Digital Signal Processing Reference
In-Depth Information
Fig. 3.1
A condensed version of the proposed framework. DT procedures that are established
from DT models observe the RT situations and map this to a scenario. The DT procedure linked
with the observed scenario is calibrated at RT by the RT learning engine
goal of the proposed framework is to dynamically select the configuration of those
knobs to address the user's (or operator's) needs. The more knobs available to effec-
tively control the system, the more gains can be achieved when using the presented
framework.
In this topic, it is assumed that knobs are available (see Chap. 1). As we are
concerned with developing the CR, a control layer, the development of protocols or
implementation techniques that make those knobs available are outside the scope of
this work.
3.4.1.2 Environment Awareness
Both Haykin and Mitola identified environment awareness as a key element for fu-
ture CR systems. Indeed, a wireless device today does not operate independently but
acts within a larger environment. Awareness of this environment does not imply you
can understand it, which is the ultimate goal of a CR. However, the opposite does
hold: understanding an environment can't be achieved without closely monitoring
it!
When a wireless device monitors the environment, it can map observations to
an environment model, created at DT. This presumes that the environment can be
at least partly analyzed and modeled at DT. Such a model should however be very
parameterized and even abstracted to a high level. This is usually feasible as we
show in subsequent case study chapters.
In this way, we avoid having to do everything at RT (which provides too much
RT execution overhead) or everything at DT (which is too complex in practice or
