无线传感网中节点能量和链路容量约束的二阶分布式流控制方法  

Node Energy and Link Capacity Constrained Distributed Flow Control in Wireless Sensor Networks: A Second-Order Method

作　　者：穆元彬 史清江[1] 徐伟强[1] 余晴里 汪亚明[1] 

机构地区：[1]浙江理工大学信息学院,浙江杭州310018

出　　处：《电子学报》2015年第5期1000-1006,共7页Acta Electronica Sinica

基　　金：国家自然科学基金(No.61374020;No.61302076;No.61272311;No.61101111);教育部重点科学技术研究项目(No.212066);浙江省自然科学基金(No.LY12F0242;No.LQ12F01009;No.LQ13F010008);浙江理工大学科研启动基金(No.1203805Y)

摘　　要：分布式流控制设计是确保无线传感网络性能的关键问题.本文针对无线传感网络中的分布式流控制设计问题,建立了一种节点能量和链路容量共同约束的流控制优化模型.由于对偶次梯度分解算法存在收敛速度慢、步长选择敏感等缺点,本文设计了基于矩阵分裂的分布式牛顿算法来实现网络效用最大化.该算法具有与对偶次梯度分解算法几乎相同的通信复杂度,但能够达到快速的二次收敛速度.仿真结果表明,该算法的收敛性能明显优于对偶次梯度算法.Distributed flow control is a key issue in wireless sensor network design to achieve good network performance. This paper presents a node energy and link capacity constrained flow control optimization model for distributed flow control in wire- less sensor networks. Since subgradient-based dual-decomposition method suffers from slow convergence rote and sensitivity in step- size choosing,this paper adopts matrix-splitting based distributed Newton algorithm to maximize the network utility. The proposed algorithm can achieve quadratic convergence rote while requiring almost the same coinmumcation overhead as the subgradient-based dual-decomposition algorithm.Moreover,it is shown by simulations that the proposed algorithm has a much better convegence per- formance than the subgradient-based dual-decomposition algorithm.

关 键 词：无线传感网络 矩阵分裂 分布式牛顿算法 网络效用 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]