Radar Target Detection Based On Information Geometry

The research of radar target detection has always been a subject of great importance.In civil navigation and military field,people try to develop new theories and methods to improve the detection performance.Information geometry is a new theoretical system which is used to solve statistical problems by using the modern differential geometry method in Riemann manifold,and it provides a new way to solve the problem of radar signal processing.In this paper,,the detection of radar target signal is studied based on the theory of information geometry,and the problem of statistical detection is transformed into the solution of geometric problem on manifold.Firstly,the principle of traditional radar target detection is analyzed in this paper,the methods to improve the detection performance and the disadvantages of them are introduced and the feasibility of introducing information geometry theory is illustrated.Secondly,some basic concepts of the information geometry theory are discussed,the distance between two probability distributions is defined based on statistical manifold,and two kinds of distance detectors based on KLD and geodesic for radar signal are put forward,then they are improved by using pulse accumulation and optimization algorithm,and application of dim target in low SNR are discussed.The simulation results show that the performance of distance detector is better than the traditional method and more reliable.Finally,the properties of positive definite matrix manifolds under Gauss clutter are studied,an algorithm of matrix CFAR Detector on the basis of Riemann geometry on manifolds is designed,and the detection threshold and detection performance for non fluctuating targets and fluctuating targets with different affecting factors is analyzed;for the non Gauss correlated clutter,the target is detected by calculating the distance between the estimated distribution on manifold obtained by observation data and the two hypotheses.Simulation results show that the performance of matrix CFAR detection based on geometric information is better than the cell average CFAR detection,and this method also has better detection performance in complex clutter background,it is easy to implement and has a definite application prospect.