《電子技術(shù)應(yīng)用》
您所在的位置:首頁 > 通信與網(wǎng)絡(luò) > 設(shè)計(jì)應(yīng)用 > FBMC系統(tǒng)中原型濾波器的迭代設(shè)計(jì)算法
FBMC系統(tǒng)中原型濾波器的迭代設(shè)計(jì)算法
2017年電子技術(shù)應(yīng)用第4期
穆亞起
桂林電子科技大學(xué) 信息與通信工程學(xué)院,廣西 桂林541004
摘要: 基于濾波器組的多載波系統(tǒng)的整體性能完全由一個(gè)原型濾波器決定,例如頻率選擇特性、符號間干擾和信道間干擾。為構(gòu)建近似完全重構(gòu)的多載波系統(tǒng),提出了一種迭代算法來設(shè)計(jì)原型濾波器。該算法將設(shè)計(jì)問題歸結(jié)為一個(gè)無約束的優(yōu)化問題,其目標(biāo)函數(shù)為符號間干擾、信道間干擾以及原型濾波器阻帶能量的加權(quán)和。通過推導(dǎo)目標(biāo)函數(shù)的梯度向量和海森矩陣,采用修正牛頓算法來迭代優(yōu)化原型濾波器,每次迭代更新中,原型濾波器都是通過閉合公式求解,因此算法的計(jì)算復(fù)雜度很低。仿真實(shí)驗(yàn)表明, 與已有的設(shè)計(jì)算法相比,所提出的算法提高了系統(tǒng)的整體性能。
中圖分類號: TN91
文獻(xiàn)標(biāo)識碼: A
DOI:10.16157/j.issn.0258-7998.2017.04.031
中文引用格式: 穆亞起. FBMC系統(tǒng)中原型濾波器的迭代設(shè)計(jì)算法[J].電子技術(shù)應(yīng)用,2017,43(4):121-125.
英文引用格式: Mu Yaqi. Iterative design of prototype filter for FBMC system[J].Application of Electronic Technique,2017,43(4):121-125.
Iterative design of prototype filter for FBMC system
Mu Yaqi
School of Information and Communication,Guilin University of Electronic Technology,Guilin 541004,China
Abstract: Filter bank based multicarrier(FBMC) modulation system is characterized by a single prototype filter which determines the overall performance, including the frequency selectivity, intersymbol interference(ISI), and interchannel interference(ICI). An efficient design algorithm is proposed to design the prototype filter for nearly perfect reconstruction (NPR) FBMC system. The design problem is formulated into an unconstrained optimization problem that minimizes the weighted sum of the ISI/ICI, and stopband energy. By deriving the gradient vector and Hessian matrix of the objective function of the problem, the modified Newton′s method is employed to iteratively optimize the prototype filter, which is updated with closed-form formula at each iteration, implying low computational complexity. Numerical examples and comparison with existing methods are included to demonstrate the improved performance of the proposed algorithm.
Key words : FBMC system;prototype filters;unconstrained optimization;modified Newton′s method

0 引言

    近年來,以正交頻分復(fù)用(Orthogonal Frequency Division Multiplexing, OFDM)技術(shù)為代表的多載波數(shù)據(jù)傳輸技術(shù)以其在頻譜效率、對抗多徑衰落、低的實(shí)現(xiàn)復(fù)雜度等方面的優(yōu)異性能得到了廣泛的應(yīng)用[1-2]。然而,OFDM子載波濾波器的旁瓣電平較大,高達(dá)-13 dB,難以具備良好的頻率選擇特性,OFDM系統(tǒng)通過在信號前端加入循環(huán)前綴使傳輸速率降低來克服此缺陷。另外,OFDM系統(tǒng)要求相鄰子載波之間嚴(yán)格滿足正交性。而濾波器組多載波系統(tǒng)(Filter Bank-based MultiCarrier,F(xiàn)BMC)只需通過設(shè)計(jì)良好頻率選擇特性原型濾波器即可,不需要在信號前端加入循環(huán)前綴和相鄰子載波之間的正交性,提高了信號的傳輸速率和系統(tǒng)設(shè)計(jì)靈活性[3-5]。因此,目前將要代替OFDM技術(shù)逐漸被公認(rèn)是基于濾波器組的多載波技術(shù)[6-7]。

    在FBMC技術(shù)中,發(fā)送端通過合成濾波器組來實(shí)現(xiàn)多載波調(diào)制,接收端通過分析濾波器組來實(shí)現(xiàn)多載波解調(diào)。這些濾波器組由原型濾波器經(jīng)調(diào)制得到[8-11]。其中,原型濾波器的設(shè)計(jì)是核心問題,現(xiàn)有的設(shè)計(jì)算法中,一類是將濾波器組的設(shè)計(jì)問題轉(zhuǎn)化為以結(jié)構(gòu)參數(shù)為變量的優(yōu)化問題,主要有頻率采樣的方法[12-16]、窗函數(shù)方法[17]。其中,窗函數(shù)方法是通過對理想濾波器加窗來構(gòu)造原型濾波器,從而將設(shè)計(jì)問題轉(zhuǎn)化為關(guān)于窗函數(shù)參數(shù)的優(yōu)化問題。頻率采樣方法是通過對理想濾波器頻率響應(yīng)進(jìn)行等間隔采樣,然后求逆傅里葉變換,得到原型濾波器的沖激響應(yīng)函數(shù)。這類方法設(shè)計(jì)簡單,可調(diào)節(jié)部分參數(shù),設(shè)計(jì)的原型濾波器具有閉合解,但是因設(shè)計(jì)自由度小導(dǎo)致性能受限。另一類方法是直接對原型濾波器的系數(shù)進(jìn)行優(yōu)化,其中代表性算法是基于半定規(guī)劃(SDP)的方法[18]。將原型濾波器的設(shè)計(jì)問題被歸結(jié)為一個(gè)帶約束優(yōu)化問題,從而能獲得更佳的原型濾波器。但是所歸結(jié)的優(yōu)化問題是關(guān)于濾波器系數(shù)的高度非線性優(yōu)化問題,求解較為困難。為了克服這些缺點(diǎn),文獻(xiàn)[19]采用基于?琢BB(?琢-based Branch and Bound)來極大地降低直接算法的求解規(guī)模。該算法通過對約束進(jìn)行有效近似,從而極大地降低了優(yōu)化變量個(gè)數(shù)。但是,所需優(yōu)化的變量個(gè)數(shù)通過求解SQP來確定,導(dǎo)致計(jì)算復(fù)雜度高[19]。該方法設(shè)計(jì)所得的原型濾波器具備高阻帶水平和較低的失真。但是該方法近似中舍去了大量的設(shè)計(jì)自由度,導(dǎo)致原型濾波器設(shè)計(jì)性能受限。

    本文所考慮的設(shè)計(jì)是快速優(yōu)化得到原型濾波器,根據(jù)FBMC系統(tǒng)的性能指標(biāo),將原型濾波器的設(shè)計(jì)問題歸結(jié)為一個(gè)無約束的優(yōu)化問題,其目標(biāo)函數(shù)是由FBMC系統(tǒng)的符號間干擾(ISI)、信道間干擾(ICI)和原型濾波器的阻帶能量所導(dǎo)出,運(yùn)用修正的牛頓迭代法,可以快速設(shè)計(jì)得到原型濾波器。與已有設(shè)計(jì)算法進(jìn)行仿真對比發(fā)現(xiàn),本算法具有更低的計(jì)算代價(jià),得到的FBMC系統(tǒng)有著較好系統(tǒng)性能,從而可以快速而有效地設(shè)計(jì)大規(guī)模通道的FBMC系統(tǒng)。

1 FBMC系統(tǒng)的基本結(jié)構(gòu)

    不失一般性,本文將以FBMC-OQAM(Offset Quadrature Amplitude Modulation)系統(tǒng)為例來闡述FBMC系統(tǒng)的設(shè)計(jì)問題。圖1給出了FBMC-OQAM系統(tǒng)的模型,其中ak(n),bk(n)是第k通道上第n個(gè)輸入信號的實(shí)部和虛部,則輸入信號xk(n)=ak(n)+jbk(n)。偏移正交振幅調(diào)制(OQAM)中,實(shí)部與虛部在時(shí)域T/2處同相交錯(cuò)的相互正交,其中T是傳輸信號的周期,N是綜合濾波器通道數(shù),每個(gè)通道之間的載波頻率是1/T。FBMC-OQAM的基帶輸入信號為[7]

tx5-gs1-3.gif

tx5-t1.gif

2 FBMC系統(tǒng)的設(shè)計(jì)

2.1 FBMC系統(tǒng)中性能指標(biāo)

    在系統(tǒng)中,可以通過減小ISI/ICI來提高系統(tǒng)性能,從而使得系統(tǒng)滿足近似完全重構(gòu)條件,而ISI/ICI可以被確定通過原型濾波器的設(shè)計(jì)[19]。根據(jù)文獻(xiàn)[19]表明ISI/ICI的整體水平可以被表示為:

tx5-gs4-8.gif

    將式(7)和式(8)寫成矩陣相乘的形式:

tx5-gs9-16.gif

2.2 原型濾波器的設(shè)計(jì)

    基于上述的分析,可以將原型濾波器的設(shè)計(jì)問題歸結(jié)為無約束的優(yōu)化問題,然后用修正牛頓算法進(jìn)行求解該優(yōu)化問題。在FBMC-OQAM通信系統(tǒng)中,原型濾波器需要滿足線性相位結(jié)構(gòu)[20],即:

tx5-gs17-22.gif

    另外,為使原型濾波器具有好的頻率特性,期望原型濾波器具備高的阻帶衰減,這可以通過控制其阻帶能量來達(dá)到:

tx5-gs23-29.gif

    (3)判斷||dk||2≤η(η是給定的很小的正數(shù))是否成立,若成立,終止該算法迭代,xk+1為最優(yōu)的結(jié)果;若不成立,令xk=xk+1,k=k+1并返回到步驟(2)繼續(xù)迭代。

2.3 計(jì)算復(fù)雜度分析

    本文通過修正牛頓迭代算法來設(shè)計(jì)一個(gè)綜合性能較好的FBMC-OQAM系統(tǒng),原型濾波器都是通過閉合公式求解,計(jì)算復(fù)雜度來自求解線性等式(28),主要有求矩陣B(xk)的逆,以及計(jì)算向量tx5-gs29-x1.gif和矩陣B(xk)。另外,如果搜索步長取最優(yōu)值,則計(jì)算復(fù)雜度高[20],所以式(29)采用了單位步長進(jìn)行計(jì)算。因此本文算法有較小的計(jì)算消耗。此算法適用于大規(guī)模FBMC-OQAM系統(tǒng)的設(shè)計(jì),為未來5G通信中發(fā)揮作用提供了設(shè)計(jì)算法方面的儲(chǔ)備。

3 仿真結(jié)果與分析

    對于FBMC-OQAM系統(tǒng)的評價(jià)指標(biāo),采用與文獻(xiàn)[20]相同的評價(jià)指標(biāo):最小均方誤差(MSE)和阻帶能量(SE)。設(shè)計(jì)例子的仿真程序使用的是MATLAB2010b軟件編程并運(yùn)行于Intel i3-M380主頻2.53 GHz的PC。

    例1:設(shè)計(jì)一個(gè)通道載波為N=256、原型濾波器長度L=3N-1和L=4N-1的FBMC-OQAM系統(tǒng)。此外,頻率采樣法[13]、優(yōu)化頻率采樣法[15]、窗函數(shù)法[17]和αBB算法[19],在本文設(shè)計(jì)方法中設(shè)定η=1×10-5和α=0.1,在L=3N-1和L=4N-1情況下,本文設(shè)計(jì)算法迭代3次便可達(dá)到終止條件,CPU運(yùn)行時(shí)間24 s(L=3N-1)和60 s(L=4N-1),表1給出了5種方法所得到的FBMC-OQAM系統(tǒng)性能對比,并且得到的原型濾波器的幅度響應(yīng)如圖2所示。從表1中可以看出本文設(shè)計(jì)方法提高了FBMC-OQAM系統(tǒng)整體性能,與文獻(xiàn)[19]中的αBB算法相比,本文設(shè)計(jì)方法損耗較小阻帶能量來得到更加小的MSE,并且當(dāng)L=4N-1、α=0.1時(shí)計(jì)算花費(fèi)的時(shí)間(60 s)遠(yuǎn)遠(yuǎn)小于αBB算法計(jì)算時(shí)間(1 803 s);與其他方法比較,本文方法通過控制權(quán)值,可以更加靈活地平衡系統(tǒng)ISI/ICI與阻帶能量之間的關(guān)系,從而得到整體性能更加良好的FBMC-OQAM系統(tǒng)。

tx5-b1.giftx5-t2.gif

    例2:設(shè)計(jì)一個(gè)大規(guī)模通道載波FBMC-OQAM系統(tǒng),其中:載波通道為1 024,原型濾波器長度L=3N-1,α=0.1,表2給出所得FBMC-OQAM系統(tǒng)性能,同時(shí)原型濾波器的幅度響應(yīng)如圖3所示。

tx5-b2.gif

tx5-t3.gif

4 結(jié)束語

    本文圍繞如何有效地設(shè)計(jì)FBMC-OQAM系統(tǒng)的原型濾波器問題,提出了一種基于無約束優(yōu)化的快速算法,優(yōu)化的目標(biāo)函數(shù)綜合考慮FBMC-OQAM系統(tǒng)的ISI/ICI和阻帶能量。基于推導(dǎo)出的目標(biāo)函數(shù)的梯度向量和海森矩陣,采用修正牛頓算法快速有效地求解了該優(yōu)化問題。原型濾波器的迭代更新有閉合解,該方法計(jì)算復(fù)雜度低,適用于大規(guī)模系統(tǒng)的設(shè)計(jì)。理論分析和仿真結(jié)果聯(lián)合表明,本文方法設(shè)計(jì)得到的FBMC-OQAM系統(tǒng)相比于現(xiàn)有方法有著更好的整體性能。

參考文獻(xiàn)

[1] FARHANG B B.OFDM versus filter bank multicarrier:development of broadband communication systems[J].IEEE Signal Process. Mag.2011,28(3):92-112.

[2] NEE R V,PRASAD R.OFDM for wireless multimedia communications[M].OFDM for Wireless Multimedia Communications.Artech House,2000.

[3] VAIDYANATHAN P P.Filter banks in digital communications[J].IEEE Circuits and Systems Mag,2001,1(2):4-25.

[4] SIOHAN P,SILCLET C,LACAILLE N.Analysis and design of OFDM/OQAM systems based on filter bank theory[J].IEEE Trans.Signal Processing,2002,50(5):1170-1283.

[5] AMINI P,KEMPTER R,F(xiàn)ARHANG B B.A comparison of alternative filter bank multicarrier methods for cognitive radio systems[C].Presented at the Softw.Defined Radio Tech.Conf..Orlando:FL,2006:13-16.

[6] FARHANG A,MARCHETTI N,DOYLE L E,et al.Filter bank multicarrier for massive MIMO[C].Vehicular Technology Conference.IEEE,2014:284-7.

[7] BANELLI P,BUZZI S,COLAVOLPE G,et al.Modulation formats and waveforms for 5G networks:who will be the heir of OFDM?:an overview of alternative modulation schemes for improved spectral efficiency[J].IEEE Signal Process.Mag,2014,31(6):80-93.

[8] PHOONG S M,CHANG Y B,CHEN C Y.DFT-modulated filterbank transceivers for multipath fading channels[J].IEEE Trans.Signal Process,2005,53(1):182-192.

[9] CVETKOVIC Z.Tight Weyl-Heisenberg frames in l2(Z)[J].IEEE Trans.Signal Process,1998,46(5):1256-1259.

[10] BEAULIEU F D,CHAMPAGNE B.Multicarrier modulation using perfect reconstruction DFT filter bank transceivers[C].In Proc.5th Int.Conf.Inf.,Commun.,Signal Process.,Bangkok,Thailand,2005:111-115.

[11] BEAULIEU F D.Multicarrier transceiver using DFT filter banks with perfect reconstruction[D].Thesis,McGill University,2007.

[12] BELLANGER M G.Specification and design of a prototype filter for filter bank based multicarrier transmission[C].In Proc.IEEE Int.Conf.Acoust.,Speech,Signal Process.,Salt Lake City,UT,May 7-11,2001:2417-2420.

[13] MIRABBASI S,MARTIN K.Overlapped complex-modulated transmultiplexer filters with simplified design and superior stopbands[J].IEEE Trans.Circuits Syst.II,Analog Digit.Signal Process,2003,50(8):456-469.

[14] CRUZ-ROLDAN F,HENEGHAN C,SAEZ-LANDETE J,et al.Multi-objective optimisation technique to design digital filters for modulated multi-rate systems[J].Electron.Lett,2008,44(13):827-828.

[15] VIHOLAINEN A,BELLANGER M,HUCHARD M.Prototype filter design for filter bank based multicarrier transmission[C].In Prof.17th Eur.Signal Process.Conf.,Glasgow,Scotland,2009,Aug.24-28:1359-1363.

[16] VIHOLAINEN A,BELLANGER M,HUCHARD M.Prototype filter and structure optimization[OL].[2016-10-05].Available:http://www.ict-phydyas.org/delivrables/ PHYDYAS-D5-1.pdf/view.

[17] MARTIN P,Bregovic R,MARTIN M A,et al.A generalized window approach for designing transmultiplexers[J].IEEE Trans.Circuits Syst.I,Reg.Papers,2008,55(9):2696-2706.

[18] CHEN D,QU D M,JIANG T.Novel prototype filter design for FBMC based cognitive radio systems through direct optimization of filter coefficients[C].Wireless Communications and Signal Processing,2010 International Conference on.IEEE,2010:1-6.

[19] CHEN D,QU D M,JIANG T,et al.Prototype filter optimization to minimize stopband energy with NPR constraint for filter bank multicarrier modulation systems[J].IEEE Trans.Signal Process,2013,61(1):159-169.

[20] JIANG J Z,SHUI P L,ZHANG Z J.Design of oversampled DFT modulated filter banks via modified Newton′s method[J].IET Signal Processing,2011,5(3):271-280.

[21] SUN W Y,YUAN Y X.Optimization theory and methods:nonlinear programming[M].Springer Science+Business Media,Inc.,2006:203-456.



作者信息:

穆亞起

(桂林電子科技大學(xué) 信息與通信工程學(xué)院,廣西 桂林541004)

此內(nèi)容為AET網(wǎng)站原創(chuàng),未經(jīng)授權(quán)禁止轉(zhuǎn)載。