Research On Blind Source Separation Technology Of Centralized MIMO Radar

Multiple input multiple output radar is a new type of system radar,which can simultaneously transmit diversity waveforms and jointly process the received echo signals.MIMO radar has many advantages such as high angle accuracy,good target detection performance,and strong anti-interference suppression ability.The correct blind source separation of observed MIMO radar signals is an important prerequisite for effective interference to enemy's radars.Existing MIMO blind source separation algorithms mainly deal with single MIMO radar signals.However,in a complex electronic warfare environment,the combatants need to confront multiple MIMO radars.The blind source separation technology of centralized MIMO radar is studied emphatically in this thesis.In the first part of this thesis,basic principle of centralized MIMO radar is studied,and the waveform design conditions of the centralized MIMO radar are analyzed.The timefrequency domain characteristics of the orthogonal frequency-divided linear frequency modulation signal,the orthogonal polyphase coded signal and the orthogonal discrete frequency coded signal are discussed,and these three typical signals are simulated.In the second part,the blind source separation technology is studied.Based on the mathematical model of blind source separation,the non-underdetermined and underdetermined separation conditions are discussed.From two perspectives,the blind source separation evaluation index is analyzed,then the blind source separation method under different conditions are studied and the underdetermined separation are verified by simulation.In the third part,the blind source separation of single centralized MIMO radar is studied.The single MIMO radar reconnaissance model under non-underdetermined conditions is analyzed.The source number estimation method based on principal component analysis combined with different criteria is discussed,thus the dimension of the observation signal are reduced and the noise are decreased.The fast blind source separation algorithm in complex value form is studied,and the problem that the signal and the mixing matrix in the MIMO radar reconnaissance model are all complex values are solved,then the separation simulations of three common used MIMO radar signals are performed to verify the effectiveness of the separation algorithm.In the fourth part,the blind source separation of multiple centralized MIMO radars is presented.Firstly,the reconnaissance model of multiple MIMO radars is established,and the mathematical expressions of the observed signals are deduced.Then the angle-of-arrival estimation algorithm based on the Capon minimum variance method is studied,thus the azimuth angles of different MIMO radars are determined.After that,the preset beam zero point technology is used to filter single MIMO radar signal from the mixed signals,so that the source signal can be recovered by using the classical blind source separation method.Finally,through the blind source separation simulation of multiple MIMO radar signals,the correctness of the corresponding algorithm is verified.Through the study of the above four parts,the correct separation of single and multiple centralized MIMO radar signals are realized,which further perfects the existing blind source separation method for centralized MIMO radar.