A view-probability-matrix approach to the modelling of gossiping protocols

This paper addresses the quantitative analysis of gossiping protocols. In contrast to existing approaches which are entirely based on the simulation of the individual nodes' behaviours, we present a new approach based on summary stochastic models for the peer sampling service. Instead of an ordinary state- and transition-based model, a matrix-based approach is presented. Starting from a basic model with static node population and without ageing of neighbourhood information, refinements of the model are presented which enable the modelling of ageing and dynamic population. The paper also contains some experimental results for the different models introduced in the paper.

• Type: Journal
• Year: 2008
• Where: SIGMETRICS
• Authors: Thomas Krieger, Martin Riedl, Johann Schuster, Markus Siegle
• Keywords: Hardware, Peer Sampling Service, SIGMETRICS 2008, Static Node Population, Summary Stochastic Models