An Efficient Minorization Maximization Approach for MIMO Radar Waveform Optimization via Relative Entropy

We propose an efficient algorithm to design waveforms via relative entropy for multiple-input-multiple-output radar in the presence of (signal-dependent) clutter. The proposed algorithm is devised under the framework of minorization-maximization. Unlike the existing algorithm employing a highly nonlinear function to minorize the objective function, the proposed algorithm uses a quadratic function as the minorizer, leading to a much lower computational cost per iteration. In addition, we exploit an accelerated scheme called squared iterative method, to enhance the convergence rate of the proposed algorithm. Moreover, we extend the proposed algorithm to deal with additional constraints (i.e., constant-modulus constraint, similarity constraint, or both). Particularly, the extension allows for a single-stage design of the constrained waveforms and achieves a larger relative entropy than the existing algorithm. Numerical results are provided to show that the proposed algorithm is computationally more efficient and outperforms the existing algorithm when designing the constrained waveforms.