Analytical Model for Semantic Overlay Networks in Peer-to-Peer Systems

The hit rate of basic Peer-to-Peer protocols can be increased by using semantic metadata to transform the network structure or construct an overlay network. These semantic protocols benefit from the observation that the pattern of the stored or queried documents depends on the fields of interest of the nodes in the network. However, the performance of such networks highly depends on boundary conditions. These conditions determine both the probability of a successful query and the network traffic generated by the queries. In this paper, we will propose an analytical model that calculates the success rate of a query in a peer-to-peer system with a semantic overlay network. With the model, it is possible to optimize these systems by finding the appropriate values for the boundary conditions when designing the network or at runtime. The model has also been validated with a series of simulations. Key-Words: Modeling, Peer-to-Peer, Semantic overlay networks, Simulation