Abstract: Human activities are increasingly based on the use of remote resources and services, and on the interaction between
remotely located parties that may know little about each other. Mobile agents must be prepared to execute on different hosts with
various environmental security conditions. The aim of this paper is to
propose a trust based mechanism to improve the security of mobile
agents and allow their execution in various environments. Thus, an
adaptive trust mechanism is proposed. It is based on the dynamic interaction between the agent and the environment. Information
collected during the interaction enables generation of an environment
key. This key informs on the host-s trust degree and permits the mobile agent to adapt its execution. Trust estimation is based on
concrete parameters values. Thus, in case of distrust, the source of problem can be located and a mobile agent appropriate behavior can
be selected.
Abstract: The paradigm of mobile agent provides a promising technology for the development of distributed and open applications. However, one of the main obstacles to widespread adoption of the mobile agent paradigm seems to be security. This paper treats the security of the mobile agent against malicious host attacks. It describes generic mobile agent protection architecture. The proposed approach is based on the dynamic adaptability and adopts the reflexivity as a model of conception and implantation. In order to protect it against behaviour analysis attempts, the suggested approach supplies the mobile agent with a flexibility faculty allowing it to present an unexpected behaviour. Furthermore, some classical protective mechanisms are used to reinforce the level of security.