Communication

Please encrypt emails addressed to me.

S/MIME certificate: download

PGP key: download

PGP fingerprint: FC02 6F24 803C F944 5CE1  32C9 E246 00E0 BC0F 483D.

Contact

Please come by at my office, or contact me via email.

Theses

The site lists open and finished topics for theses. More topics for Bachelor/Master/Diploma theses are available on the thesis website of the institute.

Upon consultation, I am also accepting own topics that relate to my fields of interests.

Open Topics for Theses


1.
default
Matousek, Matthias
Neural Network Prototyping
Bachelor Thesis, Project
Institute of Distributed Systems, Ulm University,
2017
in preparation

Keywords: AA,

Tags: AA, BA, PROJEKT, MatthiasMatousek

Abstract: Machine learning with Neural Networks — especially Deep Learning — is currently booming. In order to test the potential of such algorithms for specific use cases, it is necessary to be able to quickly prototype and evaluate neural networks. In this thesis or project, the student shall compare different neural network frameworks, such as TensorFlow or Caffe, and implement an exemplary use case. Such a use case could be an anomaly detection mechanism for connected cars.

Ongoing Thesis


Matousek, Matthias; Lang, Dominik
OBD Data-Logger for Driving Behaviour Analysis
Project
Institute of Distributed Systems, Ulm University,
2017
to be published

Tags: MatthiasMatousek, DominikLang, PROJEKT, AA

Abstract: Analysis of driving behaviour presents opportunities for security, but comes with implications for privacy. In order to conduct research in this field, driving data is required. The goal of this project is to implement tools that log driving data from a vehicle's OBD port. Further, a framework for data analysis — e.g. via machine learning mechanisms — should be provided.

Matousek, Matthias
Comparison of Homomorphic Encryption Scheme Implementations
Master Thesis, Project
Institute of Distributed Systems, Ulm University,
2017
to be published

Tags: AA, MA, PROJEKT, MatthiasMatousek

Abstract: Homomorphic Encryption (HE) -- the "holy grail" of cryptography -- allows computation on encrypted data without knowledge of the encryption keys. While it has been shown that both partially (only limited operations) as well as fully homomorphic crypto-systems (unlimited, arbitrary operations) are possible, they still come with substantial computation complexity and storage requirements. Goal of this work is to survey and compare HE scheme implementations. Properties of interest are their capabilities (partially vs. fully homomorphic), performance (speed and storage), as well as their favourable applications. The applicant should be familiar with IT security concepts (lecture "Security of IT-Systems") and cryptography (lecture "Kryptologie - Methoden und Algorithmen" and optimally "Privacy Engineering and Privacy Enhancing Technologies - PET"). They should further be interested in the application of different programming languages (such as C++, Python, Haskell, Java, and others). While no expert knowledge of programming languages is required, HE schemes are implemented in different languages, and thus, the willingness to explore these should be present. This topic is suitable for a Master thesis, or as a Master project.

Finished Theses


2016

Waldenmaier, Tobias
Privacy-Preserving Geofencing
Master's Thesis
Institute of Distributed Systems, Ulm University,
2016

Tags: AA, MA, MatthiasMatousek, privacy

Abstract: With geofencing, car holders can get notified when their vehicle leaves a predetermined area. This feature can be used by companies to keep tabs on their company vehicles, or by private individuals to gain higher security. However, when the vehicle manufacturer or a third-party provider is hosting the geofencing service, he also needs access to the vehicle location. This also means that the provider can derive movement profiles of the vehicle's drivers. So-called searchable encryption has the potential to enable geofencing without the need for the provider to know the exact vehicle locations. The goal of this thesis is to develop and evaluate a proof-of-concept that provides geofencing while preserving the drivers' privacy. Thus, several searchable encryption schemes should be compared and assessed for their applicability to this task. At least one of the schemes should be implemented and the resulting prototype evaluated regarding its adequacy and efficiency.

Kneer, Daniel
Simulation of Backend-Based Misbehaviour Detection for Connected Cars
Master's Thesis
Institute of Distributed Systems, Ulm University,
2016

Tags: AA, MA, MatthiasMatousek, itsec

2015

Gaiser, Steffen
Analysis of the Data Security and Privacy in Connected Cars
Master's Thesis
Institut für Verteilte Systeme, Universität Ulm,
2015

Tags: AA, MA, MatthiasMatousek, privacy, itsec

Abstract: Vehicular networks are becoming reality in todays cars. Modern cars are connected to the Internet using a WLAN or GSM connection. The cars are communicating with the backend of the vendor and also with other service providers on the Internet. The data contains a lot of private information about the driver. From this data, a data profile can be achieved, for example from location data, contacts and calender entries. This work analyses how the data protection and privacy is implemented by the vendors of different cars and shall give an overview of the different connected car systems. Therefore an overview of the different services of the cars will be created which categorizes the privacy terms. Another point is the analysis and categorization of the privacy terms which have to be accepted to use the services when buying or renting a car. The next step is – if possible – to analyse the technical implementation. Therefore a sniffing of the traffic of the different services will be realised. The goal of the thesis is to give an overview of the data protection and privacy of the services in current connected cars.