MIMO Radar Waveform Design for Multipath Exploitation Using Deep Learning

This paper investigates the design of waveforms for multiple-input multiple-output (MIMO) radar systems that can exploit multipath returns to enhance target detection performance. By making reasonable use information in waveform design, MIMO effectively improve signal-to-interference and noise ratio (SINR) receiver under a constant modulus (CM) constraint. However, optimizing these constraints is challenging non-linear non-convex problem cannot be easily solved using traditional methods. To overcome this challenge, we proposed novel method scenarios based on deep learning. Specifically, leveraged powerful nonlinear fitting ability neural networks solve optimization problem. First, constructed residual network transform CM constraint into phase sequence Next, used as loss function network. Finally, adaptive moment estimation (Adam) optimizer train Simulation results demonstrated our outperformed existing methods by achieving better SINR values receiver.