Abstract:
This project investigates the problem of the network congestion in a university networks, which can be occurs by using high bit rate applications as P2P, video, or online gaming. Such problems can cause higher delays and some traffic drops, so a suitable mechanism that can manage that congestion with less drops and delays is required.
To resolve that problem firstly a bandwidth control mechanism based on volume was proposed to specify restriction for the links quotas for colleges, departments, labs, and users. And to achieve fairness between users max-min fairness algorithm was used to allocate the bandwidth to the colleges, departments, labs, and users.
Secondly, to investigate of the heavy applications network traffic classification was done by using WEKA software and applying different machine learning algorithms in some internet data set collected from CAIDA and get the accuracy results for each one.
Finally, a congestion avoidance mechanism was proposed by setting the Quality of Service (QoS) parameters in the network is proposed which needs classification mechanism for the network traffic that based on the application type, then by giving the desired application traffic the higher priority on the network and the undesired one the lower priority, this will give the desired traffic the best resources , while the low priority application will be delayed and dropped while the higher priority application is served and this by using OPNET simulation.
