基于随机矩阵最大特征值分布的频谱感知算法

现有频谱感知算法在低信噪比时检测性能较低且受虚警概率影响大,针对此问题,提出了一种基于wishart矩阵样本协方差矩阵最大特征值的分布特性的频谱感知算法。该算法利用最大特征值与几何平均特征值的比值,不需要主用户的先验知识,不敏感于噪声,对相关信号和独立同分布信号均具有较高的检测性能。仿真结果表明,所提算法受虚警概率的影响较小,检测性能高,并且在采样点数、协作用户数、信噪比及虚警概率较小的情况下,也能获得较好的检测性能。

Spectrum sensing algorithm based on maximum eigenvalue distribution of random matrix

The sensing performance in existing algorithms is low at low SNR, and is highly influenced by the false-alarm probability. Aiming at this problem, using wishart random matrix theory, a spectrum sensing algorithm based on the distribution characteristics of the maximum eigenvalue of the sampled covariance matrix was proposed. The ratio of maximum eigenvalue and geometric mean eigenvalue is calculated in the algorithm, and no prior knowledges of primary signal are needed. Not sensitive to noise uncertainty, but the algorithm has high detection performance to the related signals and the independent and identically distributed signals. The simulation results show that the proposed algorithm is less affected by the false-alarm probability and high detection performance, and can also get better sensing performance even under the conditions of few number of cooperative users and samples, low SNR and false-alarm probability.