On providing on-demand interoperability solutions for IoT

Abstract

As more and more Things join the Internet, the interoperability of Things becomes a major concern. Several challenges arise when Things should be able to interoperate with other software components. Since Things enable an enormous area of application, they are highly heterogeneous actors. Things are designed to fit the application context and as a result the data that Things offer varies greatly. Additionally, Things use diverse technologies since their requirements diverge. Software systems that utilise the IoT will have to integrate cloud services, network functions and Things. Such IoT cloud systems will be major enablers of IoT applications. But in order for an IoT cloud system to function, the components of the system have to interoperate with each other. However, the Thing diversity presents a challenge when the interoperability of IoT cloud systems is pursued. Appropriate interoperability solutions are needed to manage the technology and data diversity, such that Things can interoperate with IoT cloud systems. Nevertheless, the technology and data diversity itself is not the only interoperability challenge that needs to be resolved. Things are often mobile and depend strongly on their context and environment. As Things move through the world or change contexts, they will have to dynamically enter and leave IoT cloud systems. It is therefore not sufficient to manage the interoperability problems with static means. Interoperability solutions need to be provided on-demand. Detecting interoperability problems and finding appropriate solutions is still an exhaustive and costly task. Same is true for provisioning and deploying such solutions. However, all of those steps are required to provide on-demand interoperability solutions. The aim of this thesis is to reduce the time and effort that is required to perform those tasks. By modeling the interoperability capabilities of Things, network functions and cloud services with interoperability metadata, we endeavour to detect interoperability problems and recommend solutions. Furthermore, we reuse and deploy software components that are deemed capable of solving the problem. Additionally, we discuss and define the property of on-demand interoperability and several factors that are key to providing on-demand interoperability. We introduce Interoperability DevOps as the main stakeholders of our framework, since complex interoperability problems must be resolved with human intelligence. A proof-of-concept prototype that implements the framework was built during this thesis. The prototype was then used to evaluate the appropriateness of the framework.