Sound Over-and Under-Approximations of Complexity and Information Security (SOUCIS)
Abstract
The technical keystones of this initiative were the use of sound over-approximating static analysis in conjunction with precise under-approximating analysis. For the former, new static analysis techniques for inferring program invariants in conjunction with a new technique for revealing side channels and complexity attacks in Java programs were developed. For the latter, new randomized, fuzz testing and machine learning techniques for vulnerability identification were developed. The state of the art in both areas was systematically surveyed and results were found that challenged previously published conclusions. A collaborative workbench application was developed to organize an analyst's task in using the tools.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 2018
- Accession Number
- AD1061405
Entities
People
- David Van Horn
- Dawn Song
- Eric Koskinen
- Jeff Foster
- Michael Hicks
- Timos Antopoulos
Organizations
- University of Maryland