Model-based stochastic simulation of super peer promotion in P2P VoIP using graph transformation

The concept of super peer has been introduced to improve the performance of popular P2P VoIP applications. A super peer is the strongest peer in the network that has the capacity to act as a server for a set of VoIP clients. There is no doubt that by taking benefit of heterogeneity, super peer can do improve the efficiency, without compromising the decentralised nature of P2P networks. The core issue in the formation of super peer based overlay network is the selection of super peer among the participant peers. To solve this problem a number of solutions have been proposed in the literature. Generally, super peer are selected among the best nodes in the network, for example those with the most abondant resources, such as bandwidth, CPU cycles or memory. The next issue is when the peer shall be selected for this extended role either as when peer joins the network, or at any time during the session or promotion to super peer should be subject to requirements. In order to validate these approaches of super peer selection, simulation would be an ideal choice, but most existing simulation approaches cannot cope with unbounded dynamic changes of network topology. We propose an approach to modelling and simulation of P2P systems based on graph transformations, a visual rule based formalism that has recently been supported by facilities for stochastic modelling and simulation. We are considering a P2P VoIP applications such as Skype, we model three alternative solutions to the problem of peer promotion to super peers and evaluate these through simulation.