基于Markov流量预测和改进蚁群算法的分簇自适应路由  

Adaptive Cluster Routing Based on Markov Flow Prediction and Improved Ant Colony Algorism

作　　者：孙亮[1] 

机构地区：[1]兰州文理学院电子信息工程学院,兰州730000

出　　处：《计算机测量与控制》2014年第3期820-822,共3页Computer Measurement &Control

摘　　要：为了改进传统分簇路由协议的被动轮换簇头导致的簇头过早死亡,以及寻找簇间多跳路由时仅考虑长度因素而导致路径拥塞的缺陷,提出了一种基于Markov预测节点数据流量和改进蚁群算法的分簇路由协议;在网络初始化阶段,Sink节点对整个网络进行非均匀分簇以避免"盲区";簇成员节点存储自身的状态序列并能预测在未来时刻的数据流量,当簇头需要轮换时,簇头管理节点接收节点预测数据并选择具有最小数据流量的节点作为新簇头;在寻求簇间多跳路由时,引入改进的蚁群算法,使得簇头在选择下一跳节点时,综合考虑路径长度、节点剩余能量以及路径拥挤度等因素;仿真实验证明文中的分簇路由协议能最大程度地均衡节点负载和延长网络的生命期,在运行到450轮时才出现第一个死亡节点,较其它方法具有较大的优越性。In order to improve the traditional clustering protocol passively rotating the cluster head which leads to the node death earlier, a clustering routing protocol based on Markov predicating node data flow and improved ant colony is proposed. In initial time of network, the network was clustered by Sink node to avoid ＂blind spots＂, cluster member stored its state sequence and can predict the latter data traffic. When the cluster head needing to be rotated, the cluster head manage node received the predicating data from cluster members and choose the node having the least data flow as the new cluster head. In search of the multi-- route between clusters, the improved ant colony algorism was improved to choose the route considering route length, node remain energy and route crowd extent. The Simulation experiment shows the method in this paper can realize the node load balance and prolong the network life cycle, the first dead node appeared in the 450 wheel, com- pared with the other methods, it has bigger priority.

关 键 词：分簇路由 流量预测 马尔科夫链 蚁群算法 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]