Game theory for heterogeneous flow control

Abstract

A general framework is developed for networks with flows that use all available congestion signals to regulate their rates. It is conceptually a generalization of the existing network utility maximization (NUM) theory for homogeneous congestion control. Instead of a convex optimization characterization in NUM, a game with multiple convex optimizations is formulated to characterize equilibria in such a network. Examples are provided to motivate the needs of this general theory. We also provide some basic properties of the game and point out some possible future directions along this line. © 2008 IEEE.