To keep the network secure, it is necessary to monitor network traffic timely and effectively. The traditional methods for detecting network anomalies were mainly based on such ways as sampling, counting and aggregating, but they can not solve the problem of getting accurate and effective results well. In this paper we propose a new method that is based on the basic properties of frequent pattern mining problem and makes use of the vertical mining methods to mine frequent patterns from network traffic. Based on this algorithm, we build a prototype system to evaluate our algorithm on huge net flow data of campus network. The experimental result shows that this algorithm can detect network anomalies timely and effectively and can help network administrators achieve more effective monitoring on network.
展开▼
