Artificially Diverse Operating System (ADOS)

Artificial diversity attempts to eliminate the security risks of the IT monoculture. The goal of artificial diversity is to minimize the number of common vulnerabilities across a set of systems by intentionally introducing variations to the structure and process of functionally-equivalent software. Ideally, these variations are invisible to the legitimate user but pose a challenge to the malicious user attempting to find and exploit vulnerabilities. To provide security in computing systems through artificial diversity while maximizing performance, we propose an Artificially Diverse Operating System (ADOS). In this operating system-based approach to diversity, we will apply different diversity techniques within the operating system (OS), so that a network contains a diverse set of systems and individual systems exhibit diversity over time. Because these diversity techniques will be applied within the core OS (i.e., kernel) and are not application-specific, they support any user applications and complement other artificial diversity techniques applied at the application level. To selectively and dynamically apply the diversity techniques, we will design a lightweight reasoning module that optimizes the tradeoff between performance and security.