Spectrum sensing method based on covariance detection

The invention provides a spectrum sensing method based on covariance detection. The method comprises the following steps: S1, constructing covariance detection statistics of a plurality of mobile data signals; s2, taking the constructed covariance detection statistics and the detection statistics of the covariance absolute value detection method as two-dimensional feature vectors of a support vector machine, training to generate a classifier, and obtaining a frequency spectrum classification model; and S3, inputting the two-dimensional feature vector of the to-be-detected signal into the spectrum classification model for judgment to detect whether a user signal exists in a channel, and compared with a classical covariance detection method, the method provided by the invention has better detection performance and higher reliability under a low signal-to-noise ratio. 本文提出了一种基于协方差检测的频谱感知方法，包括以下步骤：S1:构建多个移动数据信号的协方差检测统计量；S2、把构建的协方差检测统计量和协方差绝对值检测方法的检测统计量作为支持向量机的二维特征向量，训练生成分类器，得到频谱分类模型；S3、将待测信号的二维特征向量输入到频谱分类模型中进行判决，