|Contents and Learning Objectives:
Hardware Algorithms (Prof. Dr. Marcel Jacomet)
The need for higher performance and energy efficiency increase at the same time in portable applications. Hardware algorithms and hardware/software algorithms are the answer of such needs, resulting in system-on-chip microelectronic solutions for signal and data processing. This course block introduces a systematic hardware design approach using the FSMD model. Hardware arithmetic algorithms will be introduced and generic hardware design approaches deduced by the various concepts. The course block is rounded off with an introduction to the unfolding hardware parallelization concept.
Signal Processing (Prof. Dr. Thomas Niederhauser)
Real signals are often disturbed by noise, which requires digital signal processing to extract their information. Applications such as feedback control or autonomous sensor nodes, however, have strongly diverging requirements on the implemented algorithms with respect to energy, space and time constraints. In this course, the students learn how to design and realize adequate frequency-based filters in floating-point and fixed-point arithmetic. Adaptive filters are considered for systems with non-stationary noise sources. The course closes with more advanced multi-rate filter designs.
Computer Architecture (Prof. Dr. Theo Kluter)
This course block gives a broader insight in computer architectures by visiting the different types of processors, like RISC, VLIW, DSP, and GPU and their typical applications/short comings. Furthermore, the advantages and disadvantages of different memory models (shared-memory, distributed memory, caches and scratchpads) are discussed. Finally, memory coherence and consistence problems are shown with their solutions (coherence protocols in hard-/software and more advanced techniques like transactional memory models) are shown. The course closes with a brief overview of energy consumption and reduction.