基于采样协方差矩阵的混合核SVM高效频谱感知

近年来随着盲检测算法的提出，越来越多的基于采样协方差矩阵的盲检测算法应用于频谱感知。针对其检测门限是近似值，检测性能会受到影响等问题，提出了基于采样协方差矩阵的混合核函数的支持向量机（support vector machine,SVM）高效频谱感知，通过感知信号采样协方差矩阵的最大最小特征值（maximum minimum eigenvalue,MME）和协方差绝对值（covariance absolute value,CAV）提取的统计量作为SVM的特征向量并训练其生成频谱感知的分类器，无需计算检测门限并且特征提取减少了样本集的大小。利用遗传算法（genetic algorithm,GA）优化混合核函数的SVM的参数。实验结果表明，该方法比MME算法和CAV算法的检测概率有所提高，并且比SVM减少了感知时间，具有良好的实用性。

An efficient spectrum sensing of mixed kernel SVM based on sampling covariance matrix

In recent years,with the blind detection algorithms were proposed,more and more blind algorithms based on sampling covariance matrix were applied to spectrum sensing.The detection threshold was an approximation,and the detection performance would be affected for this algorithms.Thus,the mixed kernel function support vector machine (SVM) efficient spectrum sensing based on sampling covariance matrix was proposed.The statistics which were maximum minimum eigenvalue (MME) and covariance absolute value (CAV) of sensing signal sampling covariance matrices were used as the feature vectors of SVM and were trained to generate a spectrum sensing classifier.The advantage of this algorithm was that it needn’t calculate the detection threshold and the extraction of features reduces size of the sample set.The genetic algorithm (GA) was used to optimize the parameters of mixed kernel function SVM algorithm.The experimental results show that the proposed method has higher detection probability than MME and CAV algorithms,and has less sensing time than SVM,which has good practicability.