?
Reducing the Root-Mean-Square Error at Signal Restoration using Discrete and Random Changes in the Sampling Rate for the Compressed Sensing Problem
The data revolution will continue in the
near future and move from centralized big data to
“small” datasets. This trend stimulates the emergence not
only new machine learning methods but algorithms for
processing data at the point of their origin. So the
Compressed Sensing Problem must be investigated in
some technology fields that produce the data flow for
decision making in real time. In the paper, we compare
the random and constant frequency deviation and
highlight some circumstances where advantages of the
random deviation become more obvious. Also, we
propose to use the differential transformations aimed to
restore a signal form by discrets of the differential
spectrum of the received signal. In some cases for the
investigated model, this approach has an advantage in the
compress of information.
