无线传感网络中拥塞控制与路由的跨层设计:分布式牛顿法

无线传感网络应用广泛, 其性能与路由选择和拥塞控制密切相关. 致力于拥塞控制与多径路由的跨层优化, 以实现在链路容量受限和节点能量受限情况下的无线传感网络效用最大化. 针对对偶次梯度算法具有收敛速度慢与信息交互量大等缺陷, 设计了具有二阶收敛性能的分布式牛顿算法来实现网络效用最大化. 通过矩阵分裂技术, 实现了只需单跳信息交互的牛顿对偶方向的分布式求解方法. 仿真结果表明, 分布式牛顿算法的收敛性能显著优于对偶次梯度算法.

Cross-layer Congestion Control and Routing Design for Wireless Sensor Networks:Distributed Newton Method

Wireless sensor networks have a wide range of extensive applications, and their performances are strongly dependent on routing and congestion control. Therefore, in this paper we address cross-layer optimization of congestion control and multi-hop routing to achieve network utility maximization for wireless sensor networks subject to link capacity and node power constraints. Considering that dual subgradient algorithm has a very slow convergence and large amount of information exchange, we design a distributed Newton method with quadratic convergence performance to maximize the network utility. We use matrix splitting technique to distributedly solve Newton dual direction through single-hop information exchange. Extensive simulation shows that distributed Newton algorithm converges faster than the dual subgradient algorithm.