基于随机森林强化学习的干扰智能决策方法研究

Research on Jamming Intelligent Decision-Making Method Based on Random Forest Reinforcement Learning

电磁频谱的主导权是现代化电子战制胜的关键。传统的通信对抗中干扰方的干扰模式相对固定单一,干扰效率低下。因此,研究频谱对抗环境中利用强化学习智能选择干扰信道策略对用户通信进行干扰,将干扰方的信道决策过程建模为一个马尔科夫决策过程(Markov Decision Making Process,MDP),并提出了一种基于随机森林强化学习的智能干扰算法。仿真结果表明,与文献[10]所提的智能干扰算法和基于感知的随机信道选择算法相比,所提随机森林强化学习算法干扰收敛速度最快。通过在线自主学习,干扰方可以快速寻找到用户的通信规律,对用户通信实施有效干扰。

The dominance of the electromagnetic spectrum is the key to the success of modern electronic warfare. The mode of the jammer in the traditional communication confrontation is relatively fixed and the efficiency is low. In this paper, we study how to use reinforcement learning intelligent to choose jamming channel strategy for jamming users’ communication in spectrum countermeasure environment. We formulate the channel decision process of the jamming as a Markov Decision Process(MDP) and propose an intelligent jamming algorithm based on the random forest reinforcement learning. The simulation results show that compared with the intelligent jamming algorithm proposed in [10] and the sensing-based random channel selection algorithm, the proposed algorithm has the fastest convergence speed. Through online self-learning, the jammer can quickly find the users’ communication rules and effectively jam the users’ communication.