Despite the huge development and usage of devices
compliant with IEEE 802.11g, their performance is still hard to
predict and to guarantee. A lot of work has focused on
performance evaluation based on simulations, but it is not
sufficient, due to the great number of variables involved. Others
have focused on linking the application performance with the
channel conditions, however taking into consideration only one
end of communication.
This paper presents a test bed and a measurement
methodology to experimentally analyze the effects of physical
layer interference, such as AWG noise, on a scenario that entails
an actual, yet simple, IEEE 802.11g network supporting video
streaming applications. In addition, the paper presents an initial
set of experimental results that help in developing a simple
model for correlating physical layer measures to video
streaming performance. To this aim, the tests are conducted
through a cross-layer approach that allows to compare metrics
at different stack layers and predict some network behaviors.
The novelty of the model is the fact it considers the noise
conditions at both client and server and uses them in a cross–
layer approach to predict the video streaming time.