1: \begin{abstract}

2: This paper considers models of large-scale networks where nodes are characterized by a set of states describing their local behavior, and by an explicit mobility model over a two-dimensional lattice. A stochastic model is given as a Markov population process that is, in general, infeasible to analyze due to the massive state space sizes involved. Building on recent results on fluid approximation, such a process admits a limit behavior as a system of ordinary differential equations, whose size is, unfortunately, dependent on the number of points in the lattice. Assuming an unbiased random walk model of nodes mobility, we prove convergence of the stochastic process to the solution of a system of partial differential equations of reaction-diffusion type. This provides a macroscopic view of the model which becomes independent of the lattice granularity, by approximating inherently discrete stochastic movements with continuous, deterministic diffusions. We illustrate the practical applicability of this result by modeling a network of mobile nodes with on/off behavior performing file transfers with connectivity to 802.11 access points. A numerical validation shows high quality of the approximation even for low populations and coarse lattices, and excellent speed of convergence with increasing system sizes.

3: \end{abstract}

4: