Phase-coded Waveform Design for MIMO Radar Based on Nonlinear Conjugate Gradient Method

Multiple-input multiple-output (MIMO) radar can enhance its waveform diversity gain by designing waveforms with good orthogonality. The peak sidelobe level (PSL) is a key metric for evaluating the orthogonality of MIMO radar waveforms. This paper proposes the LSE-CG algorithm to address the slow convergence of traditional gradient descent in PSL optimization. It combines the nonlinear conjugate gradient (NCG) method with Newton's Downhill-based step size search. The algorithm optimizes the PSL using the Log-Sum-Exp (LSE) approximation and achieves faster convergence while PSL is closer to lower-bound compared to conventional approaches.