基于Gibbs-LDA和最小二乘支持向量机的物联网安全预测方法

Method for Safety Predicting of Internet of Things Based on Gibbs-LDA Model and LSSVM

摘要:

针对物联网中各类用户的网络行为出现复杂化、多样化和恶意化的特征和趋势,提出了一种基于Gibbs—LDA和最小二乘支持向量机的物联网安全预测方法;首先,提取通信时间、地址和内容等文中信息作为多维的通信记录样本,然后基于LDA模型,将安全事件建模为主题,获取样本特征并得到主题模型,通过Gibbs算法来估算LDA模型中的参数,从而建立了基于LDA的物联网安全多维预模型,最后,在LDA特征空间上建立了特征与安全事件分布的权重,并将此权重用于初始化各个支持向量机的预测结果,将权值最大的最小二乘支持向量的预测结果作为最终的结果;仿真实验证明了文中方法能有效地实现物联网安全预测,在NIPS和VAST数据集上进行仿真实验,结果表明了文中方法较其他方法具有预测精度高和预测时间短的优点,具有较大的优越性。

Abstract:

Aiming at the feature and tendency the complexity, diversity and malignity of various users in Internet of things, a predicting method for Internet of things safety based on Gibbs-LDA and least square SVM is proposed. Firstly, the communication time, address and content is regarded as the original multi-dimension communication sample, then based on the LDA model, the safety event is built as the theme to obtain the sample feature and the LDA model. The Gibbs algorithm is used to estimate the parameters of LDA model, therefore, the Internet of things safety multi-dimension model is built. Finally, the weight distribution of feature and safety event over the LDA feature space is obtained, and the least square SVM with the biggest weight is used as the final predicting result. The simulation experiment shows the method in this paper can realize the predicting for network security event for Internet of things with higher predicting accuracy, and compared with the other methods, it has higher predicting accuracy and less predicting time, therefore, it has big priority over the other methods.