无线通信网络入侵信号特征自动提取模型研究

Research on Automatic Feature Extraction Model of Intrusion Signal from Wireless Communication Networks

针对特征提取精准性不佳的问题,提出无线通信网络入侵信号特征自动提取模型。本文优化聚类入侵信号,定位分离入侵小信号,构建过滤模型,处理损失能量。结合新增加时间、信号衰减特征量分段处理信号,并依次标记子段序号。然后,根据小波变换确定分解层数,计算层次相对衰减率,提取复合特征向量。通过Hilbert变换提取预处理信号的包络,构建特征自动提取模型。实验结果表明模型剪网入侵信号电压波动范围为[-0.5 V-3 V],攀爬入侵信号电压波动范围为[-4.8 V-4.8 V],与实际情况一致,使用该方法提取结果精准。

Aiming at the poor accuracy of feature extraction, this paper proposes a wireless communication network intrusion signal feature automatic extraction model. This paper optimizes the clustering of intrusion signals, locates and separates intrusion small signals,builds a filtering model to deal with the loss of energy. Combined with the newly added time and signal attenuation characteristics, the signal is processed in segments, and the sub segment number is marked in turn. According to the wavelet transform, the number of decomposition levels is determined, the relative attenuation rate of the levels is calculated, and the composite feature vector is extracted. The envelope of the preprocessed signal is extracted by Hilbert transform, and the automatic feature extraction model is constructed. The experimental results show that the voltage fluctuation range of the network clipping intrusion signal is[-0.5v-3v], and the voltage fluctuation range of the climbing intrusion signal is [-4.8v-4.8v], which is consistent with the actual situation. The extraction results using this method are accurate.