao)(chao)表(biao)(biao)(biao)(biao)面(mian)(mian)的(de)(de)第(di)(di)m個(ge)(ge)反(fan)射單元(yuan)的(de)(de)反(fan)射系(xi)(xi)(xi)數(shu),l=1,…,l,m=1,…,ml,θl=vecd(θl)為(wei)(wei)(wei)第(di)(di)l塊(kuai)智(zhi)(zhi)能(neng)(neng)超(chao)(chao)(chao)(chao)(chao)(chao)表(biao)(biao)(biao)(biao)面(mian)(mian)被動波(bo)(bo)(bo)(bo)束(shu)(shu)成形(xing)向量(liang)(liang)(liang)(liang)(liang),vecd(x)表(biao)(biao)(biao)(biao)示(shi)通(tong)過(guo)取矩(ju)陣x的(de)(de)主(zhu)對角(jiao)線(xian)元(yuan)素而得到(dao)(dao)的(de)(de)列向量(liang)(liang)(liang)(liang)(liang),為(wei)(wei)(wei)所有(you)(you)(you)智(zhi)(zhi)能(neng)(neng)超(chao)(chao)(chao)(chao)(chao)(chao)表(biao)(biao)(biao)(biao)面(mian)(mian)總波(bo)(bo)(bo)(bo)束(shu)(shu)成形(xing)向量(liang)(liang)(liang)(liang)(liang),上標(·)t表(biao)(biao)(biao)(biao)示(shi)轉置,為(wei)(wei)(wei)所有(you)(you)(you)智(zhi)(zhi)能(neng)(neng)超(chao)(chao)(chao)(chao)(chao)(chao)表(biao)(biao)(biao)(biao)面(mian)(mian)反(fan)射單元(yuan)總和(he);基(ji)(ji)(ji)(ji)站(zhan)與(yu)智(zhi)(zhi)能(neng)(neng)超(chao)(chao)(chao)(chao)(chao)(chao)表(biao)(biao)(biao)(biao)面(mian)(mian)獲取的(de)(de)信(xin)(xin)道狀(zhuang)態信(xin)(xin)息(xi),包含(han)每個(ge)(ge)幀的(de)(de)第(di)(di)一(yi)個(ge)(ge)時(shi)隙(xi)獲取基(ji)(ji)(ji)(ji)站(zhan)到(dao)(dao)第(di)(di)l塊(kuai)智(zhi)(zhi)能(neng)(neng)超(chao)(chao)(chao)(chao)(chao)(chao)表(biao)(biao)(biao)(biao)面(mian)(mian)信(xin)(xin)道矩(ju)陣第(di)(di)l塊(kuai)智(zhi)(zhi)能(neng)(neng)超(chao)(chao)(chao)(chao)(chao)(chao)表(biao)(biao)(biao)(biao)面(mian)(mian)到(dao)(dao)第(di)(di)k個(ge)(ge)用戶(hu)(hu)的(de)(de)信(xin)(xin)道矢(shi)(shi)量(liang)(liang)(liang)(liang)(liang)基(ji)(ji)(ji)(ji)站(zhan)到(dao)(dao)第(di)(di)k個(ge)(ge)用戶(hu)(hu)的(de)(de)信(xin)(xin)道矢(shi)(shi)量(liang)(liang)(liang)(liang)(liang)其(qi)余時(shi)隙(xi)獲取基(ji)(ji)(ji)(ji)站(zhan)到(dao)(dao)第(di)(di)k個(ge)(ge)用戶(hu)(hu)的(de)(de)等效信(xin)(xin)道基(ji)(ji)(ji)(ji)站(zhan)到(dao)(dao)所有(you)(you)(you)ris的(de)(de)拼接信(xin)(xin)道所有(you)(you)(you)ris到(dao)(dao)用戶(hu)(hu)的(de)(de)拼接信(xin)(xin)道基(ji)(ji)(ji)(ji)站(zhan)采(cai)用主(zhu)動波(bo)(bo)(bo)(bo)束(shu)(shu)成形(xing),為(wei)(wei)(wei)基(ji)(ji)(ji)(ji)站(zhan)主(zhu)動波(bo)(bo)(bo)(bo)束(shu)(shu)成形(xing)向量(liang)(liang)(liang)(liang)(liang),其(qi)中fg,g=1,…,g代表(biao)(biao)(biao)(biao)第(di)(di)g組(zu)(zu)用戶(hu)(hu)的(de)(de)主(zhu)動波(bo)(bo)(bo)(bo)束(shu)(shu)成形(xing)向量(liang)(liang)(liang)(liang)(liang),基(ji)(ji)(ji)(ji)站(zhan)最(zui)大發射功率為(wei)(wei)(wei)pt;傳輸框架中,每一(yi)幀下(xia)(xia)包含(han)ts個(ge)(ge)時(shi)隙(xi);該方(fang)法(fa)包括以下(xia)(xia)步驟:
2.根據權(quan)利要求1所述的(de)(de)多智能(neng)超表面輔助(zhu)多播通(tong)信系統聯合波束(shu)(shu)成(cheng)形(xing)(xing)(xing)方法,其(qi)(qi)特征在(zai)于,所述步驟(zou)s1中深度展(zhan)開的(de)(de)金字塔歸一化交替投(tou)影梯度網絡(luo)包含(han)ts個(ge)時(shi)隙(xi),第一個(ge)時(shi)隙(xi)即階(jie)段(duan)一,包含(han)io層(ceng)網絡(luo),輸(shu)入為(wei)采(cai)樣信道集合,第i層(ceng)基(ji)于前一層(ceng)網絡(luo)輸(shu)出(chu)參數計算獲取(qu)的(de)(de)智能(neng)超表面被(bei)動波束(shu)(shu)成(cheng)形(xing)(xing)(xing)向(xiang)(xiang)量θ(i-1)與基(ji)站主動波束(shu)(shu)成(cheng)形(xing)(xing)(xing)向(xiang)(xiang)量f(i-1),輸(shu)出(chu){f(i),θ(i)},第i層(ceng)包含(han)可(ke)(ke)訓練變量其(qi)(qi)余(ts-1)時(shi)隙(xi)采(cai)用(yong)階(jie)段(duan)二,其(qi)(qi)輸(shu)入是第一階(jie)段(duan)的(de)(de)輸(shu)出(chu)以及(ji)等(deng)效(xiao)信道第二階(jie)段(duan)包含(han)可(ke)(ke)訓練變量{γ},輸(shu)出(chu)更新(xin)后的(de)(de)主動波束(shu)(shu)成(cheng)形(xing)(xing)(xing)向(xiang)(xiang)量ft;所有時(shi)隙(xi)的(de)(de)平均和(he)速(su)率(lv)作為(wei)損失(shi)函數。
3.根據權(quan)利要求1所述的多(duo)智能超表面輔助(zhu)多(duo)播通信系統的聯合波束成形方(fang)法,其特征在于,所述步(bu)驟(zou)s2中階(jie)段一流程包(bao)括以下步(bu)驟(zou):
4.根據(ju)權(quan)利(li)要求3所(suo)述的多智能超表面輔助多播(bo)通信(xin)系統(tong)聯(lian)合(he)波束成(cheng)形(xing)方(fang)法,其特(te)征在于,所(suo)述步驟(zou)a5)利(li)用(yong)梯度下(xia)降法更新第i次迭代的基站主動波束成(cheng)形(xing)向量(liang)f(i),具體方(fang)法為:
5.根(gen)據權利(li)(li)要求4所(suo)(suo)述(shu)的多智能(neng)超表(biao)面(mian)輔助多播通(tong)信系統(tong)的聯合波束成(cheng)形方(fang)法,其特征在于(yu),所(suo)(suo)述(shu)訓練(lian)變量集(ji)表(biao)示(shi)為訓練(lian)共分為io段(duan),第(di)j段(duan)訓練(lian)時(shi),訓練(lian)變量集(ji)為的子集(ji)每幀的第(di)一(yi)個(ge)時(shi)隙下(xia),金(jin)字塔歸一(yi)化交替投影梯度網(wang)絡輸入(ru){gl,ul,k,hk},輸出隨后利(li)(li)用上述(shu)訓練(lian)變量集(ji),基(ji)于(yu)步(bu)驟s2階段(duan)一(yi)流(liu)程獲(huo)取與令在每幀接(jie)下(xia)來(lai)的時(shi)隙下(xia),θt=θ1,網(wang)絡輸入(ru)第(di)t個(ge)時(shi)隙的輸出γ,基(ji)于(yu)步(bu)驟s2階段(duan)二流(liu)程獲(huo)取ft;第(di)j段(duan)訓練(lian)的損失函數表(biao)示(shi)為:
6.根據(ju)權(quan)利(li)要求1所(suo)述的(de)多智(zhi)能超表面(mian)輔助多播通(tong)信(xin)系統(tong)聯合波(bo)束成形方法,其(qi)(qi)特征在(zai)于(yu),所(suo)述步(bu)驟(zou)s4的(de)使用(yong)(yong)流程(cheng)(cheng)為,在(zai)每一(yi)幀第一(yi)個時隙下,利(li)用(yong)(yong)步(bu)驟(zou)s2的(de)階(jie)段一(yi)流程(cheng)(cheng)優(you)化(hua)智(zhi)能超表面(mian)被動(dong)波(bo)束成形向量θ1、基站主動(dong)波(bo)束成形向量f1,其(qi)(qi)中,優(you)化(hua)流程(cheng)(cheng)所(suo)需(xu)的(de)參(can)數(shu)由(you)網絡輸(shu)出;在(zai)之后(hou)的(de)時隙t下,令θt=θ1,利(li)用(yong)(yong)步(bu)驟(zou)s2的(de)階(jie)段二流程(cheng)(cheng)優(you)化(hua)基站主動(dong)波(bo)束成形向量ft,其(qi)(qi)中,優(you)化(hua)流程(cheng)(cheng)所(suo)需(xu)的(de)參(can)數(shu)γ由(you)網絡輸(shu)出。