Towards Data Sharing across Decentralized and Federated IoT Data Analytics Platforms

In the past decade the Internet-of-Things concept has overwhelmingly entered all of the fields where data are produced and processed, thus, resulting in a plethora of IoT platforms, typically cloud-based, that centralize data and services management. In this scenario, the development of IoT services in domains such as smart cities, smart industry, e-health, automotive, are possible only for the owner of the IoT deployments or for ad-hoc business one-to-one collaboration agreements. The realization of "smarter" IoT services or even services that are not viable today envisions a complete data sharing with the usage of multiple data sources from multiple parties and the interconnection with other IoT services. In this context, this work studies several aspects of data sharing focusing on Internet-of-Things. We work towards the hyperconnection of IoT services to analyze data that goes beyond the boundaries of a single IoT system. This thesis presents a data analytics platform that: i) treats data analytics processes as services and decouples their management from the data analytics development; ii) decentralizes the data management and the execution of data analytics services between fog, edge and cloud; iii) federates peers of data analytics platforms managed by multiple parties allowing the design to scale into federation of federations; iv) encompasses intelligent handling of security and data usage control across the federation of decentralized platforms instances to reduce data and service management complexity. The proposed solution is experimentally evaluated in terms of performances and validated against use cases. Further, this work adopts and extends available standards and open sources, after an analysis of their capabilities, fostering an easier acceptance of the proposed framework. We also report efforts to initiate an IoT services ecosystem among 27 cities in Europe and Korea based on a novel methodology. We believe that this thesis open a viable path towards a hyperconnection of IoT data and services, minimizing the human effort to manage it, but leaving the full control of the data and service management to the users' will