RAJA: a resource-adaptive Java agent infrastructure

Yun Ding, Rainer Malaka, Christian Kray, Michael Schillo

Abstract: This paper presents RAJA, a Resource-Adaptive Java Agent Infrastructure. RAJA is easily accessible to agent developers, since it allows structured program\-ming by using a multi-level architecture to clearly separate domain-specific functionality from resource and adaptation concerns. It is generic, since it is applicable to a wide range of adaptation strategies. These two key features are illustrated by several applications, where the RAJA concept has been successfully applied to solve real world problems. They stem from very different domains (video streaming and spatial reasoning), which demonstrates the wide range of application and the flexibility of the proposed infrastructure.