My research aims to explore Cloud/Fog and Edge Computing for efficient processing and storage of IoT data. By incorporating Game Theories and learning algorithms, I seek to predict the equilibrium of data provisioning, ensuring improved data quality for end consumers and enhancing their overall user experience. Additionally, I address the latency issue in IoT data by implementing Markov Chain Models and Echo State Networks to optimize data processing in big data analytics. Ultimately, my research seeks to harness the potential of IoT-Cloud integration and advanced data processing techniques to manage effectively and utilize IoT data in pervasive environments.

The latency of IoT data modeling in Mobile Cloud
The vision of ubiquitous computing in interactive mobile cloud applications and Internet of Things (IoT) based systems is still difficult to achieve. The difficulty lies in the use of cloud services in mobile devices, which impacts the issues of performance, scalability, availability, and lack of resources in mobile computing environments. Despite the astonishing advancement achieved in IoT technology, there is still much to do. Some IoT-based systems, which rely on a variety of mobile devices, need to work even when the connection is temporarily unavailable or under-degraded. Besides, mobile cloud service providers can reduce network latency by moving some of their services close to the user. We propose in this chapter the usage of small clouds known as cloudlets, and we describe two cloudlet based architectures, which allow leveraging the geographical proximity of cloud services to mobile users. We model the network latency of the different components of the two architectures using a continuous-time Markov chain (CTMC).

Adaptive CLA’s Negotiation for Pervasive Systems
To cope with the energy performance concern of pervasive and Internet-of-Thing (IoT) devices, current pervasive systems require intelligent algorithms that can change the behavior of the devices and the overall network. Making the devices aware of their states and able to adjust their operative modes using context information has the potential to achieve better energy performance. Context information is typically obtained from environmental sensors, device sensors, and from external sources. Also, the massive amount of data generated by sensors and smart objects led to a new kind of services known as Data-as-a-Service (DaaS). The quality of these services is highly dependent on the quality of sensed data (QoD), which is characterized by a number of quality attributes. DaaS provisioning is typically governed by a Service Level Agreements (SLAs) between data consumers and DaaS providers. We propose a multi-attributes and adaptive approach for Context Level Agreements (CLAs) negotiation by employing the Game Theories and implementing learning algorithm to show the convergence of the Nash Bargaining Solution (NBS). And also we propose a game-based approach for DaaS Provisioning, which relies on signaling based model for the negotiation of several QoD attributes. We use in the negotiation between the two parties a Q-learning algorithm for the signaling model and a Multi Attributes Decision Making (MADM) model to select the best signal.

IoT Data Provisioning
IoT data provisioning refers to the process of collecting, transmitting, and managing data generated by Internet of Things (IoT) devices in a systematic and efficient manner. This encompasses the entire lifecycle of data, from its creation at the edge devices to its processing, storage, and delivery to end-users or applications. IoT data provisioning presents a myriad of challenges, including scalability to handle the exponential growth of data generated by a vast number of devices, data heterogeneity from diverse sources, ensuring data security and privacy during transmission and storage, real-time processing to meet application demands, and so on. Given the large amount of sensed data by IoT devices and various wireless sensor networks. Traditional data services lack the necessary resources to store and process that data, as well as to disseminate high quality of data to variety of potential consumers. We propose a framework for Data as a Service (DaaS) provisioning, which relies on deploying DaaS services on the cloud and using DaaS agency to mediate between Data Consumers and DaaS services using publish/subscribe model.

Big Data Analytics for Patent Analysis and Business Matching
Business-Matching Systems, Data Lake Management based on the DLDS Approach, and Feature Learning of Patent Networks using Tensor Decomposition - is the application of advanced data management and analysis techniques to optimize decision-making, uncover valuable insights, and improve efficiency in different domains. Business-Matching Systems utilize data analysis and matching algorithms to connect businesses with suitable partners, while Data Lake Management based on DLDS Approach focuses on organizing and curating raw data in data lakes to make it more accessible for analytics. The research on Feature Learning of Patent Networks using Tensor Decomposition aims to extract meaningful features from patent networks, enabling valuable insights for innovation and intellectual property strategies. Overall, these topics demonstrate the significance of data-driven approaches in enhancing processes and understanding within their respective domains.

Harnessing Machine Learning and Big Data Analytics for Real-World Apps
Business-Matching Systems, Data Lake Management based on the DLDS Approach, and Feature Learning of Patent Networks using Tensor Decomposition - is the applicaHarnessing Machine Learning and Big Data Analytics for Real-World Applications involves utilizing advanced data analysis techniques to gain valuable insights and make informed decisions in various domains. Specifically, training Echo-State Networks (ESNs) for Big Data Prediction enables the accurate forecasting of future trends and patterns based on large-scale historical data. ESNs, as recurrent neural networks, efficiently handle high-dimensional and complex datasets, making them suitable for applications like stock market prediction, weather forecasting, energy demand prediction, healthcare diagnostics, and IoT applications. Through careful data preprocessing, model tuning, and evaluation, ESNs can unlock the potential of Big Data, providing valuable information for real-world decision-making and improving the efficiency and accuracy of predictions across different industries.

Data LakeHouse for IoT data processing
The Data LakeHouse for IoT data processing is a unified and scalable architecture that integrates the strengths of Data Lakes and Data Warehouses to handle the vast volume and complexity of IoT data. However, its successful implementation faces challenges such as data integration and ingestion from diverse IoT sources, ensuring Quality of Data (QoD) and Governance, scalability and performance for real-time processing, robust security and privacy measures, optimizing data processing, and enabling low-latency and real-time analytics, ensuring interoperability with various data sources and platforms. Overcoming these challenges requires a comprehensive approach, combining efficient data management, advanced analytics, and robust infrastructure to derive valuable insights and drive innovation in IoT-driven industries.