Abstract: In accordance with the current network congestion problem, this paper introduces the network utility maximization framework on the basis of the optimization theory, where the objective function is the user rates’ utility function and the constraint is the link capacity, as a result the users’ satisfaction is maximized. A distributed rate control asynchronous algorithm is presented based on the duality theory and the Newton-like method is used instead of the gradient method in the link algorithm, and the feedback delay is time varying. Simulation results demonstrate that the algorithm can get the optimal value more quickly and guarantee its fairness.

Key words: congestion control, network utility maximization, Newton-like method, asynchronous algorithm, fairness

摘要： 针对当前网络中的拥塞问题，通过最优化问题引出网络效用最大化框架，在满足链路容量的约束下，最大化关于用户速率的效用函数，从而达到用户的最大满意程度。提出一种基于对偶原理的分布式速率异步控制算法，在链路算法设计过程中，用拟牛顿算法代替梯度算法，并且允许反馈延时时变。仿真结果表明，该算法收敛速度更快且满足公平性。

关键词: 拥塞控制, 网络效用最大化, 拟牛顿方法, 异步算法, 公平性

CLC Number: 

• TP301.6