Intelligent Anti-Jamming Decision-Making Technology Based on Knowledge Graph and DQN

Recent advancements in artificial intelligence have driven significant progress in intelligent anti-jamming communications. However, existing methods still face two major limitations: reinforcement learning-based models often suffer from slow convergence, while knowledge graph-based approaches lack dynamic interaction capabilities in complex, time-varying electromagnetic environments. To address these challenges, this paper proposes a novel two-stage intelligent decision-making framework. In the first stage, an anti-jamming knowledge graph repository is constructed to enable rapid decision-making through efficient reasoning, thereby ensuring real-time responsiveness. The second stage introduces a hierarchical reinforcement learning architecture that facilitates environmental interaction for continuous model evolution and self-adaptation. By simplifying multidimensional parameter spaces into two-dimensional decision scenarios, the proposed method effectively reduces computational complexity and accelerates convergence. Experimental results demonstrate that the proposed method achieves a 4.2% increase in the anti-jamming decision success rate and a 104.8% improvement in the transmission rate compared to state-of-the-art methods. Simulation results demonstrate the superiority of the framework in both anti-jamming performance and learning efficiency, validating its practical effectiveness in dynamic electromagnetic environments.