Human activities are increasingly based on the use of distant resources and services, and on the interaction between remotely located parties that may know little about each other. Mobile agents are the most suited technology. They must therefore be prepared to execute on different hosts with various environmental security conditions. This paper introduces a trust-based mechanism to improve the security of mobile agents against malicious hosts and to allow their execution in various environments. It is based on the dynamic interaction between the agent and the host. Information collected during the interaction enables generation of an environment key. This key allows then to deduce the host's trust degree and permits the mobile agent to adapt its execution accordingly to the host trustworthiness, its behavior history and the provided Quality of Service (QoS). An adaptive mobile agent architecture is therefore proposed. It endows the mobile agent with the ability to react with an unexpected behavior.

Mobile agents must be prepared to execute on different hosts where, according to specific environmental conditions, they will adopt different behaviors.The aim of this paper is to describe how a mobile agent equipped with a flexible capacity, can adapt its execution with the request and security found (in a visited environment). Based on its security, The agent becomes able to adapt itself to the environment where it is currently running, by composing only appropriate modules. This allows enhancement of the privacy of the agent code.

No abstract

Security is a very active field, were too much terminology is vaguely defined. This leads to difficulties for applications to evaluate their security level in order to communicate in safety manner. This problem increases when these applications use mobile agents. Indeed, mobile agents have to estimate the trust of environment where they will be executed. This issue is addressed, in this paper, by the use of security ontology. The development of this ontology must follow a process which consists on a set of phases in order to leads to a typical ontology.

The use of distributed systems and IT is growing, with automation being used more and more to facilitate our daily tasks. The need to remotely monitor a patient has driven one of important results of this growth: domestic medical systems. The latter are able to follow and maintain the condition of a patient in the patient's home. Monitoring is important in terms of saving time and also money. However, the critical nature of this task requires a high level of dependability. The aim of dependability is to satisfy the user's goal, which is that whatever the state and context of the overall system, its ability to control the operation of the medical device and to transmit files reporting the patient's condition (normal, critical, alert, etc.) must be continuously assured. This can be ensured by fault tolerance techniques. The authors' objective in this paper is to present a technique for fault tolerance in a domestic medical system. Briefly, their proposal integrates a smart concept into the system: agents for controlling the operation of the medical system and tolerating the faults that can occur.