Scholarlink

<p>In multiple-input multiple-output (MIMO) radar systems, the design of strictly orthogonal waveform sets plays crucial roles in enhancing interference resistance, improving target recognition accuracy, and optimizing signal transmission efficiency. However, existing methods face notable challenges. Traditional optimization algorithms often fail to achieve satisfactory solutions because of the nonconvexity and complexity of the objective functions, whereas convolutional neural networks (CNNs) suffer from limited receptive fields, thus preventing effective extraction of global sequence features and yielding poor orthogonality. To address these limitations, this paper proposes a phase sequence-based waveform design approach that leverages a transformer-based architecture. A novel causal temporal transformer (CTT) network is introduced, wherein customized encoder and decoder modules are constructed to capture the sequential characteristics of signals. Furthermore, the loss function is enhanced by incorporating mutual information to jointly optimize waveform orthogonality. The simulation results demonstrate that the proposed method exhibits superior correlation performance and achieves acceptable computational efficiency.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Orthogonal waveform design for MIMO radars on the basis of deep learning

  • 期刊论文

  • 发表日期：

    2026年3月10日

查看全文
  • Yuan Luo,
  • Guimao Du,
  • Jiaojiao Dang

摘要

In multiple-input multiple-output (MIMO) radar systems, the design of strictly orthogonal waveform sets plays crucial roles in enhancing interference resistance, improving target recognition accuracy, and optimizing signal transmission efficiency. However, existing methods face notable challenges. Traditional optimization algorithms often fail to achieve satisfactory solutions because of the nonconvexity and complexity of the objective functions, whereas convolutional neural networks (CNNs) suffer from limited receptive fields, thus preventing effective extraction of global sequence features and yielding poor orthogonality. To address these limitations, this paper proposes a phase sequence-based waveform design approach that leverages a transformer-based architecture. A novel causal temporal transformer (CTT) network is introduced, wherein customized encoder and decoder modules are constructed to capture the sequential characteristics of signals. Furthermore, the loss function is enhanced by incorporating mutual information to jointly optimize waveform orthogonality. The simulation results demonstrate that the proposed method exhibits superior correlation performance and achieves acceptable computational efficiency.