Contents I. Introduction
One of the most fundamental aspects of wireless system level performance analysis is the modeling of propagation loss of signals. Most of the system level network analyses model the propagation effect by distance-based path loss functions [1]–[3]. While such path loss modeling is well justified by the free space propagation loss (also known as the Friis equation) and its simple variations, e.g., the two ray model [2], [4], it does not take into account the complicated structure of realworld propagation environment and is thus unable to fully characterize the properties of the path loss field. The remedy is to introduce a separate shadowing term to model the effect of blockage, scattering, reflection, diffraction, etc. The lognormal distribution is often used to model this shadowing effect. Mathematical justification of the lognormal shadowing stems from the multiplicative blockage loss and the central limit theorem [2], [5]. This model, while convenient in analyzing point-to-point links, falls short in analyzing wireless networks where nearby links are often blocked by common obstacles, and the shadowing statistics vary significantly depending on the length and angle of the links.
