Jonas Felipe Pereira de Queiroz
Date, Time and place:
June 14, 14:00, Sala de Atos of FEUP
Carlos Manuel Milheiro de Oliveira Pinto Soares, PhD, Associate Professor, Departamento de Engenharia Informática, Faculdade de Engenharia da Universidade do Porto.
Luís Filipe Santos Gomes, PhD, Associate Professor with Habilitation, Departamento de Engenharia Electrotécnica e de Computadores, Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa;
Maria Goreti Carvalho Marreiros, PhD, Coordinating Professor with Habilitation, Departamento de Engenharia Informática, Instituto Superior de Engenharia do Porto;
Paulo Jorge Pinto Leitão, PhD, Full Professor, Departamento de Engenharia Eléctrica, Instituto Politécnico de Bragança (Supervisor);
Rosaldo José Fernandes Rossetti, PhD, Associate Professor, Departamento de Engenharia Informática, Faculdade de Engenharia da Universidade do Porto.
The 4th industrial revolution is characterized by the digitization of industrial environments, based on the use of Internet of Things, Cloud/Edge Computing and Artificial Intelligence technologies. This technological advances are pushed by the ever-changing market that besides posing new business challenges, further increase the already existing complexity and related issues faced by industries at the operational environments. In this context, Cyber-Physical Systems (CPS) have been considered the new paradigm to develop the next generation of intelligent and distributed industrial automation systems that can achieve higher levels of flexibility and dynamic adaptation. Although data analysis has shown to be a key enabler of industrial CPS in the development of smart machines and products, the traditional Cloud-centric solutions are not suitable to attend the data and time-sensitive requirements. Aiming to cope with that, Edge Computing has been adopted to enable the data processing capabilities at or close to the physical components, complementing the Cloud solutions. However, defining what data analysis capabilities should be deployed along Cloud-Edge computational layers is not a straightforward task. Motivated by this challenge, this work proposes a conceptual framework that defines a modular agent-based architecture to design and develop cyber-physical components, together with a general guideline to support the system engineers to evaluate and determine the most suitable computing layer to deploy a given data analysis task. The proposed framework was implemented and evaluated considering some experiments based on a smart machine CPS testbed, and different scenarios that illustrate the benefits of distributing the data analysis capabilities along Cloud-Edge layers. The modular structure of the agent-based cyber-physical component architecture shown to be suitable, enabling to easily configure and distribute data analysis tasks along the CPS components and their deployment in the different computing layers, also providing a seamless and transparent interaction between the components. In addition, the proposed guideline and the Fuzzy Logic decision-making systems were used to recommend, in a less ad-hoc manner, the most suitable layer to deploy three data analysis tasks identified in the case study.
Keywords: Cyber-Physical Systems, Data Analysis, Multi-Agent Systems, Edge Computing