基于循环谱特征的楼宇室内频谱感知算法

针对目前楼宇室内环境中,信道多径衰落和噪声不确定性等低信噪比情况下主用户信号检测性能较低的问题,提出了一种基于支持向量机(SVM)的主用户信号频谱感知算法.该算法融合了循环平稳特征检测和SVM算法的特点,对信号循环平稳特征参数进行特征提取,作为训练样本和待测样本,再采用SVM算法分别对有无主用户情况下的信号进行分类检测.仿真实验表明与人工神经网络(ANN)和最大最小特征值法(MME)相比较,所提算法可在低信噪比情况下,有效地实现对主用户信号的感知,具有较好的稳健性.

Spectrum sensing algorithm based on cyclic spectrum characteristics in building indoor environment

According to the low accuracy rate of the primary user signal detection in the building indoor environment at the situation of low SNR, such as channel multipath fading and noise uncertainty, etc., a method based on support vector machine (SVM) for the primary user spectrum sensing is proposed. The method combines cyclostationary characteristic method and SVM. Characteristics of cyclostationary characteristic parameters are extracted from signals as training samples and testing samples. Then, signals with and without the primary user are classificatorily detected by SVM. The results of simulation experiments show that the proposed algorithm achieves a good spectrum sensing and robustness compared with artificial neural network (ANN) and maximum-minimum eigenvalue (MME) at low SNR.