Applied Algebra Group at the University of Zürich


Going Beyond the Shannon Paradigm - a new approach to digital control and signals

Prof. Dr. Yutaka Yamamoto's talk
Date: 31.10.18   Time: 16.00 - 17.00   Room: Y27H12

Since the advent of the celebrated sampling theory by Shannon, the notion of band-limited signals have become very popular and prevalent. It shows that band-limited signals can be perfectly reconstructed from their sampled values. This fact led to the common understanding that one has to limit the signal bandwidth below the so-called Nyquist frequency to bring the digital signals into successful processing or control.
This talk challenges the above mentioned band-limiting principle. While the Shannon theory guarantees perfect reconstrucion for perfectly band-limited signals, there are many practical situations where the basic band-limiting hypothesis is not met, due to physical or practical constraints on the sampling period. For example, the superresolution in image reconstruction, rejection of high frequency disturbance signals beyond the Nyquist frequency, etc. This requires to optimally reconstructing intersample signals, including high frequency components. We will show that with a suitable choice of a signal generator model, this objective is indeed attained using robust sampled-data control theory. This can lead to many new applications in control and signal processing which were deemed impossible due to the limitation of the band-limiting hypothesis. We will show various applications in image processing, tracking/rejection of high frequency disturbance signal beyond the Nyquest frequecy, e.g., in hard-disc drives, and also indicate some new areas which may lead to a new horizon of digital control and signal processing.