Digital Signal Processing Reference
In-Depth Information
•
Observe RT situation:
The challenge in sufficiently observing the RT scenario is
the most prevalent in cases where Primary Users (PU) and Secondary Users (SU)
have to coexist in an OSA scenario. In Chap. 4 the flow will hence be instantiated
in a PU/SU scenario with special focus on the RT monitoring.
•
Map RT Situation to scenario:
Very often several sources of information have
to be combined to sufficiently understand the current scenario and act effectively.
This will be shown in a case study of IEEE 802.15.4 coexisting with IEEE 802.11,
where the goal of the IEEE 802.15.4 network is to find a single channel that is
least interfered by IEEE 802.11. Routing layer information, together with PHY
layer sensing information, will be combined to enable the IEEE 802.15.4 network
to find its single best channel.
•
Execute RT Procedure:
The execution of a well-defined RT procedure does not
necessary require a radio to be cognitive. However, it will be shown that the pro-
posed flow can still be used to achieve a run-time efficient design of smart radios
that adapt to their environment. The context will be an IEEE 802.11 network in
which multiple users coexist and have to share the medium, while meeting per-
formance constraints and minimizing energy cost.
•
Calibrate and learn RT Procedure:
The most important task of a cognitive ra-
dio is learning and calibrating how it behaves in the environment. How to achieve
this efficiently is illustrated in the context of IEEE 802.11 networks that aim at
minimizing the co-channel interference in a distributed way by adapting the out-
put power, sensing threshold and transmission parameters.
This chapter illustrates the generic design flow that will then be instantiated in the
next chapters. First, the design landscape is completely described in Sect.
3.2
. While
Chap. 1 focused mainly on hardware and policy flexibility, it is shown in Sect.
3.3
that the involved challenges for the control solution are even broader. Finally, in
Sect.
3.4
the proposed design framework is then introduced.
3.2 The Design Landscape Is No Longer Flat
The purpose of this section is to explain the design landscape for the control of
wireless devices and networks. The easiest systems to optimize are those that have
limited dynamism.
1
They can be approximated at DT by static models typically
designed for the worst case. Hence, relying on these static DT models, the system
can be configured completely at DT and does not need to be changed at RT.
However, as we will further detail in this chapter, wireless networks are ex-
tremely dynamic and, hence, DT-only frameworks for wireless networks, which al-
ways need to be designed based on the worst case operating conditions, are severely
suboptimal. Therefore, wireless networks ask for a more elaborate framework that
mean
(
FoM
)
max
(
FoM
)
>(
1
1
A system is said to have limited dynamism if
−
)
, when using a static config-
uration that optimizes mean
(
FoM
)
. FoM denotes the Figure of Merit at RT and
is some small
quantity.
