A Note About the Semantics of Delegation

In many applications, mobile agents are used by a client to delegate a task. This task is usually performed by the agent on behalf of the client, by visiting various service provider's sites distributed over a network. This use of mobile agents raises many interesting security issues concerned with the trust relationships established through delegation mechanisms between client and agent, agent and service provider and client and service provider. In this paper we will explain why the traditional semantics of delegation used by existing access control mechanisms, either centralised or distributed, are generally not satisfactory to prevent and detect deception and why these problems are even more critical when these semantics are used in mobile agent paradigms.