Model-driven development of self-adaptive multi-agent systems with context-awareness

Abstract

Self-adaptive systems are complex and operate in dynamic and heterogeneous environments. They have to dynamically modify their behaviour at run-time in response to different kinds of changes. This paper presents a methodology to develop context-aware self-adaptive systems by employing the model driven architecture (MDA) and agent-oriented technology advantages. The approach aims to combine these two promising technologies to overcome the complexity of developing such systems. The methodology focuses on the key issues in the analysis and design of self-adaptive multi-agent systems. Different abstraction levels based on MDA has been proposed and mappings between models in these levels provided. These mappings bridge the gap between the high-level models produced in computation independent (CIM) and platform independent models (PIM) as well as the low-level models based on specific implementation platform called SADE (Self-adaptation Development Environment). The proposed approach has been evaluated through a case study described in the paper.