Network Utility Maximization: Advances and Applications
Network Utility Maximization (NUM) significantly extends the classical network flow problem and provides an emerging framework to design a variety of resource allocation algorithms, such as TCP congestion control, and to understand layering as optimization decomposition. This talk presents a summary of selected recent advances on open issues in the NUM framework. We present new distributed algorithms that converge to the globally optimal rate allocation for NUM problems with nonconcave utility functions representing inelastic flows, with coupled utility functions representing interference effects or hybrid social-selfish utilities, and with rate-reliability tradeoff through adaptive channel coding in the physical layer. We also discuss the equilibrium properties of co-existence of heterogeneous congestion control protocols, correspondence of different decompositions of a generalized NUM problem to different layering architectures, and remaining open issues on NUM.

These recent results substantially expand the scope of the basic version of NUM, which is a classical monotropic program, by utilizing interesting mathematical theories such as distributed subgradient algorithms based on subdifferential calculus, Poincare-Hopf theorem in differential topology, and Positivstellensatz in real algebraic geometry. The power of such mathematics enables us to rigorously understand Internet protocols and design new algorithms for communication systems such as broadband access networks.

This talk is based on several preprints co-authored with A. R. Calderbank, M. Fazel, P. Hande, J. W. Lee, S. H. Low, D. Palomar, A. Tang, J. Wang at Princeton and Caltech.