国产日韩欧美一区二区三区三州_亚洲少妇熟女av_久久久久亚洲av国产精品_波多野结衣网站一区二区_亚洲欧美色片在线91_国产亚洲精品精品国产优播av_日本一区二区三区波多野结衣 _久久国产av不卡

?

稀疏貝葉斯框架下DOA與極化參數(shù)聯(lián)合估計(jì)算法

2021-08-05 09:49徐海峰
航空兵器 2021年2期

摘要:針對(duì)傳統(tǒng)極化敏感陣列測(cè)向算法在相干信號(hào)入射條件下估計(jì)精度低、運(yùn)算復(fù)雜度大的問(wèn)題,提出一種稀疏貝葉斯學(xué)習(xí)框架下的波達(dá)方向與極化參數(shù)聯(lián)合估計(jì)算法。該算法首先將數(shù)據(jù)接收矩陣稀疏得到觀測(cè)矩陣,再利用酉變換將觀測(cè)數(shù)據(jù)矩陣從復(fù)數(shù)域轉(zhuǎn)化為實(shí)數(shù)域,并且對(duì)模型參數(shù)施加一個(gè)三層的稀疏先驗(yàn)。然后,根據(jù)變分貝葉斯理論,用得到的模型參數(shù)均值和方差構(gòu)造稀疏信號(hào)的功率譜函數(shù),通過(guò)譜峰搜索得到信號(hào)的DOA。最后,利用已估計(jì)的信號(hào)DOA和模值約束算法,獲取信號(hào)極化信息。仿真試驗(yàn)表明,本文所提算法在入射信號(hào)相干時(shí)能夠正確測(cè)向,并且具有較高的測(cè)向精度和較低的運(yùn)算復(fù)雜度。

關(guān)鍵詞: 極化敏感陣列;聯(lián)合參數(shù)估計(jì);稀疏貝葉斯學(xué)習(xí);模值約束;酉變換

中圖分類(lèi)號(hào):TJ765; TN911.7 文獻(xiàn)標(biāo)識(shí)碼: A 文章編號(hào): 1673-5048(2021)02-0113-06

0 引? 言

極化敏感陣列在雷達(dá)、通信、聲吶和生物醫(yī)學(xué)等眾多領(lǐng)域有著廣泛的應(yīng)用[1]。相比于傳統(tǒng)標(biāo)量陣列,極化敏感陣列能利用接收到的入射信號(hào)極化信息,進(jìn)一步提高系統(tǒng)的測(cè)向性能[2]?;跇O化敏感陣列進(jìn)行入射信號(hào)DOA估計(jì)時(shí),若空間信號(hào)源中存在相干信號(hào),則陣列接收數(shù)據(jù)協(xié)方差矩陣便不再滿(mǎn)秩,導(dǎo)致子空間類(lèi)算法性能大幅度下降或者直接導(dǎo)致測(cè)向失敗[3]。常用的基于極化敏感陣列的信號(hào)解相干算法可分為兩大類(lèi):空間平滑類(lèi)算法和極化平滑類(lèi)算法[4]??臻g平滑類(lèi)解相干算法往往需要陣列結(jié)構(gòu)規(guī)則,否則當(dāng)信號(hào)到達(dá)角相近時(shí),測(cè)向結(jié)果變差或者失效[5-6]。而極化平滑類(lèi)解相干算法在信號(hào)入射角相近時(shí)仍能正確估測(cè)入射信號(hào)信息,并且不受陣列空間幾何結(jié)構(gòu)影響,但該算法不能實(shí)現(xiàn)對(duì)任意相干信號(hào)完全解相干[7],解相干信號(hào)數(shù)目非常有限。以上兩類(lèi)解相干算法均需在算法中加入信源數(shù)估計(jì)和解相干等預(yù)處理操作,會(huì)導(dǎo)致算法測(cè)向精度的降低和運(yùn)算復(fù)雜度的增加。

近年來(lái),稀疏重構(gòu)理論為DOA估計(jì)開(kāi)拓了一條新的解決途徑[8-9]。與傳統(tǒng)的子空間類(lèi)算法不同,稀疏重構(gòu)類(lèi)算法不涉及信號(hào)協(xié)方差矩陣的求解,可避免由相干信號(hào)入射產(chǎn)生的秩虧損導(dǎo)致測(cè)向性能下降的問(wèn)題。稀疏重構(gòu)類(lèi)算法大致可分為貪婪追蹤

算法[10]、凸松弛類(lèi)算法[11]和貝葉斯學(xué)習(xí)算法[12]三大類(lèi)。相比于前兩類(lèi)算法,貝葉斯參數(shù)學(xué)習(xí)算法無(wú)需預(yù)知信號(hào)的稀疏度,更加靈活和簡(jiǎn)單。由于稀疏重構(gòu)類(lèi)方法天然的解相干能力,目前,已有很多學(xué)者致力于此類(lèi)DOA估計(jì)算法的研究。文獻(xiàn)[13]首先將稀疏信號(hào)重構(gòu)的觀點(diǎn)引入到DOA估計(jì)中,提出了l1-SVD算法,利用l1范數(shù)最小化的重構(gòu)算法完成DOA估計(jì),但算法運(yùn)行時(shí)間較長(zhǎng),不符合工程實(shí)用。文獻(xiàn)[14]提出了平滑重構(gòu)稀疏貝葉斯學(xué)習(xí)算法,用多測(cè)量矢量模型替換單測(cè)量矢量模型,并且對(duì)變換后的觀測(cè)矩陣進(jìn)行奇異值分解,降低了算法復(fù)雜度,但對(duì)快拍數(shù)要求較高,且陣列結(jié)構(gòu)要求為嵌套陣列。文獻(xiàn)[15]提出了基于空域平滑稀疏重構(gòu)的DOA估計(jì)算法,將空域平滑理論與壓縮感知理論相結(jié)合,提高了處理相干信號(hào)的能力,減小了計(jì)算量,但針對(duì)標(biāo)量陣列,無(wú)法估計(jì)極化參數(shù)。文獻(xiàn)[16-17]提出了一種實(shí)數(shù)域下基于稀疏貝葉斯學(xué)習(xí)的DOA估計(jì)算法,在實(shí)數(shù)和復(fù)數(shù)域下構(gòu)建的DOA估計(jì)聯(lián)合稀疏模型具有相同的基相干度,可以在保證算法測(cè)向精度的同時(shí),降低算法運(yùn)算復(fù)雜度,但僅適用于標(biāo)量陣列,無(wú)法對(duì)極化參數(shù)進(jìn)行估計(jì)。 文獻(xiàn)[18]提出的基于變分稀疏貝葉斯學(xué)習(xí)的DOA估計(jì)算法,通過(guò)最小化

Kullback-Leibler(KL)散度尋求模型參數(shù)后驗(yàn)概率分布的近似分布,實(shí)現(xiàn)DOA準(zhǔn)確估計(jì),運(yùn)算復(fù)雜度低,但僅適用于單快拍條件。

本文提出了一種稀疏貝葉斯框架下實(shí)數(shù)域DOA與極化參數(shù)聯(lián)合估計(jì)算法,在稀疏貝葉斯框架下,實(shí)現(xiàn)極化敏感陣列測(cè)向數(shù)學(xué)建模,并通過(guò)酉變換,將數(shù)據(jù)接收矩陣由復(fù)數(shù)域轉(zhuǎn)換到實(shí)數(shù)域,在保證測(cè)向精度的同時(shí),進(jìn)一步降低算法的運(yùn)算復(fù)雜度,并可實(shí)現(xiàn)對(duì)相干入射信號(hào)正確測(cè)向。

1 陣列模型

1.1 陣列接收數(shù)學(xué)模型

陣元數(shù)為M的正交偶極子對(duì)極化敏感均勻線(xiàn)陣如圖1所示,陣元間距為d。

3 算法仿真與分析

假設(shè)正交偶極子對(duì)極化敏感陣列陣元個(gè)數(shù)M=12,陣元間間距為d=λ/2,λ為入射信號(hào)的波長(zhǎng)。仿真時(shí),設(shè)置入射信號(hào)頻率f=3 GHz,信噪比10 dB,快拍數(shù)100。

3.1 DOA搜索譜峰圖

對(duì)于相干信號(hào)入射的情況,設(shè)置一組相干信號(hào)且信號(hào)個(gè)數(shù)為2,信號(hào)對(duì)應(yīng)的DOA-極化信息為(θ,γ,η)=[(6.6°,10°,45°),(23.4°,20°,55°)],衰減系數(shù)為ζ=[0.192 4+j×0.981 3,0.289 1-j×0.756 7];當(dāng)信號(hào)源中的信號(hào)兩兩之間均獨(dú)立時(shí),設(shè)置獨(dú)立信號(hào)個(gè)數(shù)為3,且信號(hào)對(duì)應(yīng)的DOA-極化信息為(θ,γ,η)=[(-6.6°,10°,25°),(15.2°,20°,55°),(57.2°,65°,33°)]。在上述兩類(lèi)信號(hào)源入射情況下,信號(hào)譜峰搜索結(jié)果如圖3所示。

由圖3可知,無(wú)論對(duì)于相干信號(hào)或獨(dú)立信號(hào),本文所提算法均能正確測(cè)向。

3.2 信噪比與快拍數(shù)對(duì)參數(shù)估計(jì)性能的影響

假設(shè)相干信號(hào)入射,信噪比設(shè)置為13 dB,其他仿真條件與3.1節(jié)一致,快拍數(shù)變化范圍為20~120,步長(zhǎng)為20。經(jīng)過(guò)200次蒙特卡洛試驗(yàn),得到DOA估計(jì)均方根誤差隨快拍數(shù)變化的曲線(xiàn)??炫臄?shù)設(shè)置為128,信噪比變化范圍2~20 dB,步長(zhǎng)為2 dB,得到DOA估計(jì)均方根誤差隨信噪比變化的曲線(xiàn)。圖4~5為本文所提的基于SBL框架的DOA與極化參數(shù)聯(lián)合估計(jì)算法(SBL算法)和在本文算法基礎(chǔ)上經(jīng)實(shí)值處理改進(jìn)后的基于SBL框架的參數(shù)估計(jì)算法(RVSBL算法),以及基于正交匹配追蹤算法(Orthogonal Matching Pursuit,OMP)[23]的DOA估計(jì)性能的對(duì)比圖。

圖4~5表明,隨著快拍數(shù)及信噪比的增加,對(duì)于方位角θ、極化輔助角γ及極化相位差η估計(jì),三種算法估計(jì)性能均越來(lái)越好。本文所提基于SBL框架參數(shù)估計(jì)算法的DOA和極化輔助角估計(jì)性能優(yōu)于OMP算法,但對(duì)于極化相位差的估計(jì)而言,三種算法估計(jì)性能差不多。在不同的快拍數(shù)及信噪比下,對(duì)于RVSBL算法及SBL算法,其參數(shù)估計(jì)結(jié)果也差不多。

3.3 角度分辨力

兩相干信號(hào)的極化信息一樣,入射角度為θ=[10°10°+Δθ],Δθ從0°開(kāi)始,以步長(zhǎng)為2°遞增。信噪比13 dB,快拍數(shù)128,經(jīng)過(guò)200次蒙特卡洛試驗(yàn),得到DOA估計(jì)均方根誤差隨角度間隔Δθ的變化曲線(xiàn)如圖6所示。

比較圖6仿真結(jié)果,本文所提算法在角度間隔大于6°時(shí),其角度估計(jì)性能高于基于OMP的極化敏感陣列參數(shù)估計(jì)算法。當(dāng)信號(hào)角度間隔小于6°時(shí),三種算法均不能正確測(cè)向。

3.4 運(yùn)算時(shí)間

為比較算法的運(yùn)算復(fù)雜度,仿真SBL和RVSBL兩種算法計(jì)算時(shí)間隨快拍數(shù)的變化情況。設(shè)置兩相干信號(hào),信號(hào)對(duì)應(yīng)的DOA-極化域信息為(θ,γ,η)=[(6.6°,10°,45°),(23.4°,20°,55°)],衰減系數(shù)為ζ=[0.192 4+j×0.981 3,0.289 1-j×0.756 7],信噪比為13 dB。計(jì)算機(jī)運(yùn)行環(huán)境:CPU 2.3 GHz,內(nèi)存8 GHz,MTLAB R2016b。單次迭代試驗(yàn)結(jié)果如表1所示。

從表1可以看出,本文所提出的SBL算法由于對(duì)模型參數(shù)施加三層稀疏先驗(yàn),促進(jìn)稀疏解,相比OMP算法,更快收斂,到達(dá)滿(mǎn)足迭代停止的條件。RVSBL算法在SBL算法基礎(chǔ)上,采用酉變換實(shí)值處理,算法運(yùn)行時(shí)間約為SBL算法的1/2,這是因?yàn)殡m然酉變換處理相當(dāng)于快拍數(shù)增加一倍,但是相比復(fù)數(shù)域運(yùn)算,實(shí)數(shù)域運(yùn)算的乘法運(yùn)算操作卻減少為原來(lái)的1/4,相比SBL算法,運(yùn)算量進(jìn)一步降低。

4 結(jié)? 論

本文提出了一種稀疏貝葉斯框架下的DOA-極化參數(shù)聯(lián)合估計(jì)算法,成功地搭建稀疏貝葉斯框架的極化敏感陣列測(cè)向模型,相比基于OMP算法的DOA與極化參數(shù)聯(lián)合參數(shù)估計(jì)算法,本文所提的基于SBL框架的參數(shù)估計(jì)算法具有更高的參數(shù)估計(jì)精度和更低的計(jì)算復(fù)雜度。RVSBL算法在SBL算法的基礎(chǔ)上,通過(guò)酉變換將數(shù)據(jù)接收矩陣由復(fù)數(shù)域轉(zhuǎn)換到實(shí)數(shù)域,進(jìn)一步降低了算法的運(yùn)算復(fù)雜度。

參考文獻(xiàn):

[1] 董延坤,葛臨東,張輝.自適應(yīng)波束形成算法的現(xiàn)狀與發(fā)展動(dòng)態(tài)[J]. 微計(jì)算機(jī)信息,2005(33): 188-189.

Dong Yankun,Ge Lindong,Zhang Hui. Study Status and Advance of Robust Adaptive Beamforming Algorithms[J]. Microcomputer Information,2005(33): 188-189.(in Chinese)

[2] 陳善繼,張銳戈,吳國(guó)慶,等. 極化敏感陣列及其應(yīng)用研究[J]. 現(xiàn)代電子技術(shù),2009,32(5): 53-56.

Chen Shanji,Zhang Ruige,Wu Guoqing,et al. Research on the Polarization Sensitive Array and Its Application[J]. Modern Electronics Technique,2009,32(5): 53-56.(in Chinese)

[3] 李會(huì)勇,劉芳,謝菊蘭,等. 一種改進(jìn)的極化敏感陣列解相干算法[J]. 電子與信息學(xué)報(bào),2014,36(11): 2628-2632.

Li Huiyong,Liu Fang,Xie Julan,et al.An Improved Decorrelation Algorithm for Polarization Sensitive Array[J]. Journal of Electronics & Information Technology,2014,36(11): 2628-2632.(in Chinese)

[4] 林智勇,陶建武. 基于矢量平滑的相關(guān)源MVDR波束成形[J]. 通信學(xué)報(bào),2013(1): 96-104.

Lin Zhiyong,Tao Jianwu.MVDR Beamforming for Correlated Signal with Vector Smoothing[J].Journal on Communications,2013(1): 96-104.( in Chinese)

[5] Iizuka Y,Ichige K. Multiple Snapshot Spatial Smoothing with Improved Effective Array Aperture for High-Resolution Direction-of-Arrival Estimation[C]∥10th International Conference on Information,Communications and Signal Processing (ICICS),2015.

[6] Sekine K,Kikuma N,Hirayama H,et al. DOA Estimation Using Spatial Smoothing with Overlapped Effective Array Aperture[C]∥Asia Pacific Microwave Conference Proceedings,2012.

[7] Rahamim D,Tabrikian J,Shavit R. Source Localization Using Vector Sensor Array in a Multipath Environment[J]. IEEE Transactions on Signal Processing,2004,52(11): 3096-3103.

[8] 田野,練秋生,徐鶴. 基于稀疏信號(hào)重構(gòu)的DOA和極化角度估計(jì)算法[J]. 電子學(xué)報(bào),2016,44(7): 1548-1554.

Tian Ye,Lian Qiusheng,Xu He. DOA and Polarization Angle Estimation Algorithm Based on Sparse Signal Reconstruction[J]. Chinese Journal of Electronics,2016,44(7): 1548-1554.(in Chinese)

[9] 李新穎,彭靜,蔣占軍,等. 基于壓縮感知的DOA估計(jì)方法研究[J]. 自動(dòng)化與儀器儀表,2014(11): 34-36.

Li Xinying,Peng Jing,Jiang Zhanjun,et al.DOA Estimation Based on Compressive Sensing[J].Automation & Instrumentation,2014(11): 34-36.(in Chinese)

[10] Wen J M,Zhou Z C,Wang J,et al.? A Sharp Condition for Exact Support Recovery with Orthogonal Matching Pursuit[J].IEEE Transactions on Signal Processing,2017,65(6): 1370-1382.

[11] 趙季紅,馬兆恬,曲樺,等. 基于加權(quán)1范數(shù)稀疏信號(hào)重建的DOA估計(jì)[J]. 北京郵電大學(xué)學(xué)報(bào),2016,39(5): 33-36.

Zhao Jihong,Ma Zhaotiao,Qu Hua,et al.DOA Estimation Based on Sparse Signal Recovery Utilizing Weighted 1 Norm[J].Journal of Beijing University of Posts and Telecommunications,2016,39(5): 33-36.(in Chinese)

[12] Si W J,Qu X G,Qu Z Y,et al. Off-Grid DOA Estimation via Real-Valued Sparse Bayesian Method in Compressed Sensing[J]. Circuits Systems & Signal Processing,2016,35(10): 3793-3809.

[13] Malioutov D,Cetin M,Willsky A S. A Sparse Signal Reconstruction Perspective for Source Localization with Sensor Arrays[J]. IEEE Transactions on Signal Processing,2005,53(8):3010-3022.

[14] 陳璐,畢大平,潘繼飛. 平滑重構(gòu)稀疏貝葉斯學(xué)習(xí)測(cè)向算法[J]. 航空學(xué)報(bào),2018,39(6): 194-204.

Chen Lu,Bi Daping,Pan Jifei. A Direction Finding Algorithm Based on Smooth Reconstruction Sparse Bayesian Learning[J]. Acta Aeronautica et Astronautica Sinica ,2018,39 (6): 194-204. (in Chinese)

[15] 蔡晶晶,宗汝,蔡輝. 基于空域平滑稀疏重構(gòu)的DOA估計(jì)算法[J]. 電子與信息學(xué)報(bào),2016,38(1): 168-173.

Cai Jingjing,Zong Ru,Cai Hui. DOA Estimation via Sparse Representation of the Smoothed Array Covariance Matrix[J].Journal of Electronics & Information Technology,2016,38(1): 168-173. (in Chinese)

[16] 高陽(yáng),陳俊麗,楊廣立. 基于酉變換和稀疏貝葉斯學(xué)習(xí)的離格DOA估計(jì)[J]. 通信學(xué)報(bào),2017,38(6): 177-182.

Gao Yang,Chen Junli,Yang Guangli. Off-Grid DOA Estimation Based on Unitary Transformation and Sparse Bayesian Learning [J]. Journal of Communications,2017,38 (6): 177-182.(in Chinese)

[17] 孫磊,王華力,許廣杰,等. 基于稀疏貝葉斯學(xué)習(xí)的高效DOA估計(jì)方法[J]. 電子與信息學(xué)報(bào),2013,35(5): 1196-1201.

Sun Lei,Wang Huali,Xu Guangjie,et al.Efficient Direction-of-Arrival Estimation via Sparse Bayesian Learning[J].Journal of Electronics & Information Technology,2013,35(5): 1196-1201.(in Chinese)

[18] 郜麗鵬,杜旭華. 基于變分稀疏貝葉斯學(xué)習(xí)的DOA估計(jì)[J]. 應(yīng)用科技,2018,45(6): 32-36.

Gao Lipeng,Du Xuhua.Direction-of-Arrival(DOA) Estimation Based on Variational Sparse Bayesian Learning [J]. Applied Science and Technology,2018,45 (6): 32-36.(in Chinese)

[19] Donoho D. Superresolution via Sparsity Constraints[J]. SIAM Journal on Mathematical Analysis,1992,23(5): 1309-1331.

[20] Grimmer J. An Introduction to Bayesian Inference via Variational Approximations[J]. Political Analysis,2011,19(1): 32-47.

[21] Seeger M W,Wipf D P. Variational Bayesian Inference Techniques[J]. IEEE Signal Processing Magazine,2010,27(6): 81-91.

[22] Yang Z,Xie L H,Zhang C S.Off-Grid Direction of Arrival Estimation Using Sparse Bayesian Inference[J]. IEEE Transactions on Signal Processing,2013,61(1): 38-43.

[23] 謝菊蘭,許欣怡,李會(huì)勇. 基于OMP算法的極化敏感陣列多參數(shù)估計(jì)[J]. 雷達(dá)科學(xué)與技術(shù),2016,14(5): 453-458.

Xie Julan,Xu Xinyi,Li Huiyong. Multi-Parameter Estimation of Polarization Sensitive Array Based on OMP Algorithm [J]. Radar Science and Technology,2016,14 (5): 453-458.(in Chinese)

Joint Estimation for DOA and Polarization Parameters in

Sparse Bayesian Framework

Xu Haifeng*

(Nanjing Research Institute of Electronics Technology,Nanjing 210039,China)

Abstract: Aiming at the problems of low precision and high computational complexity in estimating coherent signals by traditional polarization sensitive array,a joint parameter estimation algorithm based on sparse Bayesian learning framework for direction of arrival and polarization information is proposed.Firstly,the observation matrix is obtained by sparse data receiving matrix,then the observation data matrix is transformed from complex domain to real domain by unitary transformation,and a three-layer sparse prior is applied to the model parameters.Then,according to the variational Bayesian theory,the power spectrum function of sparse signal is constructed by the mean and variance of the model parameters,and the DOA of the signal is obtained by peak search. Finally,the estimated signal DOA and modulus constraint are used to obtain the polarization information. The simulation results show that the proposed algorithm can correctly locate coherent incident signals,and has higher direction finding accuracy and lower computational complexity.

Key words: polarization sensitive array;joint parameter estimation;sparse Bayesian learning;modulus constraint;unitary transformation

收稿日期:2019-05-15

作者簡(jiǎn)介:徐海峰(1977-),男,研究員,研究方向?yàn)閷拵盘?hào)檢測(cè)、極化敏感陣列信號(hào)處理。