As the Internet spreads more and more engineers have to design faster network devices. Testing such high speed devices - like routers, firewall, mobile gateway, etc. - is an unsolved problem. Internet Service Providers are not willing to take the risks of online testing and sharing recorded traffic data is also limited due to user privacy.
As a solution for this problem a framework has been developed by the co-operation of Ericsson Hungary Ltd. and High Speed Networks Laboratory at the Budapest University of Technology and Economics. The tool is able to emulate typical user Internet activities, record their network traffic and create a high speed network stream using the previously recorded data. The definition of typical user activities uses special strings which refers the applications run by the given user. The tool is able to analyze real measurement results and use the results for defining typical user scenarios. During this process the frameworks uses a scoring algorithm which can determine the similarities between two strings.
This thesis presents the general architecture of the traffic emulation framework and the operations of its main parts. The website I have developed for controlling the framework operations are also presented. I will also describe the string scoring algorithm which is evaluated via an artificially created database.