Locality abstractions and security models in a mobile agent environment

Mobile agents have attracted wide interest in distributed applications because of their flexibility and capacity of adapting to very different scenarios, a common situation over the Internet. The rapid growth of this area has forced to focus more on rapid prototyping than on other aspects that are still only partially faced. The paper addresses two main concepts connected with mobile agents: locality abstractions and security models and, in addition, their mutual relationship. The paper describes the MA environment and presents some results of its implementation in the Java language. A distributed monitoring application gives an idea of system performance and cost.