high performance computing on graphics processing units: hgpu.org

The use of computer networks is drastically growing and the need for enhancing the existing network protocols and enforcing communication security thus is increasing. Tools like network simulators are used by researchers in order to test new scenarios and protocols in a controlled and reproducible environment. They allow the user to represent various topologies, simulate network traffic using different protocols, visualize the network and measure the performances. Although they are very useful, most of the widely used network simulators do not scale. Trying to simulate medium to large networks will result in a long simulation time unsuitable for investigating protocols. Some of the current network simulators implement various methods for accelerating the simulations by means of using parallel computation or changing the data representation of the nodes. However, most of the optimization techniques require additional hardware resources as a computational grid or cluster and deep knowledge of the structure of the simulator. A wide assortment of network simulators are available and most of them share the same issues when it comes to computation time and scalability. Simulating a network of 100 nodes on a normal CPU using standard simulator takes a long time. In this project we try to alleviate these problems by building a simulator on a Graphical Processing Unit [GPU] based on parallel execution on using the CUDA programming model developed by nVIDIA.