Agent-based cyber-physical system development with SEA_ML


Challenger M., Tezel B. T., Amaral V., Goulao M., Kardaş G.

MULTI-PARADIGM MODELLING APPROACHES FOR CYBER-PHYSICAL SYSTEMS, TEKINERDOGAN BEDIR,BLOUIN DOMINIQUE,VANGHELUWE HANS,GOULÃO MIGUEL,CARREIRA PAULO,AMARAL VASCO, Editör, Academic Press , Oxford, ss.195-219, 2021

  • Yayın Türü: Kitapta Bölüm / Araştırma Kitabı
  • Basım Tarihi: 2021
  • Yayınevi: Academic Press
  • Basıldığı Şehir: Oxford
  • Sayfa Sayıları: ss.195-219
  • Editörler: TEKINERDOGAN BEDIR,BLOUIN DOMINIQUE,VANGHELUWE HANS,GOULÃO MIGUEL,CARREIRA PAULO,AMARAL VASCO, Editör
  • Dokuz Eylül Üniversitesi Adresli: Evet

Özet

Intelligent agents are software components that can work autonomously and proactively to solve the problems collaboratively. To this end, they can behave in a cooperative manner and collaborate with other agents constituting systems called Multi-agent Systems (MAS). These systems have different perspectives such as the internal structure, plan, interaction, organisation, role, environment and so on. By having these views, MASs can consider the structure, behaviour, interaction, and environment of the complex systems such as Cyber-physical Systems (CPS). Therefore, intelligent software agents and MASs can be used in the modelling and development of CPSs.

There are different Domain-specific Modelling Languages (DSMLs) to build MASs with a focus on various MAS aspects. One of the generative MAS DSMLs is SEA_ML++ which presents a thorough Model-driven Engineering practice with including the abstract syntax, graphical concrete syntax, model-to-model transformations and model-to-code transformations with the support of Platform Independent and Platform Specific levels of MAS modelling. In this chapter, we discuss how SEA_ML++ is used for the design and implementation of agent-based CPSs. An MDE methodology is introduced in which SEA_ML++ can be used to design agent-based CPS and implement these systems on various agent execution platforms. As the evaluating case study, the development of a multi-agent garbage collection CPS is taken into consideration. The conducted study demonstrates how this CPS can be designed according to the various viewpoints of SEA_ML++ and then implemented and executed on Jason platform.