A fast and excellent unimodular waveform with good autocorrelation and cross-correlation design method for the multiple-input multiple-output radar is devised. Unlike the existing methods that only optimize partial metrics or only optimize the short sequence in acceptable time, we propose a comprehensive waveform design method to minimize the weighted sum of almost entirely metrics under the constant modulus constraint. Then, a deep learning framework, named as the comprehensive optimization network, is derived to handle the problem. Numerical results show that the proposed method has superior performance and acceptable optimization time compared with the existing methods.

中文翻译：

---

通过深度学习方法设计MIMO雷达的单模波形

为多输入多输出雷达设计了一种具有良好自相关和互相关设计方法的快速，优良的单模波形。与仅在可接受的时间内仅优化部分量度或仅优化短序列的现有方法不同，我们提出了一种全面的波形设计方法，以在恒定模数约束下将几乎全部量度的加权总和最小化。然后，推导了一个名为综合优化网络的深度学习框架来处理该问题。数值结果表明，与现有方法相比，该方法具有优越的性能和可接受的优化时间。