徐國強(qiáng) 王孟 李英華 伉沛川 鮑堯 楊文翰
摘 要:應(yīng)用過零雙門限協(xié)作頻譜感知算法估計無線電頻譜的占用狀態(tài),研究了認(rèn)知用戶個數(shù)、計算所用的采樣點數(shù)、不同信噪比及判決系數(shù)對檢測概率的影響,并與傳統(tǒng)雙門限能量檢測、差分雙門限能量檢測算法的檢測性能進(jìn)行了對比研究,發(fā)現(xiàn)在信噪比為-5 dB,判決系數(shù)為0.05,虛警概率小于等于0.8時,過零雙門限檢測方法單認(rèn)知用戶檢測性能明顯優(yōu)于其他2種檢測方法;當(dāng)信噪比均為-5 dB,參與協(xié)作感知的認(rèn)知用戶個數(shù)小于6時,過零雙門限協(xié)作檢測算法性能明顯優(yōu)于其他2種算法;在虛警概率恒定的情況下,增加計算所用的采樣點數(shù)或增大判決系數(shù)能夠明顯提高檢測概率;在虛警概率恒定情況下,算法檢測概率隨著信噪比增大而增大,當(dāng)信噪比為-7 dB時,檢測概率趨近于1;在相同的檢測條件下,過零雙門限協(xié)作檢測算法的虛警概率要明顯低于差分雙門限檢測的虛警概率。結(jié)果表明:過零雙門限協(xié)作頻譜感知算法可以降低系統(tǒng)的虛警概率,有效解決噪聲不確定性問題,進(jìn)一步提高系統(tǒng)感知性能。
關(guān)鍵詞:認(rèn)知無線電;頻譜感知;過零雙門限協(xié)作頻譜感知;過零檢測;噪聲不確定性
中圖分類號:TN 92
文獻(xiàn)標(biāo)志碼:ADOI:10.13800/j.cnki.xakjdxxb.2019.0121文章編號:1672-9315(2019)01-0145-07
Zero?crossing dual?threshold cooperative
spectrum sensing algorithm
XU Guo?qiang1,WANG Meng1,
LI Ying?hua2,
KANG Pei?chuan2,BAO Yao2,YANG Wen?han2
(1.Urumqi Station,The State Radio Monitoring Center,Urumqi 830054,China;
2.The State Radio Monitoring Center,Beijing 100037,China)
Abstract:The occupation state was studied in radio spectrum by zero?crossing dual?threshold cooperative spectrum sensing algorithm.The effect of the number of cognitive users,the number of sampling points used in the calculation,the different signal?to?noise ratio,and the decision coefficient on the probability of detection was studied.The detection performance of the algorithm was compared with the conventional dual?threshold energy detection algorithm and dual?threshold spectrum sensing algorithm based on differential energy detection.It is found that when the signal?to?noise ratio is -5dB,the decision coefficient is 0.05,and the false alarm probability is less than or equal to 0.8,the single?user detection performance of the zero?crossing dual?threshold detection method is obviously better than that ofother two detection algorithms.When the signal?to?noise ratio is -5 dB and the number of cognitive users participating in cooperative sensing is less than 6,the performance of the zero?crossing dual?threshold detection algorithm is obviously better than that of other two algorithms.When the false alarm probability is constant,increasing the number of sampling points used in the calculation or increasing the decision coefficient can significantly increase the detection probability.When the false alarm probability is constant,the detection probability of the algorithm increases with the increase of SNR;when the SNR is -7 dB,the detection probability approaches 1.Under the same detection conditions,the false alarm probability of the zero?crossing dual?threshold cooperative detection algorithm is obviously lower than the false alarm probability of the dual?threshold spectrum sensing algorithm based on differential energy detection.The results indicate that zero?crossing dual?threshold cooperative spectrum sensing algorithm can reduce the false alarm probability of the system,effectively solve the problem of noise uncertainty,and further improve the system’s perceived performance.
Key words:cognitive radio;spectrum sensing;zero?crossing dual?threshold cooperative spectrum sensing;zero?crossings;noise uncertainty
0?引?言
無線電頻譜資源通常由國家無線電管理部門進(jìn)行分配和管理,但研究表明在當(dāng)前頻譜管理政策下,授權(quán)的頻譜資源利用率不高,頻譜資源的浪費(fèi)現(xiàn)象較普遍[1-3],基于此背景,提出了認(rèn)知無線電(Cognitive Radio,CR)技術(shù)。在不干擾授權(quán)用戶的正常業(yè)務(wù)情況下,非授權(quán)用戶(也稱認(rèn)知用戶)可擇機(jī)使用已分配的頻譜資源,以提高資源利用率[4]。CR能夠?qū)崿F(xiàn)動態(tài)使用空閑頻譜,是提高頻譜利用率的重要技術(shù)手段,已引起各國專家的廣泛研究[5-6]。為此,電氣和電子工程師協(xié)會專門成立了IEEE 802.2工作組,研究在廣播電視頻段實現(xiàn)頻譜共享的感知無線網(wǎng)絡(luò)接入技術(shù)。CR用戶能實時檢測允許共享頻譜的使用情況,當(dāng)檢測到授權(quán)用戶當(dāng)前未使用該段頻譜時,CR用戶可接入并使用該空閑頻段,發(fā)現(xiàn)授權(quán)用戶接入時,CR用戶必須及時停止使用該段頻譜。CR關(guān)鍵技術(shù)之一是實時頻譜檢測,當(dāng)前研究的頻譜檢測技術(shù)有能量檢測(Energy Detector,ED)、匹配濾波器檢測(Matched Filtering,MF)和循環(huán)平穩(wěn)特征檢測(Cyclostationarity?based Sensing,CS)[14-15]等。其中ED實現(xiàn)簡單、計算量小,它利用能量檢測值與設(shè)定門限比較來確定授權(quán)用戶信號是否存在,但噪聲不確定性對ED影響嚴(yán)重[16]。MF檢測性能較好,但要預(yù)先確定授權(quán)用戶的完整信息。CS檢測性能最優(yōu),但算法復(fù)雜。雙門限能量檢測方法是根據(jù)噪聲不確定性設(shè)定上下2個門限,當(dāng)檢測到的信號能量低于門限下值時,判定頻譜空閑,可動態(tài)接入頻譜,當(dāng)檢測到的信號能量高于門限上值,判定授權(quán)用戶正在使用該頻譜,不可接入。當(dāng)檢測到的信號能量處于2門限之間的時,不作判決。由此可有效降低噪聲不確定性的影響,但在多認(rèn)知用戶協(xié)作感知時,可能所有參與感知的用戶檢測到的信號能量都處于雙門限之間時,產(chǎn)生感知失敗問題。文獻(xiàn)[17]為了提高ED檢測概率,提出基于能量檢測的雙門限協(xié)作感知技術(shù),根據(jù)最大合并比準(zhǔn)則對處于雙門限之間的能量信號進(jìn)行判決;文獻(xiàn)[18]為了解決感知失敗問題,提出基于雙門限的分層協(xié)同頻譜感知技術(shù),利用能量值軟融合算法對處于雙門限之間能量信號進(jìn)行判決;文獻(xiàn)[19]提出根據(jù)多能量檢測器和自適應(yīng)雙閾值的兩比特量化算法對處于雙門限之間的信號進(jìn)行判決來提高檢測性能;文獻(xiàn)[20]性提出基于信任度的雙門限協(xié)作頻譜感知算法,根據(jù)效率函數(shù)對處于雙門限之間的能量統(tǒng)計值進(jìn)行重判決;文獻(xiàn)[21]提出了差分雙門限協(xié)作頻譜感知算法,根據(jù)差分能量檢測方法對處于雙門限之間的信號進(jìn)行判決以提高檢測性能。
文中提出了過零雙門限協(xié)作頻譜感知算法(簡稱過零雙門限檢測),將過零檢測算法應(yīng)用于頻譜感知,結(jié)合雙門限能量檢測算法,對在雙門限能量檢測檢測過程中能量統(tǒng)計值處于雙門限之間的認(rèn)知用戶采用過零算法[22]進(jìn)行判決。仿真結(jié)果表明,與傳統(tǒng)雙門限能量檢測和文獻(xiàn)[21]提出的檢測方法相比,文中所提算法提高了檢測準(zhǔn)確度,更加完善了系統(tǒng)的頻譜感知性能。
1?能量檢測
1.1?能量檢測原理
能量檢測算法根據(jù)認(rèn)知用戶收到的信號能量統(tǒng)計值建立統(tǒng)計模型,通過統(tǒng)計模型確定判決門限。比較能量統(tǒng)計值與判決門限,統(tǒng)計量大于判決門限,則判定授權(quán)用戶存在,否則判定信道空閑。頻譜感知的假設(shè)模型可由式(1)表示
式中?S(n)為授權(quán)用戶信號;W(n)為信道噪聲;H為信道增益;Y(n)為認(rèn)知用戶接收到的信號;H1為授權(quán)用戶存在;H0為信道空閑。設(shè)N為每次檢測采樣點數(shù),能量檢測中的能量統(tǒng)計值G為
G=1Nni=1[Y(n)]2(2)
假設(shè)門限值為λ,則能量檢測的二元假設(shè)模型為
H0:G≤λ
H1:G>λ(3)
假設(shè)信道噪聲為高斯白噪聲,當(dāng)N較大時,統(tǒng)計能量值G近似服從式(4)所示的高斯分布[23]
H0:G0~Nσ2w,2Nσ4w
H1:G1~Nσ2s+σ2w,2N
(σ2s+σ2w)2
式中?σ2w為噪聲方差;σ2s為授權(quán)用戶信號平均功率,W.由式(4)可計算出虛警概率Pf與對應(yīng)的檢測概率Pd分別為
式中?Q(·)為標(biāo)準(zhǔn)高斯互補(bǔ)累積分布函數(shù);Pd為授權(quán)用戶存在時,判定為H1的概率;Pf為信道空閑時,判定為H1的概率。假設(shè)虛警概率Pf恒定不變,可通過式(7)確定判決門限λ為
1.2?過零檢測算法
過零檢測算法是用時間序列理論對接收到的信號進(jìn)行處理,構(gòu)建統(tǒng)計模型,計算統(tǒng)計量Φ。文獻(xiàn)[24]研究結(jié)果表明,當(dāng)信道中只有高斯白噪聲時,統(tǒng)計量Φ近似服從非中心卡方分布,即Φ~X23(11)。該分布自由度為3,非中心參數(shù)為11.由此,可將頻譜感知轉(zhuǎn)化為擬合優(yōu)度檢測問題,當(dāng)信道內(nèi)無授權(quán)用戶時,Φ應(yīng)服從上述分布,否則Φ將偏離上述分布。
假設(shè)一個認(rèn)知用戶有N個采樣點Y(i),
i∈N={,…,N}。假定所有Y(i)都是實數(shù)值[25]。設(shè)n為Y(i)的第n階差分器,定義如下
當(dāng)n-1Y(i)的符號與n-1Y(i-1)的符號不同時,產(chǎn)生一個過零點。設(shè)Dn為第n階過零點總數(shù)。這里定義0Yi=Yi,并定義
Δj和μj分別如下
式中?E(·)為期望函數(shù)。當(dāng)N足夠大時,0 ≤Δj≤(N-1),
nj=1=N-1.通過文獻(xiàn)[14],可定義Φ為
Φ=nj=1(Δj-μj)2μj(11)
根據(jù)Φ值可建立判決器如式(12)
式中判決門限τ可通過P{Φ>τ|H0}≤α確定,α為判決系數(shù)。如果Y(i)~N(0,σ2w),當(dāng)N足夠大時,E(Dj)可通過式(13)計算得出
因此,μj可通過式(13)、式(9)和式(10)計算出。且有相關(guān)實驗表明,對于大多數(shù)情況,計算和使用Dj時,當(dāng)j≥9時,過零檢測算法性能不會有更多改善。
2?過零雙門限協(xié)作頻譜感知算法
噪聲不確定性會嚴(yán)重降低ED算法性能,雙門限能量檢測可解決噪聲不確定性問題,提高判決結(jié)果的可靠性。雙門限能量檢測根據(jù)噪聲不確定性確定判決門限λ1,λ2(λ1<λ2),然后將能量統(tǒng)計值G與判決門限進(jìn)行比較,若G≥λ2,則判決授權(quán)用戶存在;若G<λ1,則判決信道空閑;若λ1 文中利用過零雙門限方法進(jìn)行協(xié)作檢測時,使用“或”準(zhǔn)則進(jìn)行融合判決。第k個認(rèn)知用戶的檢測概率Pk,d和虛警概率Pk,f為 Qf=1-Kk=1(1-Pk,f)(17) 式中?K為認(rèn)知用戶個數(shù)。 過零雙門限協(xié)作算法檢測方法如下 步驟1:根據(jù)公式(2),計算出認(rèn)知用戶接收到的能量值G; 步驟2:根據(jù)噪聲不確定性計算判決門限λ1,λ2(λ1<λ2),若G≥λ2,則判定檢測到授權(quán)用戶;若G<λ1,則表明信道空閑;若λ1≤G<λ2,此時通過能量值G無法判斷授權(quán)用戶是否存在,采用過零檢測算法繼續(xù)對信道進(jìn)行判決,執(zhí)行步驟3; 步驟3:設(shè)n=9,利用公式(8)計算采樣值{Yi}的一階過零點總數(shù)和高階過零點總數(shù)。利用公式(13)計算一階和高階過零點總數(shù)的期望值。 步驟4:利用式(11)計算統(tǒng)計量Φ. 步驟5:給定過零檢測算法的判決系數(shù)α,并計算該判決系數(shù)下的判決門限τ. 步驟6:根據(jù)過零檢測算法,如果Φ≤τ,則表明信道空閑,否則表明授權(quán)用戶存在。 步驟7:所有認(rèn)知用戶將本地判決結(jié)果發(fā)送到判決融合中心,融合中心采用“或”準(zhǔn)則做出最終判決。 3?算法仿真分析 本節(jié)采用蒙特卡羅方法對文中提出的檢測算法進(jìn)行MATLAB仿真分析。假設(shè)授權(quán)用戶信號調(diào)制方式為BPSK,噪聲為加性高斯白噪聲且是獨(dú)立同分布的,仿真次數(shù)為5 000. 首先將文中提出的過零雙門限檢測方法與傳統(tǒng)雙門限能量檢測和差分雙門限檢測進(jìn)行檢測性能對比。圖2是信噪比SNR=-5 dB,α=0.05時,單認(rèn)知用戶檢測性能比較仿真結(jié)果。從圖2可以看出,當(dāng)虛警概率Pf ≤ 0.8時,過零雙門限檢測方法檢測性能明顯優(yōu)于其他2種檢測方法。當(dāng)Pf≥0.8時,Pd接近于1. 圖3是過零雙門限檢測與傳統(tǒng)雙門限能量檢測和差分雙門限檢測的協(xié)作感知性能比較。協(xié)作認(rèn)知用戶數(shù)為3,信噪比分別為-5,-8,-10 dB.通過比較可以看出,當(dāng)參與協(xié)作的認(rèn)知用戶信噪比不同時,傳統(tǒng)雙門限能量檢測和差分雙門限檢測協(xié)作感知性能提高不明顯,但過零雙門限檢測協(xié)作感知性能有明顯提升。當(dāng)Pf = 0.1,協(xié)作檢測概率Qd約為0.83,比單節(jié)點檢測性能提高3.7%.當(dāng)Pf≥0.7時,Qd趨近于1. 圖4是過零雙門限協(xié)作檢測與傳統(tǒng)雙門限能量協(xié)作檢測和差分雙門限協(xié)作檢測在相同信噪比情況下的感知性能比較。協(xié)作認(rèn)知用戶數(shù)分別為3,4,5,6,7,8,9,10,信噪比為-5 dB.通過仿真可以看出,3種算法的檢測概率隨著參與協(xié)作感知的認(rèn)知用戶數(shù)增加而增加。通過比較可以看出,信噪比相同時,當(dāng)參與協(xié)作感知的認(rèn)知用戶個數(shù)Num小于6時,過零雙門限協(xié)作檢測算法性能明顯優(yōu)于其他2種算法。當(dāng)Num大于等于6時,Qd趨近于1. 圖5是SNR=-5 dB,α=0.05時,不同取樣點數(shù)N對檢測概率的影響。從中可以看出,在某一恒定虛警概率下,增大計算所用的采樣點數(shù)時,能夠提高檢測概率。 圖6是SNR=-5 dB,N=1 000時,不同α值對檢測概率的影響。從中可知,虛警概率恒定時,α增大時可提高檢測概率。但如果α增大,在H0情況下,判決授權(quán)用戶存在的概率會增大,從而導(dǎo)致利用空閑頻譜的機(jī)率降低。因此,要合理設(shè)置α的大小。 圖7是N=1 000,α=0.05,Pf分別為0.01,0.03,0.05,0.08,0.10時,信噪比對Pd的影響情況。從中可知,當(dāng)Pf恒定時,Pd隨著信噪比增大而增大。當(dāng)信噪比為-7 dB時,Pd趨近于1.當(dāng)信噪比恒定時,Pf增大也會導(dǎo)致Pd增大。這是因為Pf增大會降低判決門限。 圖8是過零雙門限檢測與差分雙門限檢測在給定的判決門限下,虛警概率性能比較。N=1 000,判決門限為3組,分別為λ1=1.06,λ2=1.11;λ1=1.05,λ2=1.10和λ1=1.04,λ2=1.09.由圖7可知,判決門限增大會降低虛警概率。信噪比對虛警概率影響不明顯。當(dāng)λ1=1.06,λ2=1.11,SNR=-10 dB時,過零雙門限檢測的虛警概率為0.014,差分雙門限檢測的虛警概率為0.079.可見,在相同條件下,過零雙門限檢測的虛警概率要明顯低于差分雙門限檢測的虛警概率。 4?結(jié)?論 針對頻譜感知中噪聲不確定性降低頻譜感知性能的問題,本文在雙門限能量檢測算法的基礎(chǔ)上,通過引入過零檢測算法,有效避免了雙門限能量檢測算法中感知失敗問題,確實降低了噪聲不確定性對頻譜感知的影響。 參考文獻(xiàn)(References): [1] Federal Communication Commission.Spectrum policy task force report[R].ET Docket No.02-155,2002. [2]Yang J.Spatial channel characterization for cognitive radios[D].Berkeley:University of California,2004. [3]ZHAO Qing,Sadler B M.A survey on dynamic spectrum access:signal processing,networking,and regulatory policy[J].IEEE Signal Processing Magazine,2007,24(3):79-89. [4]Stotas S,Nallanathan A.On the throughput and spectrum Sensing enhancement of opportunistic spectrum access cognitive radio networks[J].IEEE Transactions on Wireless Communications,2012,11(1):97-107. [5]Mitola J,Maguire G Q.Cognitive radio:Making software radios more personal[J].IEEE Personal Communications,1999,6(4):13-18. [6]Sofotasios P C,Rebeiz E,Zhang L,et al.Energy detection based spectrum sensing over and extreme fading channels[J].IEEE Transactions on Vehicular Technology,2013,62(3):1031-1040. [7]Yucek T,Arslan H.A survey of spectrum sensing algorithms for cognitive radio applications[J].IEEE Communications Surveys and Tutorials,2009,11(1):116-130. [8]黃?川,梁步林,許?力,等.認(rèn)知無線電網(wǎng)絡(luò)中基于動態(tài)信譽(yù)機(jī)制的頻譜分配策略研究[J].信號處理,2014(11):1339-1344. HUANG Chuan,LIANG Bu?lin,XU Li,et al.Research on dynamic reputation mechanism based spectrum allocation strategy in cognitive radio network[J].Signal Processing,2014(11):1339-1344. [9]彌?寅,盧光躍.基于特征值極限分布的合作頻譜感知算法[J].通信學(xué)報,2015,36(1):84-89. MI Yin,LU Guang?yue.Cooperative spectrum sensing algorithm based on limiting eigenvalue distribution[J].Journal on Communications,2015,36(1):84-89. [10]陳長興,符?輝,牛德智,等.基于雙門限能量檢測的協(xié)作頻譜感知算法[J].系統(tǒng)工程與電子技術(shù),2013,35(8):1742-1746. CHEN Chang?xing,F(xiàn)U Hui,NIU De?zhi,et al.Cooperative spectrum sensing algorithm based on double?threshold energy detection[J].Systems Engineering and Electronics,2013,35(8):1742-1746. [11]申?濱,王?舒,黃?瓊,等.基于Gerschgorin圓盤理論的認(rèn)知無線電寬帶頻譜感知[J].通信學(xué)報,2014(4):1-10.SHEN Bin,WANG Shu,HUANG Qiong,et al.Gerschgorin disk theorem based spectrum sensing for wideband cognitive radio[J].Journal on Communications,2014(4):1-10. [12]賈?瓊,李兵兵.基于局部方差的MIMO頻譜感知算法研究[J].電子與信息學(xué)報,2015,37(7):1525-1530. JIA Qiong,LI Bing?bing.MIMO spectrum sensing method based on the local variance[J].Journal of Electronics and Information Technology,2015,37(7):1525-1530. [13]張?瑩,滕?偉,韓維佳,等.認(rèn)知無線電頻譜感知技術(shù)綜述[J].無線電通信技術(shù),2015,41(3):12-16. ZHANG Ying,TENG Wei,HAN Wei?jia,et al.Review of spectrum sensing techniques in cognitive radio networks[J].Radio Communications Technology,2015,41(3):12-16. [14]孫夢巍,趙?龍,許巧春,等.動態(tài)時變衰落信道下的頻譜感知算法[J].通信學(xué)報,2014,35(7):63-69.SUN Meng?wei,ZHAO Long,XU Qiao?chun,et al.New spectrum sensing method under time?variant flat fading channels[J].Journal on Communications,2014,35(7):63-69. [15]盧光躍,葉迎暉,孫?宇,等.克服噪聲不確定性的擬合優(yōu)度檢驗頻譜感知算法[J].電訊技術(shù),2016,56(1):26-31.LU Guang?yue,YE Ying?hui,SUN Yu,et al.Anovel anti?noise?uncertainty spectrum sensing method using goodness of fit test[J].Telecommunication Engineering,2016,56(1):26-31. [16]王海蓮,張士兵,郭莉莉.基于可信度的加權(quán)協(xié)作頻譜檢測算法[J].數(shù)據(jù)采集與處理,2014,29(3):472-477.WANG Hai?lian,ZHANG Shi?bing,GUO Li?li.Reliability?based weighted cooperative spectrum sensing algorithm[J].Journal of Data Acquisition and Processing,2014,29(3):472-477. [17]劉會衡,陳?偉,劉克中.一種基于能量檢測的雙門限協(xié)作頻譜感知技術(shù)[J].武漢理工大學(xué)學(xué)報,2011(4):144-148. LIU Hui?heng,CHEN Wei,LIU Ke?zhong.A dual?threshold cooperative spectrum sensing technology based on energy detection[J].Journal of Wuhan University of Technology,2011(4):144-148. [18]Liu S Q,Hu B J,Wang X Y.Hierarchical cooperative spectrum sensing based on dual thresholds energy detection[J].Communications Letters IEEE,2012,16(7):1096-1099. [19]Detection M F,Detection E.Two?stage detectors with multiple energy detectors and adaptive dual threshold in cognitive radio networks[J].International Journal of Distributed Sensor Networks,2013,2013(9):172-180. [20]張學(xué)軍,魯?友,田?峰,等.基于信任度的雙門限協(xié)作頻譜感知算法[J].物理學(xué)報,2014,63(7):78401. ZHANG Xue?jun,LU You,TIAN Feng,et al.Dual?threshold cooperative spectrum sensing for cognitive radio based on trust[J].Acta Physica Sinica,2014,63(7):78401. [21]張學(xué)軍,嚴(yán)金童,田?峰,等.基于差分能量檢測的雙門限協(xié)作頻譜感知算法[J].儀器儀表學(xué)報,2014,35(6):1325-1330. ZHANG Xu?jun,YAN Jin?tong,TIAN Feng,et al.Dual?threshold cooperative spectrum sensing algorithm based on differential energy detection[J]. Chinese Journal of Scientific Instrument ,2014,35(6):1325-1330. [22]Kedem B.Spectral analysis and discrimination by zero?crossings[J].Proceedings of the IEEE,1986,74(11):1477-1493. [23]ZHANG L L, HUANG J G,Tang C K.Novel energy detection scheme in cognitive radio[C]//IEEE International Conference on Signal Processing,Communications and Computing.IEEE,2011:1-4. [24]Welch B L.On linear combinations of several variances[J].Journal of the American Statistical Association,1956,51(273):132-148. [25] SHEN Lei,WANG H,ZHANG W,et al.Blind spectrum sensing for cognitive radio channels with noise uncertainty[J].IEEE Transactions on Wireless Communications,2011,10(6):1721-1724.