Only exemplary topics are listed here.
If you are interested in these or related topics, please explain your expertise and contact email@example.com
We offer various is topics for Bachelor theses, Master theses or project works (Forschungsprojekte) in the area of design automation, embedded systems, and formal methods. Typically, we discuss in person before fixing a specific topic.
Test Optimization for Embedded Systems
Embedded systems are not directly visible to the end user. However, they must be extremely reliable to guarantee correct behavior of larger technical systems. For this reason elaborated test procedures are applied that successively validate each embedded system before shipping. Individual hardware components, electrical interfaces and, finally, the functionality of the complete embedded system including software are incrementally tested. The goal of this work is to work on testing data from a actually running testing process at an industrial partner's facilities in Hamburg.
Technology-Level Test Vector Generation
Test vectors for integrated circuits applied to each produced chip guarantee proper functionality and highest quality standards. More recent improvement in circuit technology and the use of advance approximate processing units makes a precise differentiation of good and bad chips on the logic level more and more difficult. The topic proposed here will consider low-level technology information for generating tests for advanced production processes and approximate hardware.
Verifying Artificial Neural Networks
Formally verifying the correctness of procedures and systems used in safety related areas is a must. Artificial neural networks have been proven very effective in solving many tasks in every day life. However, methods deciding whether an artificial neural network never violates safety guarantees are only at their infancy. The goal of this work is to study the state-of-the-art in verifying correctness of artificial neural networks.
Assessing the Performance of Embedded Machine Learning
Most suppliers of advanced embedded processing devices provide software libraries and hardware support to improve the performance of machine learning in embedded systems either for training classifiers, for applying learned classifiers or both. But how efficient are these embedded platforms. The goal is to use and evaluate the performance of embedded machine learning platforms.