A new connection degree calculation and measurement method for large scale network monitoring

Traffic pattern characteristics monitoring is useful for abnormal behavior detection and network management. In this paper, we develop a framework for connection degree calculation and measurement in high-speed networks. The bi-directional traffic flow model is employed to aggregate traffic packets, which can reduce the number of flow records and capture user's alternation behavior characteristics. The first order connection degree and joint correlation degree are selected as the features to capture the characteristics of traffic profiles. To perform careful traffic inspection and attack detection, not only the abnormal changes of a single traffic feature but also the correlations between the features are analyzed in the new framework. First, the symmetry of in and out connection degrees is analyzed. And we found that incomplete flows are an important information source for abnormal behavior detection. Second, joint correlation degree can characterize the user's communication profiles and their behavior dynamics, which are employed to perform abnormal detection using measurements based on Renyi cross entropy. Finally, the reversible degree sketch is employed for querying abnormal traffic pattern sources for real-time traffic management. The experimental results based on actual traffic traces collected from Northwest Regional Center of CERNET (China Education and Research Network) show the efficiency of the proposed method. The method based on Renyi entropy can detect abnormal changing points correctly. FNR of the reversible sketch for locating abnormal sources is below 4% and time complexity is constant and less than 4 s, which is critical for real-time traffic monitoring. Crown Copyright (c) 2013 Published by Elsevier Ltd. All rights reserved.