联合循环平稳特征PCA与XGBoost的频谱感知

SPECTRUM SENSING BY COMBINING CYCLOSTATIONARY FEATURES PCA WITH XGBOOST

针对低信噪比条件下主用户信号检测概率低的问题,提出一种基于循环平稳特征主成分分析与极限梯度提升算法(eXtreme Gradient Boosting,XGBoost)的主用户信号频谱感知算法。在信号各循环频率不为零值的情况下,提取能量最大的信号循环谱,通过PCA对循环谱特征进行降维处理,生成训练样本和测试样本。利用训练完成的XGBoost算法对待检测的信号进行分类,实现主用户信号是否存在检测。实验结果表明:与支持向量机算法、随机森林算法和传统循环谱算法相比较,该算法在低信噪比和低虚警率情况下具有更优的检测性能。

Aiming at the low accuracy of primary user signal detection under low SNR conditions,this paper proposes a spectrum sensing algorithm for primary user signals based on cyclostationary feature PCA and XGBoost.Under the condition that the cyclic frequency of the signal was not zero,the signal cycle spectrum with the largest energy was extracted,and the cyclic spectrum features were reduced by PCA to generate training samples and test samples.The trained XGBoost algorithm was used to classify the signals to be detected,so as to realize the detection of the presence of the primary user signal.The experimental results show that compared with support vector machine algorithm,random forest algorithm and traditional cyclic spectrum algorithm,the proposed algorithm has better detection performance in low signal-to-noise ratio and low false alarm rate.