基于随机矩阵的新型频谱盲感知方法

A Novel Blind Spectrum Sensing Algorithm Based on Random Matrix

针对传统频谱感知算法需要预先估计噪声方差且当存在噪声不确定度时,检测性能降低的特点,提出一种基于随机矩阵的改进型频谱盲感知算法(M-CMME)。该算法通过分析协方差矩阵最大特征值极限分布特性,分析并利用采样协方差矩阵特征值与信号平均能量的关系,推导设定虚警概率条件下判决门限的闭式表达式。该算法不需要预先知道授权用户信号的先验知识,且能够有效克服噪声不确定度的影响。仿真结果显示,当噪声方差估计存在偏差的情况下,该算法具有较强的鲁棒性,且在较少采样点、低信噪比、较少阵元数情况下能够获得比CMME更优的检测性能。

In order to improve the detection performance of traditional spectrum sensing under low SNR,a spectrum sensing algorithm（ M-CMME） is proposed. The proposed algorithm analyzes the characteristic of limiting eigenvalue distribution,analyzes and utilizes the relation of energy and eigenvalues of sample matrix and then deduces the form expression of decision threshold under constant false alarm ratio. The algorithm,we believe,does not need estimating the noise power and exhibits a good robustness against noise uncertainty. Simulation results show preliminarily that,when there is a deviation of the noise estimation,the algorithm can obtain strong robustness and this algorithm can get better detection performance than CMME with fewer samples,lower SNR and fewer antennas.