【摘要】 针对在低信噪比情况下,主用户信号检测率较低的问题,提出了一种新颖的基于随机森林的频谱感知算法。随机森林算法组合多个弱分类器,使整体分类效果增强,减少过拟合现象。首先,在各循环频率不为零值情况下,提取能量最大的信号循环谱中特征参数(E(S)、D(S))作为随机森林的训练特征值;其次,选取主用户存在下的若干特征值作为正样本和主用户不存在下的若干特征值作为负样本生成随机森林;最后,利用训练完成的随机森林对待检测的信号进行分类,实现主用户是否存在的检测。实验结果表明:所提出的算法有较高的检测率和较低的虚警率。更多还原

【Abstract】 Aiming at the problem that the detection rate of primary user signal is very low in the circumstance of low signal-to-noise ratio,a novel spectrum sensing method based on random forest is proposed. The random forest algorithm is a combination of multiple weak classifiers,which can reduce overfitting and improve the overall classification results. Firstly,the characteristic parameters( E( S) and D( S)) of the signal cyclic spectrum with maximum energy are extracted under the non-zero cycle frequency condition,and used as the training eigenvalues for the random forest. Secondly,some eigenvalues when the primary users use the channel are used as positive samples,some eigenvalues when the primary users do not use the channel are used as negative samples,and the random forest is created. Finally,the input signals are classified by the trained random forests to detect the presence of the primary user. The experiment results show that the proposed algorithm has high detection rate and low false alarm rate.更多还原