期刊文献+

资源受限下基于改进Q学习的干扰自适应采样

An Adaptive Jamming Sampling Method Based on Improved Q-Learning Under Resource-Constrained Conditions
下载PDF
导出
摘要 资源受限条件下的干扰采样是限制干扰认知性能提升的瓶颈之一。将资源受限条件下的干扰采样问题建模为采样资源分配问题,提出了一种基于改进Q学习的干扰自适应采样方法。将多种典型干扰表示成Markov模型,并将资源受限条件下的采样子带选择过程建模为马尔科夫决策过程(Markov Decision Process,MDP)。针对干扰自适应采样应用需要同时具备较快收敛速率和较低稳态误差的需求,提出了一种温度参数动态调整的Q学习算法,并针对干扰时变的应用场景,在Q学习算法中嵌入干扰切换检测功能模块,改善了Q学习在干扰样式发生切换重新学习时探索不足的问题。仿真结果表明,与现有周期性采样方法相比,所提算法能显著提升多种典型干扰的有效采样效率,且能适应时变干扰的采样。 Jamming sampling under resource constraints is one of the bottlenecks that limit the enhancement of cognitive performance in jamming scenarios.The jamming sampling problem under resource-constrained conditions is modeled as a resource allocation problem,and a jamming-adaptive sampling method based on improved Q-learning is proposed.Initially,multiple typical jamming scenarios are represented as Markov models,and the process of sub-band selection for sampling under resource constraints is modeled as a Markov decision process.Subsequently,to fulfill the requirements of a rapid convergence rate and low steady-state error for adaptive jamming sampling applications,a Q-learning algorithm with dynamically adjusted temperature parameters is introduced.Furthermore,to accommodate the dynamic nature of jamming scenarios,a jamming transition detection module is incorporated into the Q-learning algorithm.This addition mitigates the issue of insufficient exploration when the Q-learning algorithm needs to relearn due to changes in the jamming pattern.Simulation results demonstrate that,compared to existing periodic sampling methods,the proposed algorithm significantly enhances the effective sampling efficiency across various typical jamming scenarios and is capable of adapting to the sampling of time-varying jamming.
作者 高科婕 朱勇刚 张凯 周展阳 GAO Kejie;ZHU Yonggang;ZHANG Kai;ZHOU Zhanyang(School of Electronics and Information Engineering,Nanjing University of Information Science and Technology,Nanjing 210044,China;Sixty-third Research Institute,National University of Defense Technology,Nanjing 210007,China)
出处 《电子信息对抗技术》 2024年第6期26-34,共9页 Electronic Information Warfare Technology
关键词 采样资源受限 干扰自适应采样 马尔科夫决策过程 Q学习 limited sampling resources jamming adaptive sampling Markov decision processes Q-learning
  • 相关文献

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部