Name
Abstract | ||
|---|---|---|
Software lineage refers to the evolutionary relationship among a collection of software. The goal of software lineage inference is to recover the lineage given a set of program binaries. Software lineage can provide extremely useful information in many security scenarios such as malware triage and software vulnerability tracking. In this paper, we systematically study software lineage inference by exploring four fundamental questions not addressed by prior work. First, how do we automatically infer software lineage from program binaries? Second, how do we measure the quality of lineage inference algorithms? Third, how useful are existing approaches to binary similarity analysis for inferring lineage in reality, and how about in an idealized setting? Fourth, what are the limitations that any software lineage inference algorithm must cope with? Towards these goals we build ILINE, a system for automatic software lineage inference of program binaries, and also IEVAL, a system for scientific assessment of lineage quality. We evaluated ILINE on two types of lineage-- straight line and directed acyclic graph--with large-scale real-world programs: 1,777 goodware spanning over a combined 110 years of development history and 114 malware with known lineage collected by the DARPA Cyber Genome program. We used IEVAL to study seven metrics to assess the diverse properties of lineage. Our results reveal that partial order mismatches and graph arc edit distance often yield the most meaningful comparisons in our experiments. Even without assuming any prior information about the data sets, ILINE proved to be effective in lineage inference--it achieves a mean accuracy of over 84% for goodware and over 72% for malware in our data sets. |
Year | Venue | Keywords |
|---|---|---|
2013 | USENIX Security | infer software lineage,software lineage inference algorithm,automatic software lineage inference,lineage inference algorithm,towards automatic software lineage,program binary,software lineage,software lineage inference,inferring lineage,lineage quality,lineage inference |
Field | DocType | Citations |
Edit distance,Data mining,Data set,Vulnerability (computing),Computer science,Computer security,Study software,Software,Artificial intelligence,Graph,Inference,Malware,Machine learning | Conference | 20 |
PageRank | References | Authors |
0.76 | 37 | 3 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Jiyong Jang | 1 | 297 | 16.23 |
| Maverick Woo | 2 | 173 | 7.47 |
| David Brumley | 3 | 2940 | 142.75 |