[Back]


Diploma and Master Theses (authored and supervised):

P. Eder:
"An infrastructure agnostic application deployment framework for the Internet of Things";
Supervisor: S. Dustdar, M. Vögler; Institute of Information Systems, Distributed Systems Group, 2017; final examination: 2017-05-30.



English abstract:
In the last years terms like Smart Grid, Smart City or Smart Vehicle, became increasingly popular. They all fall under the conception of the so called Internet of Things (IoT). It is estimated that in 2020 will be up to 34 billion devices connected with the Internet.
However, the huge potentials of IoT come with various challenges. Among issues like privacy, security, scalability, etc., the infrastructure in IoT is a significant problem. The huge amount of different devices leads to a highly heterogeneous environment. Thus,
developing applications that respect these heterogeneous infrastructures is exceedingly complex and in further consequence very error prone. A possible solution to this problem is to abstract the hardware layer through virtualization.
In the course of this thesis we investigate how operating-system-level virtualization can be used to cope with the heterogeneous environment. Therefore, commonly used IoT devices
are presented and an in-depth explanation of how operating-system-level virtualization is implemented on different operating systems (e.g., Linux, FreeBSD, Solaris, Windows) is given. Furthermore, currently available container engines with focus on Linux are introduced and compared. We also explain the concept of continuous delivery, and why this development approach facilitates agile development.
To show the feasibility of operating-system-level virtualization in the context of IoT, a prototype of an application deployment framework is developed. This framework provides distribution and deployment of applications within one click. Furthermore, it allows
to deploy on different CPU-architectures (e.g., ARM, x86) transparently to the user. Additionally, it can easily be integrated within a continuous delivery pipeline through a REST API. The features of the framework are demonstrated via the deployment of a
Building Managament System (BMS). In the last part of the thesis, the deployment time of different applications on various devices is evaluated and discussed, which shows both, feasibility and applicability of the proposed approach.

Created from the Publication Database of the Vienna University of Technology.