An Analytical Investigation of Software Mutation for Increased Information Survivability
Abstract
The work reported here has three major components. First, we report on the development of an analytic framework for judging the success of diversification schemes at retarding malicious attacks. We attempt to quantify the relationship between the amount of work needed by a defender to achieve a certain level of diversity, and relate it to the amount of work needed by an attacker to subvert the resulting system. Second, we report on a prototype system for automating the creation of diverse programs. This is, in fact, a general-purpose mutation system wherein the modification of software is controlled by mutation scripts or meta-programs that drive the mutation of source code. Finally, we report on a further prototype system for boosting information-system survivability by means of diversity. This system monitors the behavior of programs and ensures their compliance with semantic constraints describing their intended behavior, so that undesired behavior can be detected. Monitoring all aspects of a program's semantics would be infeasible, but in a diverse system, different replicas of a program can monitor different, randomly selected aspects of software behavior.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 2004
- Accession Number
- ADA423201
Entities
People
- Alexandre Hulot
- Aron Bartie
- C. C. Michael
- John Viega
- Natasha Jarymowycz