Trust Mechanisms and Reputation Systems

Abstract

Most of the approaches for ad hoc networks assume that the nodes readily and honestly follow the protocol. Unfortunately, this assumption does not always hold in reality. Misbehaving in ad hoc networks can take place at different levels of the system architecture:

• 1 On the communication level, nodes can refuse to forward messages of others in order to save bandwidth and energy. Many misbehaving nodes would lead to a low reliability of the system. But such nodes can be detected, because every node can eavesdrop the network traffic of adjacent nodes ([279], ’Watchdog’). Most of the research on reputation and trust models in ad hoc networks focus on that issue.

• 1 In contrast, misbehavior on the application level covers spoof query results and valueless services from nodes that would be able to provide high-quality ones. It is challenging to automatically detect this kind of adverse behavior. Typically it is the end-user who recognizes that a service provided by a certain node does not meet its expectations. But it is very important for open ad hoc networks to detect and expel such nodes. Otherwise, the system can become suspicious, and the users can loose their motivation to participate in the system [317].