부산대학교 도서관

Frequency agility (FA) radar can avoid jamming by changing its carrier frequency, which makes it difficult for conventional jammers to cope with. In this paper, an intelligent jamming strategy design method is proposed. The process of the jammer and the FA radar confrontation is modeled as a Markov Decision Process, in which the jammer and the FA radar act as the agent and environment, respectively. The reinforcement learning algorithm, Proximal Policy Optimization, is selected to solve the jamming strategy design problem. To overcome the problem that the jamming effect cannot be evaluated by jammers in time-sharing mode, a decision approach of periodic action-evaluation is proposed. To evaluate the performance of the jammer, we elaborately designed a reward function with three components: intercept reward, launch reward, and evaluation reward. Simulation results demonstrate that the proposed method can learn highly adaptive strategies in the FA radar countermeasure scenarios and achieve good jamming performance.