恶意攻击下改进的分布式扩散最小均方算法  

Improved distributed diffusion least mean square algorithm against malicious attacks

作　　者：魏巍[1] 李先河[2] 

机构地区：[1]西安邮电大学物联网与两化融合研究院,陕西西安710061 [2]西安邮电大学通信与信息工程学院,陕西西安710121

出　　处：《西安邮电大学学报》2016年第4期19-22,共4页Journal of Xi’an University of Posts and Telecommunications

基　　金：陕西省教育厅科学研究计划资助项目(12JK0541)

摘　　要：给出一种改进的分布式扩散最小均方算法,以改善无线传感器网络中攻击节点对网络估计性能的影响。根据攻击节点及正常节点估计值与待估目标参量的差值,确定采集数据被篡改程度,将其归一化后,作为融合权重,代替原扩散最小均方(DLMS)算法中固定的融合权重。仿真结果显示,通过动态改变融合权重,改进算法的全局均方偏差值有所降低,且对网络所受攻击强度的改变不敏感。An improved distributed diffusion least mean square algorithm is presented to weaken the influence of attack nodes on estimation performance in wireless sensor network. The extent of the data being tampered with is determined according to the difference between the target parameter to be estimated and the estimated value of the attack node or the normal node, which is normalized later, and taken as a new weight in the fusing stage to replace the original fusion weight in diffusion least mean square （DLMS） algorithm. Simulation results show that, by dynamically changing the fusion weights, the improved algorithm can reduce the global mean square deviation, and it is not sensitive to the change of the attack strength of the network.

关 键 词：无线传感网络 分布式 恶意攻击 最小均方算法 自适应权重 

分 类 号：TN911.7[电子电信—通信与信息系统]