Passive Tracking Algorithm Using Airborne Transmitter Radar With Unknown Clutter Spatial Distribution

As an important passive radar,airborne transmitter radar has the advantages of low cost,good concealment and strong anti-interference ability,which is of great significance in modern military defense.However,the movement of airborne platform seriously affects the clutter spatial distribution in the surveillance scenario.In order to address this problem,this thesis focuses on the passive tracking algorithm using airborne transmitter radar with unknown clutter spatial distribution.The main contributions are as follows:Firstly,in order to address the problem of multitarget tracking using airborne transmitter radar in strong clutter,the performance of several traditional clutter density estimation algorithms,such as the clutter density estimation algorithm based on target trajectory,clutter map algorithm,expectation maximization algorithm and fixed-order clutter sparsity estimation algorithm are analyzed.The clutter density estimation algorithm based on target trajectory is easy to implement,but the estimation error is large with unknown clutter spatial distribution.The clutter map method improves the accuracy of the clutter density estimation,but depends on spatial division.The expectation maximization algorithm can effectively improve the accuracy of the clutter density estimation by constructing the likeable function of the finite set,but it is not possible to estimate the clutter density online.The fixed-order clutter sparsity estimation algorithm estimates the clutter density online by solving the sparsity in the supersphere,but the fixed order sparsity is not suitable to the complicated clutter background.Secondly,in order to address the problem of multitarget tracking using airborne transmitter radar with unknown clutter spatial distribution,a clutter density estimator with optimized sparsity order is proposed.First,the clutter set is obtained by eliminating the potential target-originated measurements that fall within the validation gate.Thus,the effect of target measurement on clutter density estimation is reduced.Second,the samples of “sparsity order – hypercube volume” are constructed from the clutter set.Then the sparsity order is optimized online by the support vector regression and gradient method.Last,the output of the clutter density estimation is incorporated into the Gaussian mixture probability hypothesis density update.The simulation results show that the proposed algorithm can optimize the sparsity order online when the clutter distribution is unknown,and the fixed sparsity order can avoid the clutter density estimation error.Finally,in order to address the problem of multitarget tracking with unknown clutter spatial distribution,a clutter density estimator with sparsity order fusion is proposed.First,the potential target-originated measurements are rejected by the feedback of target state and clutter density,and the clutter measurement set is obtained.Second,the reinitialization based on the interacting multiple model is used to generate the clutter sparsity estimation with different models,and the clutter sparsity of the current model is predicted.Then the clutter density likelihood function is constructed using the interacting multiple model.The model-matched weight and clutter sparsity estimation are updated respectively.The simulation results show that the proposed algorithm can assign reasonable weights to the sparsity for each model and fuse the sparsity estimation of each model with corresponding model-matched weight.The target tracking accuracy of the proposed algorithm is better than clutter density estimator at the cost of more computational complexity.