A Resource Scheduling Algorithm for Multi-Target 3D Imaging in Radar Network Based on Deep Reinforcement Learning

Inverse synthetic aperture radar (ISAR) three-dimensional (3D) imaging technology enables the acquisition of clear 3D structures of targets, significantly enhancing target recognition performance. In resource-constrained environments, an effective resource scheduling algorithm is essential for achieving high-quality 3D imaging of multiple targets. However, existing algorithms often neglect the quality requirements of 3D imaging during resource allocation. A resource scheduling algorithm for multi-target 3D imaging in a radar network based on deep reinforcement learning (DRL) is proposed in this paper, achieving multi-target 3D imaging with minimal time resource consumption while ensuring the imaging quality of targets. First, based on the projection-based multi-view ISAR 3D imaging method, the impact of the radar distribution and radar number on the target imaging quality is analyzed. Subsequently, a resource scheduling model is constructed with the objective of minimizing time consumption while ensuring target imaging quality. The problem is then formulated as a Markov decision process, and the Advantage Actor–Critic (A2C) deep reinforcement learning method is employed to solve the model. By reasonably designing the reward for reinforcement learning and pruning the action space based on domain knowledge, the convergence speed of the network is significantly accelerated. An optimal scheduling strategy including a radar node allocation scheme and timing pulse allocation scheme for each radar can be obtained after convergence. The simulation experiments validate the effectiveness of the proposed algorithm.