基于动态联盟的无人机集群协同干扰方法

Collaborative Jamming Based on Dynamic Alliances in UAV Cluster

针对无人机集群对多个异构电磁目标进行协同干扰时面临的多智能体协同学习决策问题,提出了基于动态联盟的无人机集群协同干扰方法。运用马尔可夫决策对协同干扰进行建模,构建干扰联盟博弈模型,优化集群干扰中的频率和功率联合决策以及干扰资源的动态调配,实现对多个用频模式不同、压制功率需求不同的电磁目标进行干扰的系统效能优化。首先,利用协同强化学习算法对任务目标的信道决策进行学习,联盟内无人机协同干扰;然后,根据干扰效果进行干扰联盟结构的动态调整。所提算法可从强化学习进程中根据干扰效能的差异动态调整干扰资源,并能适应目标功率的变化,相同场景下,系统干扰效能值较基线算法提升约12.5%。

For the multi-agent collaborative and learning decision-making problem encountered by unmanned aerial vehicle(UAV)cluster when coordinating jamming against multiple heterogeneous electromagnetic targets,a collaborative jamming method based on dynamic alliances is proposed.The method employs Markov decision processes to model coordinated jamming and constructs an jamming coalition game model.It optimizes joint decisions on frequency and power in cluster jamming and dynamically allocates jamming resources,in order to optimize system performance in interfering with multiple electromagnetic targets with different frequency modes and suppression power requirements.Initially,reinforcement learning algorithms are utilized to learn the channel decisions of task objectives,enabling UAVs within the alliance to collaboratively select channels.Subsequently,adjustments to the alliance structure among alliances are made based on jamming effects.The proposed algorithm can dynamically adjust jamming resources based on differences in jamming effectiveness through the reinforcement learning process and adapt to changes in target power.In the same scenario,the system jamming effectiveness value is improved by approximately 12.5% compared with that of baseline algorithms.