Robust Frequency-Invariant Beamforming Method for Linear Constrained Minimum Variance Diagonal Loading

Abstract

The invention discloses a linearly constrained minimum variance diagonal loaded robust frequency-invariant beam forming method. On the basis of a linearly constrained minimum variance criterion, a balanced matrix of an array spatial response derivation function is defined through a spatial response derivation function; the matrix is introduced into a linearly constrained minimum variance beam forming method, such that a linearly constrained minimum variance wideband frequency-invariant beam forming method is obtained; and an iterative global optimal solution is obtained on a constrained boundary through a convex optimization toolbox CVX. By means of the method disclosed by the invention, the robustness when the microphone characteristic error exists is improved; the mismatch error problem due to uncertainty of a microphone, such as the gain, the phase and the position, is solved; and the method disclosed by the invention is widely applied in many fields, such as multi-channel voice enhancement, human-computer voice interactive systems, hearing-aids, vehicular hand-free voice communication, remote television conference systems and robot hearing.

Description

technical field [0001] The invention relates to the technical field of robust microphone array beamforming, in particular to a robust frequency-invariant beamforming method for linearly constrained minimum variance diagonal loading. Background technique [0002] Microphone array beamforming technology originates from the idea of ​​array antenna beamforming. In order to improve the output performance of the array, many classic beamforming methods have been proposed one after another. The traditional linear constrained minimum variance beamforming method is to make the main lobe of the array antenna pattern point to the direction of the desired signal, and make its null point to the direction of the interference signal, so as to improve the strength of the desired signal output by the array and reduce the strength of the interference signal , thereby improving the output performance of the array. However, this method is mainly used to design the weight vector of the narrowba...

AI Technical Summary

Problems solved by technology

[0003] In order to realize broadband frequency-invariant beamforming, scholars have made some attempts, among which, the least squares frequency-invariant beamforming method has too wide a passband range, and the desired signal resolution is too low; the super-directional broadband frequency-invariant beamforming method The conformal array structure is required, and the calculation is complicated; the spherical array broadband beamforming method requires a spherical array structure, which is not easy for physical realization and practical application; and in the actual application process, there are usually uncertainties caused by the gain, phase and position of the microphone. Mismatch errors, which cause distortions in the array response vector and degrade the performance of the beamformer

Images