Parameter estimation algorithm based on tensor subspace

Two methods of parameter estimation algorithm based on tensor subspace are proposed to improve the accuracy of signal parameter estimation in low signal-to-noise ratio, which are the high order singular value decomposition algorithm based on matrix stack and the algorithm based on tensor decomposition factor. According to the parametric signal model, the shift invariance property of the signal subspace and the principle of tensor Vandermonde decomposition, the construction forms of tensor model are discussed. The high order singular value decomposition algorithm based on matrix stack verifies that the tensor high order singular value decomposition is based on the matrix singular value decomposition. The algorithm of tensor decomposition factor may further improve the accuracy of signal parameter in the case of low signal-to-noise ratio. The accuracies of frequency and phase may serve as an index of the algorithms. The simulations verify that the algorithm based on tensor decomposition is superior to traditional matrix-based decomposition algorithm in low signal-to-noise ratio.