Network Monitoring Based on Community Structure Change

Based on the theory of complex networks, Internet is regarded as a directed and weighted network, hosts within Internet is defined as a node in network, edge represents packets between hosts. A novel method is presented for monitoring network analogy based on group behavior change. Specifically, we apply our methodology using data collected from a one point of observation where they are campus network boundary and backbone network link. First, we study the similarity of the host behavior. Second, we evaluate group behavior stability, that is, the stability of community structure. At last, we apply the proposed approach to monitor anomalous group behavior during the Slammer worm spread, the experimental results show that our approach can successfully detect Slammer worm outbreak.