On The Modeling Of P2P Systems as Temporal Networks: A Case Study With Data Streaming

Temporal networks are a useful tool to model complex systems' dynamics, especially when they are characterized by high dynamicity. While there is strong literature on simulation tools for complex and dynamical systems, there is a lack of viable solutions to model and exploit temporal graphs in simulation. In this work, we present a system devised to simulate complex systems and their evolution by using temporal graphs data structures. As a use case, we focus on data dissemination over peer-to-peer systems characterized by a relevant presence of churns. The simulation of dissemination algorithms on temporal graphs involves evaluating their efficiency in terms of coverage, delay, and number of messages sent. In particular, a reasonable trade-off between the speed of delivery and the generated network traffic must be found. In this work, besides traditional gossip strategies, a more complex scenario is considered, where a second overlay, structured as a tree, is built for more efficient propagation of data from the source to the interested peers only. This scenario is analyzed using multiple simulated strategies. More specifically, we investigate how the simulation methodology can be used for evaluating the efficiency of dissemination protocols in a peer-to-peer environment but with a specific focus on the modelling of churn.