Research On Knowledge-based And Waveform Agility Technology For Target Tracking

With the development of technology,the working environment of contemporary radar is increasingly complicated.In order to improve the adaptive level of radar and its ability to cope with complex and changeable working environment,cognitive radar is put forward and studied as an important development direction of next generation of new radar.The main difference between cognitive radar and traditional radar system lies in its knowledge-aided ability and adaptive ability.Based on the framework of cognitive radar,this paper focuses on the problem of target tracking and studies the adaptive selection of transmitted waveform.The main work of this paper is summarized as follows:(1)Briefly introduced the basic framework of cognitive dynamic system and related decision-making theory,analyzed the basic framework of cognitive radar,analyzed the target tracking problem of cognitive radar from the perspective of cognitive dynamic system behavior decision.(2)The related theories of cognitive radar target tracking and waveform agility are studied.The baseband model of radar signal is introduced,and the relationship between the transmitted waveform and the measured noise is deduced based on the radar echo model and classical parameter estimation theory.The state space model of target tracking is introduced,and typical tracking filters in linear,nonlinear and clutter scenes are studied.From the two perspectives of the tracking accuracy and the reasonable allocation of system resources,the criterion function of waveform judging is determined,and the complete flow of waveform agility algorithm based on the criterion function is given.Next,the waveform agility algorithm based on criterion function is extended to clutter environment.(3)Based on the introduction of relevant algorithms for behavioral decision-making in cognitive dynamic systems,the enhanced learning framework in the problem of waveform selection in cognitive radar tracking is analyzed.A cognitive tracking waveform selection algorithm based on reinforcement learning is proposed by using ideas of value function estimation,learning and planning process,exploration and utilization balance,etc.The return function is designed for different waveform selection criteria.(4)For the algorithms introduced and proposed in this paper,comparative experiments were performed in scenarios with/without clutter,linear/nonlinear,and in scenarios of using different waveform selection criteria.The characteristics and pros and cons of different algorithms were analyzed.Experiments show that compared with traditional waveform agility algorithm based on criterion function,the cognitive tracking waveform selection algorithm in this paper has advantages in tracking performance.