基於 Swin-Transformer 語意分割的雷達訊號解交織之研究

以作者查詢圖書館館藏

、以作者查詢臺灣博碩士

、以作者查詢全國書目

、勘誤回報

、線上人數：15

、訪客IP：18.117.80.241

姓名

王濰翊(Wei-Yi Wang) 查詢紙本館藏

畢業系所

通訊工程學系

論文名稱

基於 Swin-Transformer 語意分割的雷達訊號解交織之研究
(Radar Signal Deinterleaving Based On Swin-Transformer Segmentation)

相關論文

★ 基於多代理人強化學習方法多架無人機自主追蹤之研究	★ 連網無人機路徑規劃與基地台連線策略之共同設計：使用模仿增強的深度強化學習方法
★ 網路編碼於多架無人機網路以抵禦機器學習攻擊之研究	★ 使用多代理人強化學習於無線快取網路設計空中基地台三維路徑之研究
★ 空中無線感測：在無人機抖動環境下基於RSS的非接觸式人體偵測與定位	★ 協作式自編碼器嵌入最佳化方法於無人機群網路以接收訊號強度輔助定位之研究

檔案

[Endnote RIS 格式]

[Bibtex 格式]

[相關文章]

[文章引用]

[完整記錄]

[館藏目錄]

至系統瀏覽論文 (2027-8-1以後開放)

摘要(中)

在電子戰（EW）不斷發展的背景下，先進的電子支援措施（ESM）系統已變得不可或缺。傳統的解交織技術，在 ESM 中透過脈衝描述字（PDWs）進行發射源分類是基礎，但現在由於訊號參數的增加複雜性、變化以及在電子戰場景中訊號攔截幾率的降低，已不再適用。本文提出了一種新的解交織演算法，主要利用到達時間（TOA）值來克服當前技術的局限性，這些技術在沒有到達方向（DOA）資料的情況下過度依賴 PDWs。為了處理複雜的脈衝重複間隔（PRI）調變和遺失脈衝情況，我們將影像分割應用於雷達解交織中進行創新。我們的方法使用 Swin Transformer 模型，將 TOA 資料轉換成二維影像格式以進行更精細的分析。我們將這項技術稱為像素級 PRI 影像分類，它允許在複雜的 PRI 情境中準確識別脈衝序列。與最先進的基於深度學習的方法相比，我們的方法在有效解交織所需的脈衝數上顯著減少。模擬結果也表明，尤其是在遺失或虛假脈衝的場景中，我們的模型優於現有基於直方圖的和先進學習的方法。我們的雷達序列資料集可以在 https://github.com/ICAN-Lab/PRI-Frequency-Image-Dataset 上公開下載。

摘要(英)

In the evolving landscape of electronic warfare (EW), advanced electronic support measure (ESM) systems have become imperative.
Traditional deinterleaving techniques, which are fundamental in ESM for emitter categorization through pulse description words (PDWs), are now inadequate due to the increased complexity, variation in signal parameters, and lower chances of signal interception in EW scenarios.
This paper presents a new deinterleaving algorithm that primarily utilizes time of arrival (TOA) values to overcome the limitations of current techniques that depend heavily on PDWs, particularly in situations where direction of arrival (DOA) data is not available.
To handle complex pulse repetition interval (PRI) modulations and missing pulse situations, we innovate by applying image segmentation to radar deinterleaving.
Using the Swin Transformer, our method transforms TOA data into a two-dimensional image format for enhanced analysis.
We refer to this technique as extit{pixelwise PRI image classification}, which allows for accurate identification of pulse sequences in complex PRI situations.
Compared to the state-of-the-art deep learning based method, our method achieves a significant decrease in the number of pulses needed for effective deinterleaving.
Simulation results also show that our model outperforms existing histogram-based and advanced learning-based methods, especially in scenarios with lost or spurious pulses.
Our radar sequence dataset can be publicly downloaded at https://github.com/ICAN-Lab/PRI-Frequency-Image-Dataset.

關鍵字(中)

★ 解交織
★ 脈衝流
★ PRI調變
★ Transformer
★ 語意分割

關鍵字(英)

★ Deinterleaving
★ pulse streams
★ PRI modulation
★ transformer
★ semantic segmentation

論文目次

論文摘要 i
Abstract ii
目錄 v
圖目錄 vii
表目錄 ix
一、緒論 1
1.1研究背景與動機 1
1.2論文架構 4
二、文獻探討 5
2.1傳統方法 5
2.2深度學習方法 6
三、系統模型 9
3.1 TOA序列 9
3.2 PRI調變 10
3.2.1恆定PRI（ConstantPRI） 10
3.2.2交錯PRI（StaggerPRI） 10
3.2.3滑動PRI（SlidingPRI）11
3.2.4停留與切換PRI（DwellandSwitchPRI） 11
3.3雷達信號處理中雜訊影響的模擬 13
3.3.1虛假脈衝（SpuriousPulse） 13
3.3.2丟失脈衝（MissingPulse） 13
3.3.3測量誤差（MeasurementErrors） 14
3.4先進雷達信號解交織中的挑戰與解決方案 14
四、Swin-TransformerSegmentation用於雷達解交織 15
4.1數據處理方法 15
4.1.1構建DTOA矩陣 15
4.1.2構建PRI頻率矩陣及SwinTransformer輸入的影像
轉換 16
4.1.3通過目標尺寸增強來提高影像分割的準確性 17
4.1.4建構標籤矩陣 18
4.2用於解交織的SwinTransformer模型 21
4.2.1整體架構 21
4.2.2窗口自注意力和位移窗口自注意力機制 22
4.2.3 SwinTransformer模塊架構 23
4.2.4損失函數 24
4.2.5複雜度分析 24
五、性能評估 27
5.1實驗設置 27
5.2基準方法 28
5.3評估指標 28
5.4實驗一：不同調變類型的解交織性能 29
5.5實驗二：交錯與恆定PRI下的性能比較 33
5.6實驗三：交錯PRI下的性能比較 37
5.7人工智慧模型的推理權衡 40
六、結論與貢獻 41
索引 42
參考文獻 42

參考文獻

[1] Y. Liu andQ.Zhang, “Improvedmethodfordeinterleaving radar signals and estimating PRI values,” IET Radar, Sonar & Navigation, vol. 12, no. 5, pp. 506–514, 2018.
[2] R. Wiley, ELINT: The interception and analysis of radar signals. Lon don, U.K.: Artech House, 2006.
[3] D. K. Barton, Radar system analysis and modeling. Norwood, MA, USA: Artech House, 2004.
[4] D. Milojevi´ c and B. Popovi´ c, “Improved algorithm for the deinterleav ing of radar pulses,” in IEE Proceedings F (Radar and Signal Process ing), vol. 139, no. 1, 1992, pp. 98–104.
[5] R. J. Orsi, J. B. Moore, and R. E. Mahony, “Spectrum estimation of in terleaved pulse trains,” IEEE transactions on signal processing, vol. 47, no. 6, pp. 1646–1653, 1999.
[6] J. Liu, H. Meng, Y. Liu, and X. Wang, “Deinterleaving pulse trains in unconventional circumstances using multiple hypothesis tracking algo rithm,” Signal Processing, vol. 90, no. 8, pp. 2581–2593, 2010.
[7] N. Visnevski, S. Haykin, V. Krishnamurthy, F. A. Dilkes, and P. Lavoie, “Hidden Markov models for radar pulse train analysis in electronic war fare,” in IEEE International Conference on Acoustics, Speech, and Sig nal Processing, vol. 5, 2005, pp. 597–600.
[8] Y. Liu andQ.Zhang, “Improvedmethodfordeinterleaving radar signals and estimating PRI values,” IET Radar, Sonar & Navigation, vol. 12, no. 5, pp. 506–514, 2018.
[9] C. Davies and P. Hollands, “Automatic processing for ESM,” in IEE Proceedings F (Communications, Radar and Signal Processing), vol. 129, no. 3, 1982, pp. 164–171.
[10] D. Wilkinson and A. Watson, “Use of metric techniques in ESM data processing,” in IEE Proceedings F (Communications, Radar and Signal Processing), vol. 132, no. 4, 1985, pp. 229–232.
[11] T. Tian, J. Ni, and Y. Jiang, “Deinterleaving method of complex stag gered PRI radar signals based on EDW fusion,” The Journal of Engi neering, vol. 2019, no. 20, pp. 6818–6822, 2019.
[12] H. Mardia, “New techniques for the deinterleaving of repetitive se quences,” in IEE Proceedings F (Radar and Signal Processing), vol. 136, no. 4, 1989, pp. 149–154.
[13] Y. Xi, Y. Wu, X. Wu, and K. Jiang, “An improved SDIF algorithm for anti-radiation radar using dynamic sequence search,” in 36th Chinese Control Conference (CCC), 2017, pp. 5596–5601.
[14] Z.-M. Liu and P. S. Yu, “Classification, denoising, and deinterleaving of pulse streams with recurrent neural networks,” IEEE Transactions on Aerospace and Electronic Systems, vol. 55, no. 4, pp. 1624–1639, 2019.
[15] X. Li, Z. Liu, and Z. Huang, “Deinterleaving of pulse streams with de noising autoencoders,” IEEE Transactions on Aerospace and Electronic Systems, vol. 56, no. 6, pp. 4767–4778, 2020.
[16] W. Chao, S. Liting, L. Zhangmeng, and H. Zhitao, “A radar signal deinterleaving method based on semantic segmentation with neural net work,” IEEE Transactions on Signal Processing, vol. 70, pp. 5806 5821, 2022.
[17] Z. Liu, Y. Lin, Y. Cao, H. Hu, Y. Wei, Z. Zhang, S. Lin, and B. Guo, “Swin transformer: Hierarchical vision transformer using shifted win dows,” in IEEE/CVF International Conference on Computer Vision (ICCV), 2021, pp. 9992–10002.
[18] W. Y. Wang, Y. J. Chen, H. Y. Huang, and W. Chen, “Radar signal deinterleaving based on swin-transformer segmentation,” 2024. [Online]. Available: Image-Dataset https://github.com/ICAN-Lab/PRI-Frequency
[19] K. Nishiguchi and M. Kobayashi, “Improved algorithm for estimating pulse repetition intervals,” IEEE Transactions on Aerospace and Elec tronic Systems, vol. 36, no. 2, pp. 407–421, 2000.
[20] Y. Mao, J. Han, G. Guo, and X. Qing, “An improved algorithm of PRI transform,” in WRI Global Congress on Intelligent Systems, vol. 3, 2009, pp. 145–149.
[21] A. Mahdavi and A. M. Pezeshk, “A fast enhanced algorithm of PRI transform,” in IEEE International Symposium on Parallel Computing in Electrical Engineering, 2011, pp. 179–184.
[22] W. Cheng, Q. Zhang, J. Dong, C. Wang, X. Liu, and G. Fang, “An en hanced algorithm for deinterleaving mixed radar signals,” IEEE Trans actions on Aerospace and Electronic Systems, vol. 57, no. 6, pp. 3927 3940, 2021.
[23] J. Sun, G. Xu, W. Ren, and Z. Yan, “Radar emitter classification based on unidimensional convolutional neural network,” IET Radar, Sonar & Navigation, vol. 12, no. 8, pp. 862–867, 2018.
[24] Z.-M. Liu, “Pulse deinterleaving for multifunction radars with hierar chical deep neural networks,” IEEE Transactions on Aerospace and Electronic Systems, vol. 57, no. 6, pp. 3585–3599, 2021.
[25] A. Ata’a and S. Abdullah, “Deinterleaving of radar signals and PRF identification algorithms,” IET radar, sonar & navigation, vol. 1, no. 5, pp. 340–347, 2007.
[26] X. Li, Z. Liu, Z. Huang, and W. Liu, “Radar emitter classification with attention-based multi-RNNs,” IEEE Communications Letters, vol. 24, no. 9, pp. 2000–2004, 2020.
[27] X. Li, Z. Liu, and Z. Huang, “Attention-based radar PRI modulation recognition with recurrent neural networks,” IEEE Access, vol. 8, pp. 57426–57436, 2020.
[28] Y. Mao, W. Ren, X. Li, Z. Yang, and W. Cao, “Sep-RefineNet: A dein terleaving method for radar signals based on semantic segmentation,” Applied Sciences, vol. 13, no. 4, p. 2726, 2023.
[29] M. A. Nuhoglu, Y. K. Alp, M. E. C. Ulusoy, and H. A. Cirpan, “Image segmentation for radar signal deinterleaving using deep learning,” IEEE Transactions on Aerospace and Electronic Systems, vol. 59, no. 1, pp. 541–554, 2023.
[30] Y. Liu, J. Xie, M. Tao, S. Tang, Z. Tang, and L. Wang, “Stagger PRI radar signal deinterleaving based on image semantic segmentation,” in IEEE5thInternational Conference on Electronic Information and Com munication Technology (ICEICT), 2022, pp. 599–602.
[31] M. Contributors, “MMSegmentation: Openmmlab semantic seg mentation toolbox and benchmark,” 2020. [Online]. Available: https://github.com/open-mmlab/mmsegmentation.

指導教授

陳昱嘉(Yu-Jai Chen)

審核日期

2024-7-22

推文