Sponsor
Portland State University. Department of Computer Science
Document Type
Technical Report
Publication Date
2009
Subjects
Malware (Computer software) -- Prevention, Algorithms, Computer networks -- Security measures
Abstract
We present an algorithm for detecting IRC-like chat networks that does not rely on Layer 7 payload information. The goal is to extract only those meshes from conventional flows where long-term periodic data is being exchanged between an external server and multiple internal clients. Flow data is passed through a series of filters that reduce the memory requirements needed for final candidate mesh sorting. Final outputs consist of two sorted lists including the fanout list, sorted by the number of client hosts in the mesh, and a secondary list called the evil sort. The latter consists of meshes with any host with a high TCP work weight1 [3] indicating significant counts of scanning hosts. We are currently able to discover SSL-encoded IRC meshes as well as other chatlike meshes including MSN chat. Therefore we believe that the new algorithm will prove useful in detecting botnet meshes encrypted at Layer 7.
Persistent Identifier
https://archives.pdx.edu/ds/psu/30755
Citation Details
Dua, Akshay; Binkley, Jim; and Singh, Suresh, "Finding IRC-like Meshes Sans Layer 7 Payloads" (2009). Computer Science Faculty Publications and Presentations. 221.
https://archives.pdx.edu/ds/psu/30755
Description
Portland State University Computer Science Department Technical Report #09-01, 2009.