Class based Modelling of End-to-End Packet Delay in Peer-to-Peer Networks

Thorsten Kisner

Keywords

Statistical and Probabilistic Modelling, Network Simulation, EndtoEnd Packet Delay, Quality of Service, PeertoPeer Networks

Abstract

In many research areas the knowledge of end-to-end packet delay (transmission time) is very important to get a better understanding of network or application behavior. Especially Peer-to-Peer (P2P) applications and their management protocols require a routing of a single message over multiple nodes. The overall transmission time from the source node to the final destination node is the accumulated transmission time of all single hops. In this case different links with different delay and bandwidths are used, to predict the accumulated transmission time, all single links must be simulated. While sophisticated simulation environments can simulate transmission characteristics at the network level, these environments are in most cases not embeddable into special purpose simulation tools. Creating a P2P network topology with a large number of nodes at the network level in a simulation environment is a challenging task, while researcher or developer are often only interested in the upper protocol levels and not in the network layer. This proposal offers a simulation model on an high abstraction level considering aggregated geographical information and class based statistical predictions. The model can be used and implemented with the provided parameters with a low effort.

Important Links:



Go Back