Spatial Spectrum Estimation Based On Subspace Fitting Algorithm

The subspace fitting algorithm is an important algorithm in the DOA estimation.It is widely used in radar,military,communications,medical,and other fields because of its advantages such as high precision and resistance to multipath.The weighted subspace fitting(WSF)algorithm is a typical representation of the subspace fitting algorithm,and it is well received by many scholars.However,because of the multi-dimensional nonlinear search process,the algorithm has high computational complexity and is not suitable for application in real world.Therefore,how to reduce the computational complexity becomes a key issue of the WSF algorithm.Based on the existing researches,this topic analyzes the reasons for the high computational complexity of the WSF algorithm and combines the spectral peak characteristics of the WSF algorithm and proposes a modified WSF algorithm(CSWSF)by limiting the GA search space.The CSWSF algorithm uses the rotational invariant subspace algorithm and the theoretical minimum error of the unbiased estimator to bound the search space of the algorithm.The algorithm effectively confines the search space to a bounded square region containing the WSF global optimal solution,and then uses the modified genetic algorithm on the shrunk search region.Therefore,CSWSF algorithm can effectively reduce the computation time of the algorithm by limiting the WSF search space.Based on the theory of the CSWSF algorithm,by introducing the dynamic continuous constraint search space search strategy,a new algorithm based on narrow-band weighted subspace fitting(NBWSF)was proposed.The NBWSF algorithm dynamically plans multiple circular search spaces along the gradient direction and gradually approaches the search space to the global optimal solution of the WSF.The NBWSF algorithm narrows the space for solving the WSF algorithm by dynamically limiting the search space,which further reduces the computation time.In this topic,the CSWSF algorithm and the NBWSF algorithm are simulated using MATLAB,and compared with common WSF optimization algorithms.The simulation results show that under the same target accuracy,the real-time performance of the proposed algorithm is better,and the computational complexity is significantly reduced.With 0dB SNR and 4 sources,the computation time of the NBWSF algorithm is about 30% lower than that of the CSWSF algorithm.