The
University of Adelaide
Home |
School EEE 
ELEC ENG 4039A/B
FINAL YEAR HONOUR PROJECT
OPTIMISATION WLAN FOR BROADBAND ACCESS
Related Link
Current Development
Access point is one of the fundamental components in wireless LAN infrastructure mode. Most of access points are driven by embedded Linux based firmware. Hence, the optimisation is feasible if the firmware could be rebuilt.
We are currently working on modifying UltraWAP access point and LinkSys WAP54g to add optimisation modules to improve speed and reliability of network with the two optimisation techniques- dynamic channel switching and dynamically changing fragmentation threshold.
UltraWAP Access point (AP) is a direct replacement for MNWAPB which is currently used in school WLAN. The source code for the firmware was not available during the time limits of this project. For this reason the kernel development was not fully supported and a generic RTL 8186 Software Development Kit was used for firmware production. The firmware image was compiled, however it failed to boot at the final stage, hence the optimisation techniques were not able to be tested on this device. The failure at the final stage may due to the firmware not being in a binary format, second stage boot loader memory mapping or the checksum used in implementation may have been hard coded at the first stage boot loader. This problem has not yet been resolved, but a good basis has been provided for further development. Click here for more detail.
Linksys WAP54G is one of the popular IEEE802.11b/g access point, the source code of this firmware is available. The firmware was based on OpenWrt – linux distribution for Linksys routers and access points. In order to add the optimisation module on the AP, the entire kernel and firmware must be rebuilt and this is feasible. Therefore, the optimisation techniques were mainly implemented on this AP.
Both dynamic channel switching and dynamic changing fragmentation threshold techniques implemented base on the link quality of the channel. This link quality could not extract from the WLAN driver of WAP54G nor wireless tools, hence ping statistics was used. From ping statistics, the percentage of packet loss could be extracted. As the percentage of packet loss is inverse proportional to the link quality, DCS switch the channel according to the packet loss value. This value would then save in temporary file created from fragmentation module so that the fragmentation module could read the value from the file to determine the link quality for finding the optimum fragmentation threshold. Click here for more detail.
In brief, the optimisation implementations were completed and tested. They both functioned correctly and can be used to optimise the network performance or further development.
Back HOME
Copyright © The University of Adelaide 2006
Copyright | Disclaimer | Privacy