现有的基于特征值的合作频谱感知方法要求认知用户各感知节点接收到的授权用户信号具有相关性。针对这个问题,提出了一种基于广义特征值的合作频谱感知方法。该方法利用过去不存在授权用户的感知周期采样协方差矩阵与当前感知周期采样协方差矩阵之间的最大广义特征值(MGED,maximum generalized eigenvalue de—tection)作为检验统计量,以此判决当前感知周期是否存在授权用户信号,从而实现频谱感知。所提方法不需要授权用户信号和噪声功率的先验信息。当认知用户各感知节点上的授权用户信号不相关时,现有的基于特征值的频谱感知方法均失效,而所提频谱感知方法仍然具有较高的检测性能。最后仿真验证了所提方法的有效性。
The available eigenvalue based cooperative spectrum sensing detectors require that the primary signals among sensing nodes were correlated. A maximum generalized eigenvalue based cooperative spectrum sensing detector was proposed. The proposed detector exploited the received signals from the previous sensing durations without primary user. The maximum generalized eigenvalue between the sample covariance matrices from current and previous sensing dura- tions was taken as test-statistic to implement spectrum sensing. No prior knowledge of primary signal and noise power was needed in the proposed detector. The proposed detector still has high detection performance while the primary signals among sensing nodes are uncorrelated, although the available eigenvalue based detectors fail. Finally, the validity of the detector is proved by simulations.