Ameni Kallel PhD thesis defense

Title: Une approche de déploiement des processus métier à l’ère IoT dans des environnements hybrides de fédération (Fog-Cloud computing).
Supervisors: Dr. Mahdi Khemakhem & Dr. Molka Rekik
Defense Date: 01 March 2024
The evolution of Fog Computing complements Cloud Computing by extending its boundaries to the edge of the network, creating an ideal ecosystem for integrating the Internet of Things (IoT). Our proposal aims to bridge the gap between modeling and architecture for IoT applications in Fog-Cloud hybrid environments (HFC). First, we introduce a BPMN 2.0 extension that handles the complexities of IoT and non-IoT resources, taking into account capabilities and quality of service (QoS) constraints. Furthermore, our IoT-Fog-Cloud architecture provides a solution to the challenges of IoT architectures. We demonstrate it through two prototypes - a child autism monitoring system and a COVID-19 monitoring system - with the goal of improving healthcare and social services. Rigorous experiments validate the efficiency of our prototypes. Amidst the COVID-19 pandemic, we focus on combating transmission by integrating Machine Learning (ML) technology with Fog, Cloud, and IoT for a COVID-19 surveillance and prediction system, offering advanced solutions. Our system uses an HFC Framework to leverage continuous data, enabling precise disease detection. Our federated ML approach includes two types of ML for real-time (stream-ML) and long-term (batch-ML) predictions.
Furthermore, we address the challenges of IoT applications, highlighting their needs for reliable network connections and agile data management. Our proposal for containerbased microservices scheduler in an HFC environment, named DRL4HFC, tackles this complexity by using Deep Reinforcement Learning (DRL) for optimal microservices placement, improving resource efficiency and QoS.