Exercising Qualitative Control in Autonomous Adaptive Survivable Systems

We seek to construct autonomous adaptive survivable systems that use active trust management to adapt their own behavior in the face of compromises in the computational environment. Active trust management maintains probabilistic trust models that indicate the trustworthiness of different resources for different tasks, and uses these models in rationally adapting allocations of computational resources to tasks. Flexible adaptation of allocations to changing circumstances places great demands on the methods used to represent the utility information needed by rational decision-making mechanisms. This paper explains how to use qualitative preference specifications to exercise effective control over quantitative trust-based resource allocation by facilitating convenient specification and adaptation of the stable foundations of the trust manager’s utility judgments.

• Type: Conference
• Year: 2001
• Where: IWSAS
• Authors: Jon Doyle, Michael McGeachie
• Keywords: Active Trust Management, IWSAS 2001, Probabilistic Trust Models, Software Engineering, Trust Manager Utility