|
Fundamentally, the problem lies in the fact that collaborative applications often
require end-user computing devices to perform tasks that happen in the background
and are not directly advantageous to the user. This is in stark contrast
to the underlying principle of most battery-saving techniques: enter a low-power
state when possible. In this work, we seek to better understand how laptop users
use the batteries attached to their devices and explore how these usage patterns
can be used to guide how aggressively a device should participate in tasks from
which the user does not derive direct benefit.
In particular,we focus our attention on a TCP IP network configuration
ultrapeer selection algorithm that determines which collaborating peers are responsible
for maintenance tasks such as routing and caching. The goal of our algorithm
is to devote system maintenance tasks to devices that have themost stable
battery characteristics. Our results are not only applicable to building aGnutellastyle
overlay, but also provide insight into how to integrate battery-awareness into
applications such as collaborative data gathering.
|
