一种基于时频谱熵特征的频谱感知算法

A Spectrum Sensing Algorithm Based on Time-Frequency Spectrum Entropy Feature

提出一种基于时频谱熵特征的频谱感知算法。充分利用噪声与授权信号谱熵数值的差异，将归一化处理的信号时频谱熵值作为检测统计量，选择噪声谱熵值作为判决门限检测信号。该算法无需信号先验信息和噪声功率信息，能够克服能量检测由于噪声不确定性所带来的不利影响，且实现简单，运算复杂度低。仿真结果表明在高斯白噪声信号环境下，基于时频谱熵特征的频谱感知算法优于分形盒维数检测和能量检测方法，尤其提高了低信噪比情况下的频谱检测率。

A spectrum sensing algorithm based on time-frequency spectrum entropy feature is presented. Making use of the difference of spectrum entropy between noise and signals mostly, the normalized time-frequency spectrum entropy is used as detection statistics, noise spectrum entropy is set as decision threshold in this algorithm. The algorithm can overcome the disadvantage effect because of noise uncertainty in the energy detection algorithm, without knowing any priori information of signal and noise power, meanwhile it is easy to carry out and low computational complexity. Simulation results show that spectrum entropy detection algorithm performs better than fractal box dimension detection and energy detection under Gaussian white noise environment, especially higher detection probability in the condition of low SNR.