Abstract
This thesis is concerned with the development of new mathematical models
for the IEEE 802.11’s access mechanisms, with a particular focus on
DCF and EDCA. Accurate mathematical models for the DCF and EDCA
access mechanisms provide many benefits, such as improved performance
analysis, easier network capacity planning, and robust network design.
A feature that permeates the work presented in this thesis is the application
of our new models to network environments where both saturated
and non-saturated traffic sources are present. The scenario in which
multiple traffic sources are present is more technically challenging, but
provides for a more realistic setting.
Our first contribution is the development of a new Markov model for
non-saturated DCF in order to predict the network throughput. This
model takes into account several details of the protocol that have been
hitherto neglected. In addition, we apply a novel treatment of the packet
service time within our model. We show how the inclusion of these effects
provides more accurate predictions of network throughput than earlier
works.
Our second contribution is the development of a new analytical model
for EDCA, again in order to predict network throughput. Our new
EDCA model is based on a replacement of the normal AIFS parameter
of EDCA with a new parameter more closely associated with DCF. This
novel procedure allows EDCA to be viewed as a modified multi-mode
version of DCF.
Our third contribution is the simultaneous application of our new Markov
models to both the non-saturated and the saturated regime. Hitherto,
network throughput predictions for these regimes have required completely
separate mathematical models. The convergence property of our
model in the two regimes provides a new method to estimate the network
capacity of the network.
Our fourth contribution relates to predictions for the multimedia capacity
of 802.11 networks. Our multimedia capacity analysis, which is based
on modifications to our Markov model, is new in that it can be applied
to a broad range of quality of service requirements. Finally, we highlight
the use of our analysis in the context of emerging location-enabled
networks.