作者: Mohsen Rezvani , Verica Sekulic , Aleksandar Ignjatovic , Elisa Bertino , Sanjay Jha
DOI: 10.1109/TIFS.2015.2455414
关键词:
摘要: Detection of high risk hosts and flows continues to be a significant problem in security monitoring throughput networks. A comprehensive assessment method should consider the propagation among risky flows. In this paper, is achieved by introducing two novel concepts. First, an interdependency relationship scores network flow its source destination hosts. On one hand, score host depends on initiated or terminated at host. other Second, which we call provenance, represents considers likelihood that particular caused Based these concepts, develop iterative algorithm for computing We give rigorous proof our rapidly converges unique estimates, provide extensive empirical evaluation using real-world data sets. Our shows effective detecting sufficiently efficient deployed