A Joint Spatio-Temporal Resource Allocation Algorithm for Multi-Target ISAR Imaging in a Radar Network Based on Utility Maximization

Inverse synthetic aperture radar (ISAR) imaging enables high-resolution two-dimensional reconstruction of non-cooperative moving targets, serving as a crucial tool for radar target recognition and situational awareness. The radar network faces the coupled optimization problem of spatial matching and temporal sequencing in order to execute multi-target ISAR imaging tasks. Firstly, the spatial matching scheme directly determines the observational geometric relationship, significantly affecting the two-dimensional resolution and image quality of ISAR imaging. Secondly, the temporal domain sequencing scheme sets the starting time of each imaging task. It dynamically changes the coherent integration time (CIT) of the task and influences the overall scheduling efficiency. Traditional methods can only obtain suboptimal allocation schemes, with low imaging utility, and the running time significantly increases when facing large-scale scenarios. In response to the above issues, this paper designs quality functions and efficiency functions to quantify the quality of spatial domain matching and the execution efficiency in the temporal domain. It also combines task priority to design a utility function for balancing quality and efficiency, and a joint allocation constraint optimization model for multi-target ISAR spatio-temporal resource scheduling under utility maximization is constructed. At the same time, the proximal policy optimization (PPO) is introduced to realize sequential online decision-making for the coupled scheduling of spatio-temporal resources, and a complete resource allocation scheme is generated step by step. Experimental results in a double-scale simulation scenario show that the proposed algorithm has stronger training convergence stability compared to the Double Deep Q-Network (DDQN) algorithm, and outperforms the random policy algorithm (RPA), genetic algorithm (GA), and particle swarm optimization (PSO) in imaging utility, with reasonable running time, demonstrating that the proposed algorithm balances optimization accuracy and real-time solution efficiency.