[Back]

M. Filz:
"A framework for dynamic configuration of IoT nodes based on events";
Supervisor: S. Dustdar, M. Vögler; Institute of Information Systems, Distributed Systems Group, 2017; final examination: 2017-05-29.

The Internet of Things (IoT) represents an evolving field of technical, social, and economic significance. At its core, it connects a plethora of devices able to collect and sense their environment by using embedded sensors and exchange gathered information via the Internet. The enormous amount of data generated by these devices produces both new possibilities and new challenges. Opening up an endless array of new opportunities for innovative applications, the high volume of data demands the development of new tools and technologies. One such technology is Complex Event Processing (CEP), designed to detect patterns by analyzing streams of data in real-time. Applied to the IoT, it allows applications to make decisions tailored to their specific environment. As a result, applications can change their configuration depending on different requirements such as location, weather, and many other factors.
However, CEP in the traditional sense of a central processing unit does hardly scale across thousands of nodes. This limitation can be overcome by applying a distributed processing model, allowing to add components as required. Yet while distribution allows such applications to meet their requirements, it comes at a cost: the effort necessary for implementation of distributed systems are very high due to their increased complexity. Therefore, this thesis proposes a framework for supporting and facilitating the construction of CEP based IoT applications. By design, the framework leverages a modular design of the concrete application and employs well-defined communication structures to satisfy aforementioned requirements. It also cares about low-level tasks of distributed systems such as the coordination of different components, the distribution of rules, and measures for improving fault tolerance. Moreover, this thesisproposes a custom language for defining queries and rules. The proposed language is optimized for the IoT context and enables users to express patterns in an intuitive and succinct way.
Additionally, this thesis presents a prototype implementation of the framework. This prototype implementation is consequently used to carry out evaluations. Besides functional use cases in the field of smart buildings, the evaluation focuses on quantitative measurements. By discussing and analyzing multiple test cases, the thesis discusses the feasibility and applicability of the proposed framework.