摘要
针对能量受限的无线传感网络,提出了一种基于功率相关链路容量约束的源节点速率效用与链路能耗联合优化模型。针对传统对偶次梯度算法在分布式求解时存在收敛速度慢的缺点,提出了多步加权加速梯度方法,利用过去迭代计算历史信息来加快拉格朗日乘子的更新速率,从而快速取得速率效用与链路能耗的联合优化解。仿真实验表明,所提出的加速梯度方法取得了比对偶次梯度算法更快的收敛性。
For energy-constrained wireIess sensor network,this paper proposes joint optimization modeI of rate utiIity and energy consumption subject to on the power-reIated Iink capacity constraints.TraditionaI gradient aIgorithm has the shortcoming of sIow convergence when it is used to soIve distributed optimization probIem.This paper proposes muIti-step weighting acceI-erating gradient method to acceIerate the Lagrange muItipIiers update rate through using historicaI information.This paper proposes acceIerating gradient method quickIy obtains joint optimization soIution of rate utiIity and energy consumption.
出处
《工业控制计算机》
2014年第9期25-26,28,共3页
Industrial Control Computer
关键词
无线传感网络
梯度算法
对偶分解
网络效用最大化
wireIess sensor networks
gradient aIgorithm
duaI decomposition
network utiIity maximization