云环境中改进FCM和规则参数优化的网络入侵检测方法

Network intrusion detection method based on improved FCM and rule parameter optimization in cloud environment

针对云环境中的网络入侵检测问题,提出一种基于模糊推理的网络入侵检测方法。首先,利用互信息特征选择对样本特征进行降维。然后,利用提出的改进模糊C均值聚类(IFCM)方法对训练样本集进行聚类,根据各样本特征与集群的对应关系获得初始模糊规则库。接着,对每个规则的前件参数和后件参数进行调优,以此获得准确的规则库。最后,基于规则库对输入连接数据进行模糊推理,对其进行分类以实现入侵检测。在云入侵检测数据集上的实验结果表明,该方法能够准确检测出网络入侵,具有可行性和有效性。

Aiming at the network intrusion detection problem in cloud environment,a method of network intrusion detection based on fuzzy inference was proposed.Firstly,it used the mutual information feature selection to reduce the feature of the sample.Then,the improved fuzzy C-means clustering method was used to cluster the training sample set,and the initial fuzzy rule base was got by the correspondence between each sample feature and cluster.After that,the refine parameter and consequent parameters of each rule were tuned to obtain an exact rule base.Finally,fuzzy inference was carried out on the input connection data based on the rule base,and it was classified to realize intrusion detection.Experimental results on the cloud intrusion detection dataset show that this method can detect the network intrusion accurately,and it is feasible and effective.