Digital Signal Processing Reference
In-Depth Information
Chapter 6
Anticipative Energy and QoS Management:
Systematically Improving the User Experience
6.1 Energy Efficiency for Smart Radios
6.1.1 Minimum Energy at Sufficient QoS
While the need for flexibility and intelligent control is well studied in the context
of smart or cognitive radios for the sake of spectrum efficiency, it can used for op-
timizing the usage of other scarce resources such as energy. The design framework
for smart radios of Chap. 3 is instantiated here to achieve smart, energy efficient
software radios. More specifically, it is shown how it can be used to achieve smart,
energy efficient IEEE 802.11a devices, provided their hardware is designed to be
energy scalable. An energy scalable radio can be reconfigured or tuned at run-time
in a way that energy consumption is impacted, so such a radio can benefit from
a smart control strategy to save energy. Of course, this energy saving should not
come at the expense of the user experience, so the objective of the smart control in
this chapter is: how to adapt the radio so that the application performance is met at
minimal possible energy consumption.
The focus of this chapter is on wireless networks where all users are in the
same collision domain with an access point (AP) to arbitrate exclusive channel ac-
cess (Fig.
6.2
). An uplink real-time video streaming application is considered. The
challenge in energy management for these systems is to determine how the sys-
tem should sleep or scale and still meet per-packet QoS timing requirements. The
instantiation takes advantage of the fact that the system will operate in dynamic en-
vironments where a single energy management solution is not sufficient. The main
dynamics are encountered in the channel state and application load requirements.
Furthermore, in a network of such systems, an efficient energy management algo-
rithm should exploit the variations across users to minimize the overall network
energy consumption.
Therefore the problem explored here could be stated as follows:
How does one
decide what system configurations to assign to each user at run-time to minimize
the overall energy consumption while providing a sufficient level of QoS? This must
