SATNAC 2009 Conference Papers
Network Engineering
Title: Automated Traffic Class Prediction and Prioritization
on 802.11 Using Machine Learning
Authors: Christiaan Brand (
Abstract: With the steady
increase of Internet access rates (DSL, LTE, FTTH) and aging local access
technologies (802.11) QoS problems are starting to manifest themselves on the
inside of SOHo (Small Office / Home) networks. Bandwidth bottlenecks are slowly
moving to the inside of our networks and we need a solution to solve this
problem without investing in other access technologies. Substantial capital
investment has been made into some of these (802.11 WiFi Access Points at
airports and coffee shops) because of their availability in consumer computer
equipment. Our solution should be able to solve the problems experienced when
dealing with multiple traffic classes from the same node in a single cell. In
this paper we will examine the impact of implementing a traffic class
prediction module on a congested 802.11 network. A novel node prioritizing
scheme for wireless networks operating in DCF mode will also be presented.
Title: End-to-end QoS Control
For an ALL-IP (NGN) Platform using WiMax as Access Layer Network
Authors: Bessie Malila
(University of Cape Town), Neco Ventura (University of Cape Town)
Abstract: Telecommunications
networks are evolving towards an integrated ALL-IP architecture, commonly known
as the Next Generation Network (NGN). The architecture consists of Broadband
transport technologies both on the access and
core networks. One of the problems in NGN architecture is the ability to
guarantee end-to-end Quality of Service (QoS) provisioning to the integrated
network services in the form of voice, video, data, fax and IPTV. These
services present complex QoS requirements on the transport networks. In this
paper we present an integration of the main components of a NGN on a test bed.
We build an IP core network based on Linux routers and integrate it with an
existing WiMax access network, an existing IP Multimedia Subsystem (IMS)
network as the control layer component and a content server as an application
layer entity. We further propose a mapping of QoS classes between the two
networks to achieve end-to-end QoS guarantees. This project builds on previous
and current research work in the Communications Research Group at the