Self-adaptive air filtering method based on iterative shrinkage weighted fusion

Abstract

The invention discloses a self-adaptive air filtering method based on iterative shrinkage weighted fusion. The method comprises the steps that an array antenna receives data to build a signal model, and a sampling covariance matrix of the received data is obtained; a self-adaptive iteration way is adopted to update a priori covariance matrix on line; weighting coefficients of the sampling covariance matrix and the priori covariance matrix are calculated according to the minimum mean-squared error criterion, and an estimated covariance matrix is obtained through the shrinkage weighted fusion processing method; finally, the self-adaptive weight vector is calculated to carry out air filtering. The self-adaptive air filtering method can estimate the covariance matrixes with high precision on the condition of small samples, can effectively relieve the priori knowledge and current data model mismatch problem, avoids subspace dimensionality determination, has the advantages of being high in output signal to interference plus noise power ratio and convergence rate, and is effective for self-adaptive air filtering actual application.

Description

technical field [0001] The invention belongs to the field of signal processing, and relates to array signal processing technology, in particular to an adaptive air filtering method using shrinkage weighted fusion and based on iterative update, which is used to estimate the covariance matrix with high precision under the condition of small samples , to improve the signal-to-interference-noise ratio output by the antenna array. Background technique [0002] Array signal processing is a hot research direction in the field of signal processing. It is widely used in radar, sonar, communication, earthquake monitoring and other fields. Adaptive air filter is an important research content of array signal processing. Its purpose is to enhance The target signal power is suppressed at the same time as interference and noise, thereby improving the signal-to-interference-noise ratio of the array antenna output. The essence of adaptive air filtering is to adaptively weight each array ele...

AI Technical Summary

Problems solved by technology

The covariance matrix inversion (SMI) algorithm is a commonly used and effective algorithm, but when the array data contains expected signals, the SMI algorithm will cause a serious decrease in the output signal-to-noise ratio; subspace projection algorithms are not suitable for environments with strong expected signals , the filtering effect is better, but this type of algorithm needs to accurately estimate the signal subspace and noise subspace, and the subspace estimation accuracy is limited under the condition of small samples; the diagonal loading algorithm adds a diagonal matrix to the covariance matrix, which can Effectively overcome the distortion of the array pattern caused by the small eigenvalue disturbance of the array covariance matrix, but how to select the diagonal loading is not easy to determine; Griffiths et al. published the article "An In "alternative approach to linearly constrained adaptive beamforming", a generalized sidelobe canceller (GSC) is proposed, which can overcome the signal cancellation problem caused by the desired signal contained in the covariance matrix in the SMI algorithm, but due to the existence of antenna array errors, When the signal-to-noise ratio is high, the problem of signal cancellation will also occur, resulting in a decrease in the output signal-to-interference-noise ratio; Goldstein et al. published the article "A multistage representation of the wiener filter based on the IEEE Trans.Information theory in 1998 In "orthogonal projections", a reduced-rank multi-level Wiener filter based on the generalized sidelobe canceller framework is proposed. This method does not require eigenvalue decomposition and has low computational complexity. However, this method needs to determine the dimension of the processor. In the case of small samples, the estimation accuracy of the processor dimension is not high

Images