基于Cauchy-Schwarz逼近的网络坐标导引

Cauchy-Schwarz approximating bootstrap algorithm for network coordinate

针对拓扑结构突变时网络坐标系统存在的剧烈抖动问题,定义节点导引步和误差修正步,通过坐标的延迟发布实现突变抑制。在导引步节点保持坐标广播静默,并基于Cauchy-Schwarz不等式生成迭代序列,逼近节点坐标的较优初值;在修正步执行Vivaldi算法,修正自身坐标并广播更新。仿真实验显示,该方法能在保证定位精度的前提下,有效抑制拓扑结构突变对网络坐标系统产生的剧烈影响,显著提升网络坐标系统泛化能力。

For strong oscillation occurs for topology mutation,this paper splited node＇s life cycle into a bootstrap phase and an error-celebration phase,and mitigated mutation through delaying release node＇s coordinate.Node keeped radio silence in bootstrap phase,and generated an iterative sequence by using Cauchy-Schwarz inequality to approximate its coordinate＇s optimal value.In error-celebration phase,node executed Vivaldi algorithm to update and broadcasted its coordinate.Experiments and simulations show that this method could mitigate the strong influence caused by topology mutation without losing locating accuracy,and improved the generalization capability of NCS dramatically.