Farid Feyzi

Thesis Title: A Model-Driven Approach for Developing Self-Adaptive Multi-Agent Systems                        


A growing number of distributed information systems need to operate in open, dynamic and heterogeneous environments, and are required to adapt themselves to environmental changes in order to satisfy specific design objectives. These self-adaptive systems are inherently complex, mainly due to the complexity of their environments and the effects of dynamic adjustments made to their structure and behavior. Developing such complex systems is still an important challenge in software engineering.
Agent technology is considered as an appropriate and powerful paradigm for developing complex and dynamic distributed systems. Intelligent/autonomous agents have been proved useful for enhancing flexibility and adaptability in many different domains. However, agent technology has not been used to its utmost potential for developing complex self-adaptive systems; the main problem is the lack of effective and easy-to-use agent oriented programming languages and supported platforms.
Model-Driven Development (MDD) promotes the production of models with sufficient detail so that they can be used for generating executable software. In MDD, models are pivotal elements of the developments process, rather than its byproducts. Changes to models can be applied to code automatically, without redevelopment needed. It has been observed that models can be utilized for capturing the knowledge used by agents for providing flexible behavior in a platform-independent context, and can thus be reused continuously. MDD is therefore a promising approach for developing self-adaptive multi-agent systems.
The purpose of this research is to propose a new model-driven approach for developing self-adaptive multi-agent systems. To this aim, we will first study agent-oriented software development concepts and methods. MDD and agent-oriented MDD concepts, algorithms and tools for model transformation will then be explored. New developments in MDD and agent-oriented technology will ultimately be used for proposing a new approach that addresses the problems currently encountered in developing self-adaptive systems. The proposed approach will be evaluated through a case study.    

(Progress Chart)

Contact Information

   Email: ffeyzi [at] ce [dot] sharif [dot] edu