大数据网络数据规模巨大，对入侵过程痕迹数据进行监测的效率通常较低，一些带有入侵痕迹的数据特征在大数据环境下，特征逐渐淡化，当前方法无法在淡化的情况下准确采集痕迹数据的特点，无法形成待监测数据与痕迹数据之间的关系，导致监测效率和精度低下。提出一种基于模糊聚类概率的大数据网络入侵过程的痕迹数据监测方法，将采集的痕迹数据转换成频域信号，对其进行频谱或功率谱分析，依据时间变化的幅值将其转换成随频率变化的功率。采用核主元分析对痕迹数据信号特征进行提取，利用非线性转换将样本痕迹数据信号从输入空间映射至高维特征空间，在高维特征空间中通过PCA进行痕迹数据信号的频域特征提取。构建一个数学模型对特征模糊聚类概率进行描述，对待监测数据和痕迹数据之间的特征模糊聚类概率进行计算，通过衡量理论进行对比分析，使大数据网络入侵过程中的痕迹数据被完整的监测。实验结果表明，所提方法不仅所需时间少，而且监测精度高。%Big data network data size, traces the process of intrusion data monitoring efficiency is low, often some data with invasion of trace characteristics under the environment of big data, characteristic gradually fade out, under the condition of current method can,t played down the characteristics of accurate gathering trace data, unable to form for monitoring data and trace data, the relationship between the monitoring efficiency and low accuracy. A kind of big data network was put forward based on fuzzy clustering probability of process data monitoring method, the trace of the collected data into frequency domain signal, the spectrum and power spectrum analysis, according to the time change amplitude convert them to change with frequency power. Using Kernel principal component anal-ysis to trace data signal characteristic extraction, using nonlinear transformation to trace sample data signals from the input space is mapped to high-dimensional feature space, in the high dimensional feature space by PCA to trace data signal in the frequency domain feature extraction. Build a mathematical model to simulate the characteristics of fuzzy clustering probability description, treatment of monitoring data and trace data between the characteristics of the fuzzy clustering probability calculation, by comparing the measure theory makes big data in the process of net-work intrusion trace monitoring data is complete. The experimental results show that the proposed method is not on-ly less time required, and monitoring of high precision.
展开▼
