Future generations of wireless systems require opportunistic
spectrum access techniques to effectively detect
and access temporarily unused spectrum bands. Cognitive
radios, with their ability to learn and adapt to their
environment, promise to possess such powerful capabilities.
As a consequence, the spectrum allocation of a wireless
system could quickly and appropriately auto-adapt to react
to a sudden traffic variation.
In this article, we propose an innovative and efficient
distributed spectrum allocation algorithm, whose objective
is to maximize the system UL capacity by exploiting multiuser
diversity. The algorithm is capable of learning over
time and of adapting the spectrum allocation when changes
occur in the radio environment. Such an algorithm finds
its application in the scope of future WLAN systems (e.g.:
802.11x).
