A service-oriented architecture for IoT infrastructure and Fog-minded DevOps

The huge and steady growth in terms of the number of distributed devices con- nected to the Internet, the so-called Internet of Things (IoT), calls for newly developed infrastructure management techniques to deal with the complexity of emerging IoT deployments, especially in light of the growing impact of the sharing economy. In this context, most of the management platforms tackle IoT issues from a high level where IoT data is managed using Cloud oriented solutions. In such a scenario, the approach adopted can be categorized under the data-centric approach where IoT devices are considered as simple mere data generators uploading data towards Cloud platforms that provide, afterwards, processed data to the users. In order to challenge this mainstream consensus on the relationship between the Cloud and IoT, what is interesting to investigate is the adoption of the Cloud "as- a-Service" approach from a low level when dealing with IoT infrastructure. Indeed, the as-a-Service paradigm provides well-investigated mechanisms for infrastructure and service provisioning; thus the challenge then is to adapt this approach to fit the management of a dynamic, possibly virtualized, infrastructure of sensing and actuation resources. Cloud providers can extend then their offerings portfolios by providing access to shareable IoT resources according to the utility model using access at the lowest level where possible. Besides providing access to virtualized IoT nodes, an interesting capability to enable is related to computing at the network edge (even on the IoT nodes themselves) to meet the requirements of typical IoT applications, such low processing delays and data privacy. The thesis presents the design and implementation of a set of mechanisms to integrate IoT within the Cloud wisdom. In particular, the approach enables the capa- bility of offering IoT resources (e.g., sensors and actuators) as virtualized resources. Therefore, the virtual IoT instances can take benefits of the resources (e.g., storage, networking and compute) offered by the Cloud, Fog or the edge-based IoT nodes. The premise then lies in engaging the research from a device-centric perspective using the Stack4Things framework.