Trustworthy Infrastructure, Mechanisms, and Experimentation for Diffuse Computing


Diffuse computing is concerned with managing and maintaining a computational infrastucture that is distributed among many heterogeneous nodes that do not trust each other completely and may have differing incentives, needs, and priorities. As commercial, academic, civilian, government, and military systems become increasingly diffuse, the challenges of providing reliable and trustworthy diffuse systems become increasingly important. We propose to augment the research agenda of the ONR CIP/SW SPYCE team with a specific but multifaceted problem: effective, timely, and confidential sharing of security-related information. Diffuse systems must respond to security events in order to provide robust and reliable service, but coordinating a response across a large distributed system raises particular challenges, especially when the parties involved do not trust each other completely. The lack of trust may be inherent, or arise simply as a prudent means of containing subsystem failure or compromise. Our approach builds on recent work (within and beyond the ONR CIP/SW SPYCE group) exploring ways to combine multi-party computation protocols and incentive-compatible mechanisms and to develop language-enforced security methodology, including policy and programming language aspects. The proposed effort also builds on recent work in computer-virus and maliciouscode threat detection, monitoring, analysis, and mitigation.