Identification of Network-wide Super Nodes in High Speed Networks
Aiping Zhou, Qi Zhao, Jin Qian, Chengang Zhu · 2019
Identifying super nodes is to determine hosts with large connection number in the measurement interval. This is a basic problem with many practical applications, such as traffic measurement and security detection. Due to massive amounts of network traffic data and source limitation, it is a large challenge to identify super nodes in high speed networks efficiently. To solve the problem, we propose a novel solution to identify super nodes in high speed networks, called ISN. It constructs a novel data structure, where bit arrays shared bit vectors, and only needs some simple computing. Moreover, it doesn’t use extra memory to generate bit arrays. So that the estimation precision and identification accuracy of our algorithm is improved obviously. Our method mainly consists of data updating, estimation of connection number and super node identification. The experiments are conducted on real traffic traces and their results show that our algorithm is superior to the related ones in terms of estimation precision, identification accuracy, and memory utilization.