MIMO Radar Phase-Coded Waveform Design for Spectrum Sharing: A Neural Network Approach

Spectrum sharing technology significantly enhance the utilization of radio frequency (RF) spectrum resource in radar and communication systems. MIMO radar has higher degrees of freedom in transmitting notched waveforms and is more likely to coexist with in band users. Unlike traditional optimization algorithms, this paper utilizes a neural network to address spectral constrained waveform design problem. Based on the proposed solution mapping network (SMN), unsupervised learning is employed to minimize a weighted sum of waveform orthogonality metrics and spectrum metric. Numerical results demonstrate that this network effectively designs waveforms with spectral notches while maintaining high waveform orthogonality.