A Source Number Estimation Method Based on Improved Eigenvalue Decomposition Algorithm

In order to seperate signal subspace from noise subspace accurately, the super-resolution DOA estimation algorithms based subspace need to obtain the number of signal sources first. In this paper, an improved eigenvalue decomposition algorithm is proposed to estimate the number of signal sources. Firstly, the sample covariance matrix calculated from received data is modified by matrix reconstruction. Then, the number of signal sources is estimated by eigenvalue decomposition algorithm. Simulation results show that the proposed algorithm has better detection accuracy and reliability than traditional methods in the case of low signal-to-noise ratio (SNR) and coherent signal. The proposed algorithm complexity is low. Based on the above advantages, the improved eigenvalue decomposition algorithm is a promising method for source number estimation.