Improving the Security of Cyber Networks Based on Community Detection Using Spectral Clustering Algorithm

Cyber networks are considered to be a type of complex and free-scale networks due to the structure and mode of communication within the network. Identifying communities is one of the most important methods of network analysis in order to understand the structure and relationships between network members. With the development of cyber networks, new challenges have been created for users in terms of information securityOne of the goals of identifying the communities in cyber networks is to prevent the spread of malware and cyber attacks. For this purpose, in order to prevent and deal with network attacks and intrusions, the communities in the network should be identified in order to significantly reduce the damage and attacks by attackers by securing and reviving the communities as well as implementing defense policies appropriate to each community. In this article, a method for detecting cyber communities by spectral clustering algorithm is presented. Also, by using the property of the normalized Laplace matrix in this algorithm, it is possible to predict the number of suitable cyber communities. In order to evaluate the detection process, two criteria Silhouette Value and Jaccard index are used. The results obtained from the evaluation criteria confirm the effectiveness of the proposed method.