基于雙卷積自編碼器的自適應(yīng)波束形成

蔣伊琳; 李帥; 鄭沛; 唐元博

doi:10.11999/JEIT240486

基于雙卷積自編碼器的自適應(yīng)波束形成

doi: 10.11999/JEIT240486 cstr: 32379.14.JEIT240486

蔣伊琳^{1, 2},
李帥^1, ,,
鄭沛³,
唐元博¹

1.
哈爾濱工程大學(xué) 哈爾濱 150001
2.
航空工業(yè)電磁頻譜協(xié)同探測(cè)與智能認(rèn)知聯(lián)合技術(shù)中心哈爾濱 150001
3.
試驗(yàn)物理與計(jì)算數(shù)學(xué)國(guó)家級(jí)重點(diǎn)實(shí)驗(yàn)室北京 100876

基金項(xiàng)目: 國(guó)防科技基礎(chǔ)加強(qiáng)計(jì)劃 (2019-JCJQ-ZD-067-00)

詳細(xì)信息

作者簡(jiǎn)介:
蔣伊琳：男，博士，副教授，研究方向?yàn)樯疃葘W(xué)習(xí)、信號(hào)處理

李帥：男，碩士生，研究方向?yàn)樯疃葘W(xué)習(xí)、信號(hào)處理

鄭沛：男，高級(jí)工程師，研究方向?yàn)殡姶蓬l譜感知

唐元博：男，碩士生，研究方向?yàn)樯疃葘W(xué)習(xí)、信號(hào)處理

通訊作者:
李帥　lishaui@hrbeu.edu.cn

中圖分類號(hào): TN911.7
計(jì)量
- 文章訪問(wèn)數(shù): 237
- HTML全文瀏覽量: 94
- PDF下載量: 54
- 被引次數(shù): 0
出版歷程
- 收稿日期: 2024-06-14
- 修回日期: 2025-02-11
- 網(wǎng)絡(luò)出版日期: 2025-02-19
- 刊出日期: 2025-02-28

Adaptive Beamforming Based on Dual Convolutional Autoencoder

JIANG Yilin^{1, 2},
LI Shuai^{1
, ,},
ZHENG Pei³,
TANG Yuanbo¹

1.
College of Information and Communication Engineering, Harbin Engineering University, Harbin 150001, China
2.
AVIC United Technology Center for Electromagnetic Spectrum Collaborative Detection and Intelligent cognition, Harbin 150001, China
3.
State Key Laboratory of Experimental Physics and Computational Mathematics, Beijing 100876, China

Funds: The National Defense Science and Technology Strengthening Program (2019-JCJQ-ZD-067-00)

摘要

摘要: 在低信噪比環(huán)境下，陣列天線獲取空域信號(hào)的來(lái)波方向極其困難，導(dǎo)致一般的波束形成方法無(wú)法準(zhǔn)確形成正對(duì)入射信號(hào)的波束。針對(duì)上述問(wèn)題，該文提出了一種基于雙卷積自編碼器的盲接收自適應(yīng)波束形成(Dual Convolutional AutoEncoder-Adaptive Beamforming, DCAE-ABF)方法，該方法在基于大量空域統(tǒng)計(jì)信息的情況下，以時(shí)域-頻域聯(lián)合條件作為約束，利用兩個(gè)獨(dú)立的卷積自編碼器(CAE)分別對(duì)陣列接收信號(hào)與輻射源信號(hào)進(jìn)行特征提取，并使用深度神經(jīng)網(wǎng)絡(luò)(DNN)將兩個(gè)CAE的特征編碼進(jìn)行連接，構(gòu)建DCAE網(wǎng)絡(luò)，實(shí)現(xiàn)在低信噪比環(huán)境下，面對(duì)未知頻率和來(lái)波方向的入射信號(hào)時(shí)，也能夠自適應(yīng)形成正對(duì)入射信號(hào)的波束，達(dá)到盲接收的效果。仿真實(shí)驗(yàn)結(jié)果表明，在低信噪比環(huán)境下，單信號(hào)與雙信號(hào)入射時(shí)所帶來(lái)的信噪比增益均高于常規(guī)波束形成(CBF)方法與基于最小均方誤差的自適應(yīng)波束形成(Minimum Mean Square Error-Adaptive BeamForming, MMSE-ABF)方法，以及基于卷積神經(jīng)網(wǎng)絡(luò)的自適應(yīng)波束形成方法(Convolutional Neural Networks- Adaptive BeamForming, CNN-ABF)，且該增益在入射信號(hào)頻率、角度變化時(shí)仍具有良好的穩(wěn)定性。
- 自適應(yīng)波束形成 /
- 卷積自編碼器 /
- 盲波束形成 /
- 信噪比增益
Abstract: Objective Most traditional beamforming techniques and adaptive beamforming methods rely on reference signals. These methods require prior knowledge of the signal frequency and Direction of Arrival (DOA) at the array for beamforming. However, in low Signal-to-Noise Ratio (SNR) environments, obtaining the frequency and DOA of the incident signals is extremely challenging. This difficulty leads to significant performance degradation in reference-signal-based beamforming, limiting its applicability in tasks such as electronic reconnaissance and electronic countermeasures in low SNR conditions. This paper addresses the challenge of enabling antenna arrays to perform adaptive beamforming for incident signals with unknown frequencies and DOAs in low-SNR environments. Methods This paper proposes a Dual Convolutional AutoEncoder-Adaptive Beamforming (DCAE-ABF) method for blind reception. The approach leverages dual Convolutional Autoencoders (CAEs) to extract features from both the array-received signal and the radiation source signal, utilizing extensive air-domain statistical information with joint time-frequency domain constraints. A Deep Neural Network (DNN) connects the feature encodings from the two CAEs to construct the DCAE network. This method enables adaptive beamforming in low SNR environments, even when the incident signal’s frequency and DOA are unknown, facilitating blind reception. Results and Discussions Simulation results demonstrate that the proposed DCAE-ABF method can rapidly and accurately adjust the beam direction for incident signals with unknown frequencies and directions of arrival in a low SNR environment, effectively orienting the beam towards the incident signals for optimal reception. This method improves the output signal’s SNR, with the SNR gain significantly exceeding that of traditional beamforming techniques (Fig. 4, Fig. 6). Furthermore, the SNR gain achieved by this method remains stable even when the frequency and angle of the incident signal vary (Fig. 5). Conclusions This paper presents an adaptive beamforming method based on dual convolutional autoencoders. The method outperforms the other three approaches discussed in this study when applied to incident signals with unknown directions of arrival in low SNR environments. Even when the DOA is unknown, the method effectively utilizes the spatial information accumulated during autoencoder training. It can extract features from the array signals and adaptively form beams directed at the incident signals, achieving optimal reception. This approach enables blind adaptive beamforming for signals with unknown frequencies and directions of arrival, significantly improving the SNR of the incident signals.
- Adaptive beamforming /
- Convolutional autoencoder /
- Blind beamforming /
- Signal-to-noise ratio gain

HTML全文

圖 1 信號(hào)接收模型示意圖

下載: 全尺寸圖片幻燈片

圖 2 DCAE-ABF示意圖

下載: 全尺寸圖片幻燈片

圖 3 DCAE網(wǎng)絡(luò)結(jié)構(gòu)圖

下載: 全尺寸圖片幻燈片

圖 4 4種方法處理后的單信號(hào)頻譜對(duì)比

下載: 全尺寸圖片幻燈片

圖 5 4種方法處理后信噪比變化圖

下載: 全尺寸圖片幻燈片

圖 6 4種方法處理后的雙信號(hào)頻譜對(duì)比

下載: 全尺寸圖片幻燈片

表 1 超參數(shù)對(duì)輸入測(cè)試結(jié)果帶來(lái)的部分影響

自編碼器層數(shù)	訓(xùn)練時(shí)間	CDAE 學(xué)習(xí)率	DNN 學(xué)習(xí)率	CAE 學(xué)習(xí)率	信噪比增益(dB)
3層	適中	0.00007	0.0003	0.00003	20.9
		0.00008	0.0004	0.00004	21.08
		0.00009	0.0005	0.00005	21.56
		0.0001	0.0006	0.00006	21.13
4層	較長(zhǎng)	0.00007	0.0003	0.00003	16.12
		0.00008	0.0004	0.00004	16.13
		0.00009	0.0005	0.00005	16.17
		0.0001	0.0006	0.00006	16.68

下載: 導(dǎo)出CSV

表 2 DCAE網(wǎng)絡(luò)訓(xùn)練過(guò)程中的超參數(shù)

神經(jīng) 網(wǎng)絡(luò)	初始學(xué)習(xí)率	學(xué)習(xí)率衰減率	網(wǎng)絡(luò)層數(shù)
CDAE	0.00009	0.95	3層
DNN	0.0005	–	3層
CAE	0.00005	–	3層

下載: 導(dǎo)出CSV

表 3 DCAE網(wǎng)絡(luò)的特征結(jié)構(gòu)

		輸入形狀	輸出形狀
陣列接收信號(hào)自編碼器網(wǎng)絡(luò)的解碼器部分	輸入層	1×3200	–
	卷積層1	1×3200	1×3200×64
	池化層1	1×3200×64	1×800×64
	卷積層2	1×800×64	1×800×64
	池化層2	1×800×64	1×400×64
	卷積層3	1×400×64	1×400×32
	池化層3	1×400×32	1×200×32
DNN連接部分	數(shù)據(jù)展平	1×200×32	1×6400
	全連接層1	1×6400	1×4800
	全連接層2	1×4800	1×3200
	全連接層3	1×3200	1×1600
	數(shù)據(jù)重構(gòu)	1×1600	1×50×32
輻射源原始信號(hào)自編碼器網(wǎng)絡(luò)的解碼器部分	反池化層1	1×50×32	1×100×32
	反卷積層1	1×100×32	1×100×64
	反池化層2	1×100×64	1×200×64
	反卷積層2	1×200×64	1×200×64
	反池化層3	1×200×64	1×400×64
	反卷積層3	1×400×64	1×400
	輸出層	1×400	1×400

下載: 導(dǎo)出CSV

參考文獻(xiàn)(18)

[1]	唐敏, 齊棟, 劉成城, 等. 基于多級(jí)阻塞的穩(wěn)健相干自適應(yīng)波束形成[J]. 電子與信息學(xué)報(bào), 2019, 41(7): 1705–1711. doi: 10.11999/JEIT180332. TANG Min, QI Dong, LIU Chengcheng, et al. New adaptive beamformer for coherent interference based on multistage blocking[J]. Journal of Electronics & Information Technology, 2019, 41(7): 1705–1711. doi: 10.11999/JEIT180332.
[2]	沈季, 萬(wàn)顯榮, 易建新, 等. 復(fù)雜干擾場(chǎng)景下的穩(wěn)健自適應(yīng)波束形成[J]. 系統(tǒng)工程與電子技術(shù), 2023, 45(4): 941–949. doi: 10.12305/j.issn.1001-506X.2023.04.01. SHEN Ji, WAN Xianrong, YI Jianxin, et al. Robust adaptive beamforming in complex interference scenarios[J]. Systems Engineering and Electronics, 2023, 45(4): 941–949. doi: 10.12305/j.issn.1001-506X.2023.04.01.
[3]	WU Xun, ZHANG Shurui, MA Xiaofeng, et al. The adaptive wideband beamforming using convolutional neural network[C]. Proceedings of 2022 International Conference on Microwave and Millimeter Wave Technology, Harbin, China, 2022: 1–3. doi: 10.1109/ICMMT55580.2022.10022889.
[4]	王兆彬, 鞏朋成, 鄧薇, 等. 聯(lián)合協(xié)方差矩陣重構(gòu)和ADMM的魯棒波束形成[J]. 哈爾濱工業(yè)大學(xué)學(xué)報(bào), 2023, 55(4): 64–71. doi: 10.11918/202107104. WANG Zhaobin, GONG Pengcheng, DENG Wei, et al. Robust beamforming by joint covariance matrix reconstruction and ADMM[J]. Journal of Harbin Institute of Technology, 2023, 55(4): 64–71. doi: 10.11918/202107104.
[5]	GODARA L C. Application of antenna arrays to mobile communications. II. Beam-forming and direction-of-arrival considerations[J]. Proceedings of the IEEE, 1997, 85(8): 1195–1245. doi: 10.1109/5.622504.
[6]	ZAHARIS Z D, GRAVAS I P, LAZARIDIS P I, et al. An effective modification of conventional beamforming methods suitable for realistic linear antenna arrays[J]. IEEE Transactions on Antennas and Propagation, 2020, 68(7): 5269–5279. doi: 10.1109/TAP.2020.2977822.
[7]	SENTHILKUMAR K S, PIRAPAHARAN K, HOOLE P R P, et al. Single perceptron model for smart beam forming in array antennas[J]. International Journal of Electrical and Computer Engineering (IJECE), 2016, 6(5): 2300–2309. doi: 10.11591/ijece.v6i5.10719.
[8]	EL ZOOGHBY A H, CHRISTODOULOU C G, and GEORGIOPOULOS M. A neural network-based smart antenna for multiple source tracking[J]. IEEE Transactions on Antennas and Propagation, 2000, 48(5): 768–776. doi: 10.1109/8.855496.
[9]	RAMEZANPOUR P, REZAEI M J, and MOSAVI M R. Deep‐learning‐based beamforming for rejecting interferences[J]. IET Signal Processing, 2020, 14(7): 467–473. doi: 10.1049/iet-spr.2019.0495.
[10]	SALLOMI A H and AHMED S. Multi-layer feed forward neural network application in adaptive beamforming of smart antenna system[C]. Proceedings of 2016 Al-Sadeq International Conference on Multidisciplinary in IT and Communication Science and Applications, Baghdad, Iraq, 2016: 1–6. doi: 10.1109/AIC-MITCSA.2016.7759925.
[11]	MOHAMMADZADEH S, NASCIMENTO V H, DE LAMARE R C, et al. Robust beamforming based on complex-valued convolutional neural networks for sensor arrays[J]. IEEE Signal Processing Letters, 2022, 29: 2108–2112. doi: 10.1109/LSP.2022.3212637.
[12]	MALLIORAS I, ZAHARIS Z D, LAZARIDIS P I, et al. A novel realistic approach of adaptive beamforming based on deep neural networks[J]. IEEE Transactions on Antennas and Propagation, 2022, 70(10): 8833–8848. doi: 10.1109/TAP.2022.3168708.
[13]	ZAHARIS Z D, SKEBERIS C, XENOS T D, et al. Design of a novel antenna array beamformer using neural networks trained by modified adaptive dispersion invasive weed optimization based data[J]. IEEE Transactions on Broadcasting, 2013, 59(3): 455–460. doi: 10.1109/TBC.2013.2244793.
[14]	梁夢(mèng)薇, 何勁, 舒汀, 等. 陣元位置互質(zhì)的線性陣列: 陣列校正和波束形成[J]. 電子與信息學(xué)報(bào), 2024, 46(1): 240–248. doi: 10.11999/JEIT221539. LIANG Mengwei, HE Jin, SHU Ting, et al. Linear coprime sensor location arrays: Array calibration and beamforming[J]. Journal of Electronics & Information Technology, 2024, 46(1): 240–248. doi: 10.11999/JEIT221539.
[15]	陳鵬, 景曉簪, 陳洋, 等. 穩(wěn)健的特征空間基變換自適應(yīng)波束形成[J]. 哈爾濱工業(yè)大學(xué)學(xué)報(bào), 2023, 55(5): 71–77,113. doi: 10.11918/202112102. CHEN Peng, JING Xiaozan, CHEN Yang, et al. Robust eigenspace bases transition technique for adaptive beamforming[J]. Journal of Harbin Institute of Technology, 2023, 55(5): 71–77,113. doi: 10.11918/202112102.
[16]	ZHU Yi, WU Xindong, QIANG Jipeng, et al. Representation learning via an integrated autoencoder for unsupervised domain adaptation[J]. Frontiers of Computer Science, 2023, 17(5): 175334. doi: 10.1007/s11704-022-1349-5.
[17]	LU Cai, MU Zuochen, ZONG Jingjing, et al. Unsupervised VSP up-and downgoing wavefield separation via dual convolutional autoencoders[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 5900315. doi: 10.1109/TGRS.2023.3334309.
[18]	AL KASSIR H, ZAHARIS Z D, LAZARIDIS P I, et al. A review of the state of the art and future challenges of deep learning-based beamforming[J]. IEEE Access, 2022, 10: 80869–80882. doi: 10.1109/ACCESS.2022.3195299.