Reduced-Complexity CRB Optimization for Dual-Function Radar-Communication Systems Using Hybrid Linear-Nonlinear Precoding

This paper proposes a waveform design technique for dual-function radar-communication (DFRC) systems that employ hybrid linear-nonlinear precoding (HLNP). The HLNP signal is a superposition of linear precoding of the communication symbols that enables conventional coherent decoding and a nonlinearly precoded auxiliary signal that introduces additional degrees of design freedom that can be used to improve system performance. Our design goal is to obtain accurate direction of arrival (DOA) estimation and satisfactory waveform ambiguity properties, and hence we optimize a weighted sum of a Cramer-Rao bound (CRB) on DOA estimation and a waveform similarity metric. To simultaneously enable effective communication at the chosen data rates, we set lower bounds on the communication SINRs, and to facilitate implementation, we constrain the total transmission power, the per-antenna power, and the peak-to-average power ratio (PAPR) on each antenna. We deploy successive convex approximation to solve the resultant nonconvex design problem, while leveraging feasible point pursuit to provide a feasible initial point. To reduce the computational cost, we introduce sub-block and null space variants of our design technique. Simulation results verify the effectiveness of the proposed algorithm and its variants, and validate their performance advantages over regular nonlinear precoding schemes.