Theses in the field of Telematics/Computer Networks

The selection of open topics is usually not limited to the compilation presented here. Instead, various other current and interesting tasks are usually available on request.

If you are interested, please send a mail to michael[dot]rossberg[at]tu-ilmenau[dot]de or david[dot]schatz[at]tu-ilmenau[dot]de.


Topic group: VPN autoconfiguration

The secure transmission of sensitive information via the Internet currently requires the utmost care. In particular, the networking of different offices presents companies, but also public authorities, with the task of setting up and maintaining IPsec networks. The long-term goal is to eliminate this very costly and error-prone process and to introduce a system that is almost as easy to use as a toaster.

Within the scope of the project, a number of very different problems have to be solved: from A for address assignment to Z for condition monitoring. The relationship between theory and practice is just as variable. From theoretical safety verification to simulation studies on robustness to experimental implementation, everything is involved. The only thing in common is the relation to the secure operation of communication infrastructures.Ideal prerequisites are C++ experience, OMNeT++ knowledge and attendance of the lectures "Network Security" or "Protection of Communication Infrastructures".

Topic group: Anonymous, interactive real-time communication

Voice over IP (VoIP) applications are very popular for real-time interactive communication. While the confidentiality and integrity of voice data can be guaranteed by conventional cryptographic algorithms/protocols, the anonymity of users (such as communication relationships) is not sufficiently protected by popular VoIP applications. However, knowledge of such metadata may already constitute a serious invasion of user privacy.

Existing anonymization systems (e.g. Tor) redirect connections via several so-called relays, so that a direct identification of communication relations is no longer possible. An open problem in this context is to select the paths through the anonymization system in such a way that as many of the following properties as possible can be achieved:

  • An end-to-end latency of < 400ms for acceptable speech quality.
  • A utilization of the relays proportional to their available bandwidth
  • As little impact as possible of compromised/untrusted relays on user anonymity.

While several proposals for path selection already exist, they all have their advantages and disadvantages. Therefore, the main question is how to achieve the best possible compromise of the desired properties. A possible starting point for this would be to formulate the problem (or parts of it) as an optimization problem. An ideal prerequisite for this is a visit to the lecture "Network Algorithms". If necessary, a further simulation of the user behaviour is necessary to determine anonymity metrics. Ideal prerequisites for this are C++ experience as well as OMNeT++ knowledge (or similar simulation frameworks).