doc. RNDr. Ing. Petr Zemánek, CSc.

The kernels of monolithic operating systems are currently probably the most complex centralized programs that undergo a process of continuous development and tuning of both functionality and performance. They have an architecture of hundreds of modules. A number of tools are currently available to diagnose the functionality and performance of these kernels, but the level of usability of the outputs of these tools for effective kernel debugging is highly dependent on the human factor, the maturity and experience of the development team.

The aim of the dissertation is to research methods of automatic detection of the causes of performance problems of the operating system kernel and kernel scalability for multiprocessor computer systems. Investigate which known techniques of artificial intelligence and machine learning can be used to automate these methods.

doc. RNDr. Ing. Petr Zemánek, CSc.

Ing. Alexandru Moucha, Ph.D.

Department of Theoretical Computer Science

The objective of the master's thesis is a study on the integration of time-sensitive networking into the PikeOS real-time operating system. The analytical part of the thesis introduces real-time operating systems and the PikeOS. It then presents core concepts of time-sensitive networks, emphasizing time synchronization using the Precision Time Protocol. The implementation chapters explain how PikeOS drivers are structured, focusing on Ethernet drivers. Their core outcome is a driver for the ENET_QOS Ethernet controller. In addition to the essential frame transfer functionalities, the driver's features include advanced filtering of IEEE 802.1Q tagged frames, which presented a new standard API proposal, multicast handling and the Precision Time Protocol integration. Testing the driver verified its correctness and measured the error of the integrated time synchronization in the expected range of tens of microseconds.

Thesis on DSpace