Probability Model for User Datagram Protocol (UDP) Upstream Throughput for Single User in Real Time and Non-Real Time Scenarios in IEEE802.11b/g WLAN

Abstract

Use of Wireless Local Area Networks (WLAN) for all brand of
activities have spiraled in recent times as internet activities have
shot up and data requirements have gone through the roof. In
addition, the implementation of the user datagram protocol (UDP)
in WLAN’s has also seen an increase. UDP has a number of
advantages over the Transmission Control Protocol (TCP) as it has
faster response times due to its lack of non-inherent error
correcting capabilities which creates the problem of instability. A
number of researches though few, have tried to tackle this trend but
with little results. This paper attempt to develop a probability model
that would enable the prediction of UDP Throughput in a WLAN
environment. Experimental data was obtained for different kinds of
traffic scenarios real time and non-real time (RT and NRT) for a
single user on the network. The work focuses on Signal to Noise
Ratio (SNR), which was varied while measuring the throughput and
then obtained a model for the upstream throughput (UDP
uploading). Also, a probability model was developed using the
cumulative distribution function (CDF) based off on different
predefined throughput thresh holds. Using the empirical data
obtained, a CDF model of UDP upstream throughput was
developed and it showed that despite the erratic characteristic
behavior of UDP, a normal cumulative distribution function (CDF)
can be obtained and its throughput can be predicted at different
intervals of poor, good and outstanding signal classifications of
SNR.