Research Of Topological Folding In Large-Scale Network Simulation

Network simulation with its low cost and high accuracy, is more and more popular between network researchers, it has been a key method to do research on network behavior. However, due to hardware resource constraints, the existing network simulation tools cannot be completed on the large-scale network simulation. Therefore, under the conditions of the existing resources, how to reduce consumption of resources needed by large-scale network simulation, improve performance to make it possible to quickly simulate thousands or millions of nodes, or even the Internet is an issue.This paper presents a theory called "topological folding" which can substantially increase the scale of network simulation by using a node to simulate a region in the original topology. By analyzing how the parallel network simulator PDNS (Parallel/Distributed Network Simulator) works, find out the shortcomings of the existing optimization strategy and make improvements. On the premise of the authenticity, reduce as much consumption of resources in the network simulation as possible to achieve the objective of improving the scale network simulation. This paper proposes the following implementation strategy:First, abstract the actual link to an abstract link mathematical model and remain the data packet transmission process in the link unchanged.Second, abstract a tree region in the original topology to an abstract region model consisting of more than one abstract link models and ensure that the regional model can restore various information of the original topology.Third, introduce routing mechanism in the abstract regional models to make the packet transmission in the abstract region unchanged.By using the above strategy, "topological folding"-based simulation system can substantially increase the scale of network simulation and the performance.