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
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we focus on the problem of dynamic rate and power allocation in wireless ad hoc networks with slow-fading channels, where a mixture of elastic and inelastic traffic is supported. A stochastic optimization problem incorporating the quality of service requirements of the two types of traffic is formulated, which aims to maximize the network performance by allocating the power for each link and controlling the service rates of all flows in the network. Since the utility functions of inelastic flows may be non-concave, which are difficult to be readily transformed to be concave, the proposed original problem is non-convex. However, despite the existing difficulty, a dynamic rate and power allocation algorithm (named DRPAA), is proposed to solve the original optimization problem. In DRPAA, both the stochastic duality theory and the particle swarm optimization approach are used, and this dynamic algorithm provides a good approximation to the optimal solution when the variation of the channel condition of each link gets larger. By using DRPAA, flow rates and link powers are dynamically allocated in each network state without the need for the distribution of the network states. Simulation results show that DRPAA has a good convergence speed and can efficiently utilize network resources to improve the network performance.
    Wireless Personal Communications 05/2013; 70(1). DOI:10.1007/s11277-012-0702-7 · 0.98 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 · 0.70 Impact Factor
  • [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 · 0.75 Impact Factor

Full-text (2 Sources)

1 Download
Available from
Mar 31, 2015