Research On Blind Source Separation Of MIMO Radar Signals In Underdetermined Mixing Model

Radar reconnaissance is the key tool of military intelligence acquisition in modern electronic warfare.As the key technology of radar reconnaissance and confrontation system,blind source separation of radar signals separates any number of aliasing radar pulse observation signals by using the independence and orthogonality between different transmitted signals to provide high-precision front-end information for sorting and identifying the signal of radiation source signals,so as to provide reliable support for the determination of the working status and characteristic parameters of the radar.With the rapid development of military industry and technology,MIMO radar with anti-stealth,high-resolution,multi-path clutter suppression and damage resistance has become a hot issue in radar field.However,the existing radar reconnaissance algorithms can only obtain the pulse stream of the MIMO radar signals which can not be used to improve the interference and the anti-interference performance.In view of the above problems,this paper uses the blind source separation algorithm in the radar signal reconnaissance,and focuses on the analysis of the separation method in underdetermined conditions.The main contents are summarized as follows:?1?For the three types of radar signal models,the features of time-domain,frequency-domain,time-frequency domain and their correlation characteristics are analyzed.And the discrete frequency coding radar signal is selected as the main research object.Discrete frequency coding radar signals have different frequencies in different time slots.In-pulse frequencies vary with the coding sequence,and sparse features exist in the time-frequency domain.This paper presents a detection method of single-source principal value of complex angular detection.By filtering outliers and noise points in time-frequency observed signals,the linear clustering characteristic of signals is improved.?2?In the process of estimating the mixing matrix,aiming at the selection of the number and location of initial cluster centers which affects the fuzzy mean clustering method,the data field uses the potential energy information to analyze the particle distribution,thus a data field aided fuzzy C-means clustering center selection method is proposed.Particle swarm optimization is introduced to improve iteration process,which can simultaneously search for more regions in the solution space of the objective function to be optimized,thus improving the estimation accuracy of the mixing matrix.The simulation results show the effectiveness of the proposed method.?3?In the process of recovering the source signals,the ₁l norm minimization algorithm based on sparse model is used to estimate the discrete frequency coding signals.However the signal with less sparseness can not reach the theoretical optimal value.Therefore,an improved frequency coding estimation algorithm for MMO radar signals is proposed,which realizes accurate recovery in the waveform dimension.Through the comparative analysis of average recovery signal to noise ratio and related characteristics,it shows the superiority of the proposed algorithm.