Hidden markov models hmms for communcations networks

Info icon This preview shows pages 77–79. Sign up to view the full content.

View Full Document Right Arrow Icon
Hidden Markov models (HMMs) for communcations networks are dis- cussed in Section 3.4. Salamatian and Vaton (2001) discuss using HMMs to examine packet loss in the internet. The data we work with, however, is internal network data, and the challenges of modeling edge data are entirely different than those of modeling highly aggregated flows over the internet. There is some interesting work on identifying groups in social networks using HMMs in the literature (Baumes et al. , 2004, 2006). While these methods are geared to social networks, do not use edge data, and use prelabeled examples, not anomaly detection, we feel this work to be inspirational, as it tackles many similar issues. Finally, Ye et al. (2000) present a Markov model for audit data from Unix machines to perform node-based anomaly detection. No network modeling is done, and focusing on a set of Unix machines is no longer of particular interest, since modern networks have a variety of operating systems, and a general approach to network anomaly detection cannot focus on a single operating system. Paths have been examined in the context of vehicular traffic in Lu et al. (2009), using a similarity metric to compare paths, and then clustering to find outliers. This method, however, assumes we observe path values that can be clustered. On the contrary, in this chapter we propose a statistically rigorous method to infer anomalous shapes from the network without any prior knowledge about traversal by individual actors. Many of the methods mentioned above are intended for much smaller graphs than our method proposes to address. We have a data set that is difficult to come by: a record of all of the communications between individ- ual computers on a large corporate-sized network. These communications are recorded at fine timescales (1 second or finer), and have been archived for the past decade in some cases. The objects (paths) we monitor num- ber in the hundreds of millions per 30-minute window, which we are able to test in under 5 seconds. With the exception of the telephone network literature (Lambert et al. , 2001; Lambert and Liu, 2006) (that does not monitor at a graph level, but at individual aggragation points), the sheer size of this endeavor separates it from most other work. The formal statement of the scan statistic is given in Section 3.2. Inde- pendence between the edges within a path is covered in Section 3.3. Mod- eling of edge data is discussed in Section 3.4. Stars and paths are compared Copyright © 2014. Imperial College Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted under U.S. or applicable copyright law. EBSCO Publishing : eBook Collection (EBSCOhost) - printed on 2/16/2016 3:37 AM via CGC-GROUP OF COLLEGES (GHARUAN) AN: 779681 ; Heard, Nicholas, Adams, Niall M..; Data Analysis for Network Cyber-security Account: ns224671
Image of page 77

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
78 J. Neil, C. Storlie, C. Hash and A. Brugh on a variety of simulations in Section 3.5. Finally, we present results of scanning on actual computer network data in Section 3.6.
Image of page 78
Image of page 79
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

What students are saying

  • Left Quote Icon

    As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

    Student Picture

    Kiran Temple University Fox School of Business ‘17, Course Hero Intern

  • Left Quote Icon

    I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

    Student Picture

    Dana University of Pennsylvania ‘17, Course Hero Intern

  • Left Quote Icon

    The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

    Student Picture

    Jill Tulane University ‘16, Course Hero Intern