The Study On Interference Cancellation And Direction Of Arrival Estimation Based On Blind Source Separation

Blind source separation(BSS)technology is widely used in wireless communication,biomedical signal processing,speech enhancement tasks,etc.This paper focuses on the joint(block)diagonalization(JBD)BSS algorithm and its application in the fields of interference cancellation and direction of arrival(DOA)estimation.The main research work of this paper is summarized as follows:1.Blind source separation for interference cancellation based on phased array radar:Estimation of Signal Parameters via Rotational Invariance Techniques(ESPRIT)is usually used to estimate the initial value of the mixing matrix in the traditional instantaneous BSS algorithms,but the traditional initial value estimation algorithm cannot perfectly match the instantaneous mixing model at the sub-array level of phased array.In order to solve the problem,an energy detection-based mixing matrix initial value estimation method is proposed.This method maintains a high degree of matching with the linear instantaneous mixing model at the sub-array level of phased array.Besides,in order to solve the problem that weak target signals are difficult to be detected,a two-stage BSS method for interference cancellation is proposed.Two techniques are used to realize the rapid detection of salient target and the accurate detection of weak target: one is the combination of rough and meticulous search strategy;the other is the combination of signal processing in sampling time dimension and pulse time dimension.Experimental results show that compared with interference cancellation algorithms,the algorithm in this paper can improve the interference-noise canceling ratio(INCR)by 7d B.2.The convolutive BSS of speech signal based on microphone array: it is necessary to pre-whiten the target matrix group for the orthogonal joint block diagonalization(JBD)convolution BSS algorithm,and the frequency domain convolutive BSS algorithm has serious scale and arrangement ambiguities.In order to solve problem,a non-orthogonal joint block diagonalization method(DBIA-JBD)that can be solved directly in the time domain is proposed.The target matrix group of Toeplitz structure between blocks and diagonalization structure within blocks is innovatively constructed.The convolutive symmetric fitting cost function matching the structure is designed,and the Bi-Iterative algorithm is used to optimize the solution.The optimization algorithm not only guarantees the efficiency of the algorithm,but also takes into account the performance of BSS.Simulation experiments show that the proposed algorithm has the characteristics of low computational complexity,good convergence and good separation performance.3.Angle estimation at sub-array level based on large phased array radar: A fast angle estimation algorithm based on fitting echo wave surface function is proposed.This method mainly solves the problem of multi-target angle estimation under strong interference of large phased array.First,the wavefront shape of radar echo signal in synthetic beam domain is analyzed.Then,the parameters of the mixing matrix solved by the BSS algorithm are used to fit the wavefront function of the echo.Finally,the peak point of the wavefront is solved to estimate the angle of the target.Compared with the sum-difference beamforming angle estimation method,the proposed algorithm has the advantages of multi-target angle estimation in the main lobe,anti-strong interference and anti-track splitting.The experimental and simulation results show the effectiveness and superiority of the method.