Study On Distributed Radar Fusion Detection Method

Due to the increasingly complex battlefield environment and target characteristics,modern radar needs to deal with the "four threats" of electromagnetic interference,stealth technology,anti-radiation missile and low-altitude penetration.The detection performance of single-node radar system is sometimes difficult to meet the demand,so it is urgent for radar system to improve its detection performance from all aspects.Radar is gradually developing towards multi-base station and distributed,which makes multi-node radar cooperate in detection,thus improving detection capability.This thesis studies the fusion detection algorithm of distributed radar.The spatial diversity gain is used to suppress the fluctuation characteristics of the target.Reduce the degradation of detection performance caused by the fluctuation of the scattering cross section of the target.The main research contents are as follows:1,the performance of fusion detection algorithm in different backgrounds is studied.Aiming at Gaussian background,the algorithms of decision-level fusion and signal-level fusion are studied.In the decision-level fusion algorithm,the fusion center uses ―K‖ rank criterion,and no matter what value K takes,it can get better detection performance than single-node radar.The weighted fusion can improve the performance loss caused by the difference of signal-to-noise ratio of each node.In the signal-level fusion algorithm,when the performance of each node is quite different,the detection performance of the incoherent accumulation detector will be reduced or even lower than that of a single node,while the signal-to-noise ratio weighted detector will not.Aiming at the non-Gaussian background,the performance of fusion detection algorithm under the background of lognormal distribution,Weibull distribution and K distribution is analyzed.The change of shape parameters will lead to the change of clutter power.The clutter power of lognormal distribution and K distribution is directly proportional to shape parameters,while the clutter power of Weibull distribution is inversely proportional to shape parameters,thus affecting the detection performance.When the fusion center uses ―K‖ rank criterion,no matter what value K takes,the performance of fusion detection is better than that of single node,and the optimal K value is different in different backgrounds.The weighted fusion can improve the performance loss caused by the difference of signal-to-noise ratio of each node.2,a decision-level fusion detection algorithm with adaptive threshold is studied.Firstly,according to Niemann-Pearson criterion,the objective function model to be optimized is established,and the Lagrange multiplier method is used to solve it,and the relationship between local threshold and signal-to-noise ratio(or signal-to-noise ratio)is obtained.The implicit function about local threshold is obtained,which reduces the search space in a certain way,thus reducing the amount of calculation.Finally,the simulation verifies the effectiveness of the algorithm in Gaussian and non-Gaussian background,and the simulation results show that the detection performance of the algorithm is better than that of the existing fixed threshold fusion detection.In this paper,a method for calculating the optimal adaptive threshold is given.When the fusion center adopts AND criterion and weighted fusion criterion,the formulas for calculating the first derivative of global detection probability and false alarm probability about local nominal factor are provided.