Conference Paper

A Unified Approach to Optimizing Performance in Networks Serving Heterogeneous Flows.

Conference: INFOCOM 2009. 28th IEEE International Conference on Computer Communications, Joint Conference of the IEEE Computer and Communications Societies, 19-25 April 2009, Rio de Janeiro, Brazil
Source: DBLP
Download full-text


Available from: Atilla Eryilmaz, Mar 31, 2015
  • G. Sun · H. Yu · L. Li · V. Anand · H. Di ·
    [Show abstract] [Hide abstract]
    ABSTRACT: Ideally, networks should be designed to accommodate a variety of different traffic types, while at the same time maximizing its efficiency and utility. Network utility maximization (NUM) serves as an effective approach for solving the problem of network resource allocation (NRA) in network analysis and design. In existing literature, the NUM model has been used to achieve optimal network resource allocation such that the network utility is maximized. This is important, since network resources are at premium with the exponential increase in Internet traffic. However, most research work considering network resource allocation does not take into consideration key issues, such as routing and delay. A good routing policy is the key to efficient network utility, and without considering the delay requirements of the different traffic, the network will fail to meet the user’s quality of service (QoS) constraints. In this paper, we propose a new NUM framework that achieves improved network utility while taking into routing and delay requirements of the traffic. Then, we propose a decomposition technique-based algorithm, D-NUM, for solving this model efficiently. We compare our approach with existing approaches via simulations and show that our approach performs well. KeywordsMathematical programming–Optimization–Network resource allocation–Network utility maximization–Optimal provisioning
    Photonic Network Communication 10/2011; 22(2):180-190. DOI:10.1007/s11107-011-0318-y · 0.79 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: TCP Reno can be regarded as a distributed algorithm over the Internet to maximize aggregate utility subject to capacity. The equilibrium reached by the interaction of many competing TCPs, each implemented only on the source and destination computers, is effectively maximizing an objective function for the entire network. In this paper, a new utility function is derived for TCP Reno under the precise end-to-end drop probability. The results show theoretically that the TCP Reno model proposed by Kelly [1] is reasonable. And simulation results are provided.
    Sciece China. Information Sciences 01/2012; 56(1). DOI:10.1007/s11432-012-4703-7 · 0.85 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We study throughput-optimum localized link scheduling in wireless networks. The majority of results on link scheduling assume binary interference models that simplify interference constraints in actual wireless communication. While the physical interference model reflects the physical reality more precisely, the problem becomes notoriously harder under the physical interference model. There have been just a few existing results on link scheduling under the physical interference model, and even fewer on more practical distributed or localized scheduling. In this paper, we tackle the challenges of localized link scheduling posed by the complex physical interference constraints. By cooperating the partition and shifting strategies into the pick-and-compare scheme, we present a class of localized scheduling algorithms with provable throughput guarantee subject to physical interference constraints. The algorithm in the linear power setting is the first localized algorithm that achieves at least a constant fraction of the optimal capacity region subject to physical interference constraints. The algorithm in the uniform power setting is the first localized algorithm with a logarithmic approximation ratio to the optimal solution. Our extensive simulation results demonstrate correctness and performance efficiency of our algorithms.
    IEEE Transactions on Parallel and Distributed Systems 01/2013; 25(10). DOI:10.1109/TPDS.2013.210 · 2.17 Impact Factor
Show more