Cognitive Radar Antijamming via Synergistic Frequency–Polarization Adaptation
To address the limitations of single-domain antijamming techniques against dynamic mainlobe suppression jamming, this article proposes a cognitive radar framework based on synergistic frequency–polarization adaptation. Conventional frequency agility can become ineffective under wideband barrage jamming, while passive polarization filtering may lead to considerable target signal attenuation. To alleviate these limitations, the proposed framework jointly optimizes transmit frequency and polarization to improve the tradeoff between jamming suppression and target energy retention. The antijamming problem is formulated as a partially observable Markov decision process and solved within a multiagent reinforcement learning (RL) framework, where discrete frequency selection and continuous polarization control are handled in a coordinated manner. This synergistic optimization allows frequency selection and polarization control to cooperate in shaping both the target scattering condition and the signal-jamming coupling, rather than acting as two independent or sequential countermeasures. In addition, a policy distillation-based deep RL mechanism is introduced to improve adaptation efficiency under limited online interaction. Numerical simulations show that the proposed framework achieves more than 3-dB signal-to-interference-plus-noise ratio improvement over conventional single-domain methods and accelerates policy convergence. Additional evaluations under several nonideal factors further illustrate the robustness of the proposed framework under more practical conditions.