认知无线电中基于拟合优度的多天线协作频谱检测

Cooperative spectrum sensing based on Goodness of Fit in Multi-antenna cognitive radio

本文提出一种噪声方差未知时,基于拟合优度的多天线协作频谱检测方法。通过计算多个天线采样样本均值和方差比的分布函数与经验分布函数之间的Anderson-darling(AD)距离,并与备选的门限值比较,实现频谱检测。论文分析了衰落信道下,基于AD检验的协作频谱检测的虚警概率和检测概率下界。理论分析和仿真结果表明,基于AD检验的协作检测比基于随机矩阵的协作检测具有更好的性能,特别是在低信噪比和小样本条件下。

By applying Goodness of Fit testing in statistics into cooperative spectrum sensing, this paper gives a new cooperative sensing of existence of unknown signals in Multi-antenna cognitive radio fading channel. First the ratio of the mean value to the variance of received samples from multi-antenna is calculated, then the Anderson-Darling distance between the empirical distribution and the cumulative distribution of the ration is calculated. The calculated distance is compared with a given level for spectrum sensing. The performance of the proposed Anderson-Darling based cooperative sensing is analyzed in the flat fading channel, while the false alarm probability and the lower bounder of detection probability are also got. Both analysis and numerical results show that AD based cooperative sensing has much higher performance than random matrix theory（RMT）based cooperative sensing, especially in a case where the received signal has a low signal to noise ratio （SNR） or a small sample numbers.