On Estimating Message Diffusion Dynamics in Epidemic Broadcasting

In this paper, we propose a method for estimating message diffusion dynamics in epidemic broadcasting called DIFER (Diffusiveness Inference From Epidemic broadcasting Records). Information diffusion in epidemic broadcasting is affected by several factors such as node mobility and the algorithm for epidemic broadcasting. In the literature, properties of epidemic broadcasting have been investigated with mathematical analyses and simulation experiments. In those studies, however, it is generally assumed that node mobility, the algorithm of epidemic broadcasting, and system parameters such as the wireless communication range are known. In this paper, we therefore propose DIFER for estimating message diffusion dynamics in epidemic broadcasting under situations such that internal properties of epidemic broadcasting are not known. We also investigate the effectiveness of DIFER through simulations. Our finding includes that DIFER can accurately estimate the message diffusion matrix of epidemic broadcasting with P-BCAST (PUSH-based Broad CAST) under different numbers of nodes, measurement intervals and node velocities.