Managing trust in distributed environments

Summary form only given. In pervasive computing the involved agents and services must organize themselves dynamically both in cooperative and non-cooperative environments in order to enable sharing of distributed resources and knowledge discovery. Provisioning of trusted services appears to be an important factor to enable the success of this new computing paradigm. Reputation-based trust management models are becoming a focus of the research community as a viable solution. On the one hand, the trust models must satisfy constraints imposed by the environment, such as scalability, on the other hand, they must comply with the unique properties of trust as a social and economic phenomenon. In this paper we classify and compare approaches that have been proposed to fulfil this task. The identified classes are: social networks, probabilistic estimation techniques and game-theoretic reputation systems. Social networks and probabilistic estimation techniques operate on so called probabilistic behavior, while game-theoretic models are suitable for rational behavior. Probabilistic techniques normally imply smaller implementation overheads than social networks and enable a more intuitive transition of reputation to trust, while social networks are believed to be able to better detect misbehavior in a broader range of settings. Game-theoretic models, as appropriate for rational economic agents maximizing their utilities, can under specific circumstances enforce trustworthy behavior.