Distributed Large Scale Network Utility Maximization

被引:25
作者
Bickson, Danny [1 ]
Tock, Yoav [1 ]
Zymnis, Argyris [2 ]
Boyd, Stephen P. [2 ]
Dolev, Danny [3 ]
机构
[1] IBM Haifa Res Lab, IL-31905 Haifa, Israel
[2] Stanford Univ, Dept Elect Engn, Stanford, CA 94305 USA
[3] Hebrew Univ Jerusalem, Sch Comp Sci & Engn, IL-91904 Jerusalem, Israel
来源
2009 IEEE INTERNATIONAL SYMPOSIUM ON INFORMATION THEORY, VOLS 1- 4 | 2009年
关键词
DECOMPOSITION;
D O I
10.1109/ISIT.2009.5205655
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
Recent work by Zymnis et al. proposes an efficient primal-dual interior-point method, using a truncated Newton method, for solving the network utility maximization (NUM) problem. This method has shown superior performance relative to the traditional dual-decomposition approach. Other recent work by Bickson et al. shows how to compute efficiently and distributively the Newton step, which is the main computational bottleneck of the Newton method, utilizing the Gaussian belief propagation algorithm. In the current work, we combine both approaches to create an efficient distributed algorithm for solving the NUM problem. Unlike the work of Zymnis, which uses a centralized approach, our new algorithm is easily distributed. Using an empirical evaluation we show that our new method outperforms previous approaches, including the truncated Newton method and dual-decomposition methods. As an additional contribution, this is the first work that evaluates the performance of the Gaussian belief propagation algorithm vs. the preconditioned conjugate gradient method, for a large scale problem.
引用
收藏
页码:829 / +
页数:2
相关论文
共 30 条
  • [1] [Anonymous], 1997, Applied Numerical Linear Algebra
  • [2] [Anonymous], IEEE INT S INF THEOR
  • [3] [Anonymous], 1998, Network optimization: Continuous and discrete models
  • [4] [Anonymous], 1999, SPRINGER SCI
  • [5] BICKSON D, 2008, IEEE INT S INF THEOR
  • [6] BICKSON D, 2008, 46 ALL C COMM CONTR
  • [7] BICKSON D, 2007, P 45 ALL C COMM CONT
  • [8] Boyd S., 2004, CONVEX OPTIMIZATION, DOI DOI 10.1017/CBO9780511804441
  • [9] Layering as optimization decomposition: A mathematical theory of network architectures
    Chiang, Mung
    Low, Steven H.
    Calderbank, A. Robert
    Doyle, John C.
    [J]. PROCEEDINGS OF THE IEEE, 2007, 95 (01) : 255 - 312
  • [10] DECOMPOSITION PRINCIPLE FOR LINEAR-PROGRAMS
    DANTZIG, GB
    WOLFE, P
    [J]. OPERATIONS RESEARCH, 1960, 8 (01) : 101 - 111