Reasearch On Intelligent Jamming Decision-making And Evaluation Method For Multi-functional Radar Network

Multi-functional radar networks can coordinate multiple radar devices to achieve more efficient,accurate,and reliable target monitoring and detection.Moreover,with networking and multi-functional radar features,multi-functional radar networks possess strong anti-interference capabilities.To address the interference problem in multi-functional radar networks,this thesis is divided into three parts,analyzing the weak links of vulnerability to interference in multi-functional radar networks,how to make interference decisions,and how to conduct interference evaluations.The main contents of this thesis are as follows:1.Analyzing the characteristics and signal processing flow of multi-functional radar networks to lay the theoretical foundation for analyzing their weak links of vulnerability to interference.From three aspects of the overall system,signal processing,and information processing,this thesis analyzes the weak links of single radar and the radar network system to interference.For example,the radar still uses conventional energy detection in signal detection,making it susceptible to forwarding deception interference.When subjected to cooperative deception interference and false target position errors within the radar measurement accuracy range,the radar network is difficult to rely on homogeneity testing to eliminate false targets,etc.When subjected to dense false target interference,the data fusion of the radar network system consumes a lot of computational resources,leading to a decline in the overall system performance.2.Analyzing the counter-environment of multi-functional radar networks and using a game theory-based approach to make interference decisions,combining the profit matrix of both sides to solve the interference strategy.To address the shortcomings of the game theory algorithm,this thesis researches intelligent interference decision-making algorithms based on deep reinforcement learning and establishes a Markov decision model for interference confrontation in multi-functional radar networks.Leveraging the characteristics of reinforcement learning and the advantages of neural networks,the DQN algorithm is used to select interference methods to obtain the optimal interference strategy.3.Based on the DQN algorithm,this thesis studies improved algorithms such as Double DQN,Dueling DQN,D3 QN,and Priority Experience Replay,and simulates and analyzes their performance improvements compared to the DQN algorithm.When the scale of confrontation increases,the performance of the DQN and its improved algorithms significantly decreases when dealing with problems of large action space dimensions.The Branching DQN algorithm is used for interference decision-making,and simulation verification of the algorithm’s efficiency is conducted.4.To address the interference evaluation problem in multi-functional radar networks,as a non-cooperative party,interference is evaluated based on reconnaissance feedback information before and after interference.A evaluation index system is established from the time domain,spatial domain,frequency domain,energy domain,and modulation domain.Based on the evaluation index system,indicators are extracted,and samples are constructed.Using evaluation methods based on Random Forest and Support Vector Machines,interference effectiveness evaluation is conducted,and the accuracy of the evaluation is simulated and verified.Leveraging the ability of CNN to extract data features and LSTM to extract time-series features,this thesis proposes an intelligent evaluation method based on CNN-LSTM,and the accuracy of the evaluation is simulated and verified.