A method to improve the detection sensitivity of eigenvalue signals in the environment of low signal-to-noise ratio

Abstract

The invention discloses a method for improving the detection sensitivity of a characteristic value signal in a low signal to noise environment. The method comprises the following steps: correlating asampling signal matrix X with a to-be-tested signal s, wherein the method is to add X with a signal matrix having a certain amplitude (the formula is described in the specification), the matrix (the formula is described in the specification) is composed of (the formula is described in the specification) signals related to the to-be-tested signal s (the (the formula is described in the specification) signals related to the to-be-tested signal s are called characteristic value displacement signals), so that a characteristic value signal lambda s corresponding to the to-be-tested signal s in theobtained matrix (the formula is described in the specification) is distributed in an area (the formula is described in the specification), so that the presence of a target signal component is judged easily. According to the method disclosed by the invention, the statistical characteristics of the data are adopted, the method is suitable for the processing of distributed big data, no data synchronization is required, the detection method is simple, and higher detection performance can be obtained. Under the condition that the characteristic parameters of the to-be-tested signal cannot be completely determined, related signal traversal can also be performed according to the parameter range of the to-be-tested signal to obtain high-sensitivity signal discovery.

Description

technical field [0001] The invention relates to signal detection, in particular to a method for improving detection sensitivity of characteristic value signals in a low signal-to-noise ratio environment. Background technique [0002] Signal detection and estimation are widely used in communication, radar, fault diagnosis, security and many other fields. Signal detection based on random matrix theory is an efficient and high-sensitivity detection technology based on the statistical characteristics of big data, and through the dimensionality reduction processing of big data, the computational complexity is greatly reduced. Existing signal detection methods in this field include maximum eigenvalue detection, minimum eigenvalue detection, trace detection, logarithmic determinant detection, eigenvalue higher-order moment detection, eigenvalue matching detection and other methods of sampling covariance matrix. [0003] According to the random matrix theory, for a high-dimensional...

AI Technical Summary

Problems solved by technology

[0010] In the latter two cases, it is difficult to detect due to low-amplitude signal eigenvalues ​​submerged in the M-P rate and Tracy-Widom distribution characteristics

[0011] Although the existing maximum eigenvalue detection uses λ s The Tracy-Widom distribution distributed near the edge b of the M-P rate performs signal resolution, but the sensitivity is not ideal

Images