Characterizing wireless networks by spatial correlations

We show that n-point correlation functions are powerful tools to characterize the structure of wireless networks. We argue that they convey the most interesting correlation structures in node locations, while being significantly easier and faster to compute than the classical correlation functions of stochastic geometry. As a case study we analyze the spatial structure of the WLAN access point locations in the east and west coasts of the USA. It is found that measured access point locations feature power-law or scale-free behavior in their correlation structures. Node location models commonly used in simulations and analytical calculations are also studied, and found to be unrealistic in this regard