基于强化学习的自适应干扰波形设计

Adaptive Interference Waveform Design Based on Reinforcement Learning

针对传统干扰算法无法适应信号环境变化的问题,提出将Q-Learning算法与“切割”假设法相结合应用到干扰波形设计中,使干扰波形能够达到自适应雷达信号长度变化的效果。该算法主要针对雷达检测环节进行干扰,其中采用恒虚警概率(constant false alarm rate,CFAR)作为环境交互模型,通过强化学习自适应地调整间歇采样信号的采样时间与转发时间,在此基础上对未知长度雷达信号进行“切割”处理以达到最佳干扰的目的。最后进行仿真,实现了对未知雷达信号的干扰。仿真结果表明:在信号模型不定的条件下,强化学习算法在决策时可以充分利用历史数据,相对于传统算法,强化学习算法可以达到更好的干扰效果。

Aiming at the problem that the traditional jamming algorithm cannot adapt to the change of signal environment,the Q-learning algorithm and the hypothesis method of"cutting"are combined to apply to the design of jamming waveform,so that the jamming waveform can achieve the effect of adaptive radar signal length change.The algorithm is mainly aimed at jamming the radar detection link.Constant false Alarm rate(CFAR)is adopted as the environment interaction model,and the sampling time and forwarding time of intermittent sampling signals are adjusted adaptively through enhanced learning.On this basis,the radar signals of unknown length are processed by"cutting"to achieve the purpose of optimal interference.Finally,simulation is carried out to realize the jamming of unknown radar signal.The simulation results show that the reinforcement learning algorithm can make full use of the historical data when making decision under the condition of uncertain signal model,and can achieve better interference effect compared with the traditional algorithm.