Cooperative Device Cloud: provisioning embedded devices in ubiquitous environments

Cloud Computing and the Internet of Things (IoT) are well known principles addressing core challenges of the distributed systems the Internet is founded on. Cloud Computing is already widely adopted in productive environments and has evolved from a “buzzword” to a commonly used technology. In contrast, IoT currently suffers from the lack of foundational design principles that allow overcoming the segmentation into solutions dedicated to particular IoT application domains, such as Automation, E-Health, Smart Homes, or Smart Cities. IoT applications are often tightly coupled to challenges like sensor integration, sensor management, semantics or, the heterogeneity of data and technologies in general. Besides the heterogeneity, the proliferation of IoT related solutions will lead to a continuously increasing amount of the sensors and devices, which are foundational to IoT applications. This leads to the challenge of efficiently managing and provisioning the resources provided by the devices. Looking at the Cloud Computing domain, popular related concepts like on demand provisioning, elasticity or, resource pooling are already available and well investigated. This work presents the Device Cloud concept, which aims at applying these Cloud Computing concepts to the IoT domain. Sensors and devices will be organized in resource pools and dynamically provisioned to users that can benefit from the resources offered. The Device Cloud will eliminate static bindings between sensors and users and allow accessing any kind of physical IoT devices on demand, similar to the Pay As You Go paradigm popular within the scope of Cloud Computing. The core question addressed by this thesis is how a dynamic, on-demand provisioning of the devices can be realized independently from the application domain or the technologies a device is based on. A generic, application independent architecture model will be introduced and discussed with respect to the participating actors, their interactions and, security and privacy. Based on an E-Health use case, the generic model will be applied to a specific application domain.