A Caltech Library Service

Optimal zero-queue congestion control using ADMM

Matni, Nikolai (2017) Optimal zero-queue congestion control using ADMM. In: 2017 American Control Conference. IEEE , Piscataway, NJ, pp. 5598-5604. ISBN 978-1-5090-5992-8.

Full text is not posted in this repository. Consult Related URLs below.

Use this Persistent URL to link to this item:


We propose an alternating direction method of multipliers (ADMM) based solution to a network utility maximization (NUM) problem, and show that it leads to a protocol for congestion control that converges rapidly to utility maximizing transmission rates while guaranteeing zero congestion throughout the network. Our approach hinges on a novel decomposition of the NUM problem that leads to easily solvable iterate update subproblems - we provide closed form solutions for these subproblems for proportional fairness and minimum delay fairness utility functions. We further show that this decomposition lends itself to a distributed implementation that naturally allows for the generation of a set of feasible transmission rates at each iteration of the algorithm, thus ensuring that queues remain empty throughout the network. We formalize these notions in the form of a congestion control protocol, and comment on their usefulness and implementation in the context of recent developments in software defined networking. Finally we compare the performance of our approach to state-of-the-art algorithms from the networking and optimization communities via a datacenter network simulation.

Item Type:Book Section
Related URLs:
URLURL TypeDescription
Matni, Nikolai0000-0003-4936-3921
Additional Information:© 2017 IEEE.
Record Number:CaltechAUTHORS:20170705-162343972
Persistent URL:
Official Citation:N. Matni, "Optimal zero-queue congestion control using ADMM," 2017 American Control Conference (ACC), Seattle, WA, USA, 2017, pp. 5598-5604. doi: 10.23919/ACC.2017.7963826
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:78782
Deposited By: Kristin Buxton
Deposited On:06 Jul 2017 21:30
Last Modified:03 Oct 2019 18:12

Repository Staff Only: item control page