中文摘要：

当不存在噪声不确定性时，频谱感知算法中能量检测法是较优的，但实际中噪声是无法确切估计的。针对这一问题，提出了一种适合多天线认知无线电系统的盲感知算法。该算法不需要主用户的任何先验知识，利用采样协方差矩阵的最大特征值和矩阵迹之比作为判决统计量来检测主用户信号。根据随机矩阵理论，当给定虚警概率时，可以求出判决门限的渐近式表达式。仿真结果表明，与同类算法相比，所提算法能获得较高的检测概率和较低的虚警概率，同时克服了能量检测法对噪声的敏感性。

英文摘要：

When there is no noise uncertainty, energy detection （ED） scheme is better among spectrum sensing methods. But in practice noises couldn＇t be exactly estimated. Aim to the noise uncertainty, this paper presented a blind sensing algorithm for multiple antennas cognitive radio system. The algorithm used the ratio of the maximum eigenvalue to trace （RMET） of the sampled covariance matrix as the statistic to detect the primary user signal, which did not require any prior knowledge of the pri- mary user. Given a target probability of false alarm, the asymptotic expression of the decision threshold was obtained using the random matrix theory. Simulation results show that, compared with similar algorithms, the proposed scheme have a higher proba- bility of detection and lower probability of false alarm, and overcome the sensitivity tO noise of energy detection scheme as well.