Interference constrained network control based on curvature

Abstract

This paper proceeds from the premise that the topology of interference constrained wireless networks heavily impacts their node-to-node delay, routing energy, and capacity region. We quantitatively analyze how the discrete Ollivier-Ricci curvature of a network affects the performance metrics of several routing protocols. Since different protocols are optimal relative to different metrics under different topologies, an adaptive control system is proposed that identifies the topology curvature and selects the best protocol under current circumstances subject to user needs. Also, we analyze how sensitive the four routing protocols (Heat Diffusion, Dirichlet, Back Pressure and Shortest Path Routing) under examination are to varying topological environment, as it would commonly be encountered in wireless networks.