Digital Signal Processing Reference
In-Depth Information
Fig. 6.9
Control dimension mapping
the system state by fetching the correct configurations from memory. This operation
is cheap compared to the cost of calibration that only has to be carried out once.
In the next subsection we detail how this information can be combined efficiently
into a global set representing the network trade-off. Profiling each user separately
and combining the information at run-time is optimal for independent users or when
the correlation is unknown. When correlation is present and known, the number of
system states to calibrate and the run-time combination could be further reduced.
6.3 Managing the User Experience
We first formally state the energy efficient cognitive radio resource allocation prob-
lem to be solved at run-time. Then we propose a very efficient algorithm to achieve
the network-wide optimal configuration.
6.3.1 Smart Resource Allocation Problem Statement
We recall that our goal is to assign transmission grants via the AP, resulting in an
optimal setting of the control dimensions to each flow and node such that the per-
flow QoS constraint for multiple users are optimally met with minimal energy con-
sumption. For a given set of resources, control dimensions and QoS constraints, the
scheduling objective can be formally stated as:
n
min
C
net
ω
i
C
i
,
1
≤
m
≤
s
(6.5)
i
=
1
s.t.
